Method of adaptive interactive learning control and a lithographic manufacturing process and apparatus employing such a method Number:7,181,296 from the United States Patent and Trademark Office (PTO) owispatent

Title: Method of adaptive interactive learning control and a lithographic manufacturing process and apparatus employing such a method

Abstract: By applying time-frequency analysis to a given standard iterative learning control or ILC an adaptive filter for the learned feed-forward loop is designed. This time varying filter varies according to the momentary frequency content of the error signal and allows to discriminate between areas of deterministic and stochastic error. Its application results in selective application of ILC to those intervals where error signals of high level are concentrated and allows application of a single ILC acquisition to different setpoint trajectories. The adaptive filter finds particular use in lithographic scanning systems where it is used for varying scan length.

Patent Number: 7,181,296 Issued on 02/20/2007 to Rotariu,   et al.

---

Inventors:	Rotariu; Andreea Iuliana (Sydney, AU), Ellenbroek; Rogier (Den Haag, NL), Steinbuch; Maarten (Helmond, NL), Van Baars; Gregor Edward (Eindhoven, NL)
Assignee:	ASML Netherlands B.V. (Veldhoven, NL)
Appl. No.:	10/857,450
Filed:	June 1, 2004

---

Foreign Application Priority Data

---

Aug 06, 2003 [EP]			03077461

Current U.S. Class:	700/44 ; 318/607; 318/638; 318/687; 700/121; 700/60; 700/63; 700/65; 706/21; 706/22; 706/23
Current International Class:	G05B 13/02 (20060101); G05B 1/06 (20060101); G05B 11/00 (20060101); G05B 13/04 (20060101); G05B 19/18 (20060101); G06F 15/18 (20060101); G06F 19/00 (20060101); G06G 7/00 (20060101)
Field of Search:	700/17,28-31,44,47,55,56,60,61,63,65,117-121 706/21-23 318/560,607,638,652,671,687

---

References Cited [Referenced By]

---

U.S. Patent Documents

	6373033		April 2002		de Waard et al.
	6449369		September 2002		Carme et al.
	6563663		May 2003		Bi et al.
	6686716		February 2004		Predina et al.

Other References

Time-frequency Analysis of a Motion System with Learning Control, written by Rotariu et al., Jun. 4-6, 2003, pp. 3650-3654. cited by other . Stability of a Novel Iterative Learning Control Scheme with Adaptive Filtering, written by Danian Zheng and Andrew Alleyne, Jun. 4-6, 2003, pp. 4512-4517. cited by other . Synthesis of a Robust Iterative Learning Controller Using an H.sub.--Approach, written by D. de Roover, Dec. 1996, pp. 3044-3049. cite- d by other . European Search Report in reference to EP 03 07 7461. cited by other.

Primary Examiner: Barnes; Crystal J.
Attorney, Agent or Firm: Pillsbury Winthrop Shaw Pittman LLP

---

Claims

---

What is claimed is:

1. A method of controlling a system or process using a learned feed-forward signal, comprising: executing a feed-forward control of said system or process based on a first setpoint signal that is indicative of a first setpoint trajectory; determining an error signal between said first setpoint signal and an output signal of said system or process; performing a time-frequency analysis of said error signal; identifying frequency components and corresponding levels of said error signal as a function of time based on said time-frequency analysis; determining a time-varying bandwidth for a filter based on the frequency components as a function of time, such that repetitive dynamic attributes exhibited by said system or process, as indicated by said frequency components, are incorporated into said learned feed-forward signal at the appropriate time instants; applying said filter to said learned feed-forward signal; and generating a control signal in response to said applying of said filter.

2. The control method of claim 1, wherein said filter is a low-pass filter and said bandwidth function is determined by a cut-off frequency which varies in accordance with said frequency content and level of said error signal.

3. The control method of claim 1, wherein in said determining a time-varying bandwidth, said the bandwidth is lowered when said system or process does not exhibit said repetitive dynamic attributes.

4. The control method of claim 1, wherein said learned feed-forward signal has a zero value for those time instants corresponding to points along said first setpoint trajectory where the system or process does not exhibit said repetitive dynamic attributes.

5. The control method of claim 1, further including: controlling said system or process over a second setpoint trajectory, indicated by a second setpoint signal, said second setpoint trajectory differing from said first setpoint trajectory only with respect those parts of said first setpoint trajectory which do not cause said system or process to exhibit said repetitive dynamics, wherein said feed-forward signal corresponding to said second setpoint signal is developed from said learned feed-forward signal corresponding to said first setpoint signal by inserting or removing values for said part of said second setpoint trajectory that differs from that of said first setpoint trajectory.

6. The control method of claim 5, wherein said first setpoint trajectory causes a scanning movement along a first scan length and said second setpoint trajectory causes a scanning movement along a second scan length, which is different than said first scan length.

7. The control method of claim 1, wherein said system is a lithographic scanning system, in which a beam of radiation is scanned via a patterning device onto a substrate having a partial resist layer, said control method being used to control a scanning position of either the substrate or the patterning device according to a scanning trajectory.

8. A user terminal comprising a mechanism operable to perform the method of claim 1.

9. A control system for a system or process using a learned feed-forward signal, said system comprising: an iterative learning control element configured to execute a feed-forward control of said system or process according to a predetermined setpoint; an analysis element configured to perform a time-frequency analysis of an error signal between said setpoint and an output of said system or process and to identify frequency components and corresponding levels of said error signal as a function of time; and an adaptive filter having a time varying bandwidth based on said frequency components of said error signal as a function of time, said filter being applied to said learned feed-forward signal to obtain a control signal to control said system or process, wherein in response to identifying repetitive dynamic attributes exhibited by said system or process based on said frequency components, said bandwidth is adjusted so that the repetitive dynamic attributes are incorporated into said learned feed-forward signal at appropriate time instants.

10. The control system of claim 9, wherein said process is a lithographic manufacturing process, in which a projection beam of radiation is scanned via a patterning device onto a substrate having a partial resist layer, said control system is adapted to control the scanning position of either the substrate or the patterning device according to a scanning trajectory.

11. A lithographic apparatus, comprising: an illumination system configured to condition a beam of radiation; a support configured to support a patterning device that imparts a desired pattern to the beam of radiation; a substrate holder configured to hold a substrate; a projection system configured to project the patterned beam onto a target portion of the substrate; and a control system configured to control movement of said support structure and/or said substrate holder along a predetermined trajectory using a learned feed-forward signal, said control system comprising: an iterative learning control element configured to execute a feed-forward control of the movement of said support structure and/or said substrate holder according to a setpoint signal corresponding to said predetermined trajectory; an analysis element configured to perform a time-frequency analysis of an error signal between said setpoint signal and an output signal indicative of the movement of said support structure and/or said substrate holder and to identify frequency components and corresponding levels of said error signal as a function of time; and an adaptive filter having a time varying bandwidth based on said frequency components of said error signal as a function of time, said filter being applied to said learned feed-forward signal to obtain a control signal to control said system or process, wherein in response to identifying repetitive dynamic attributes exhibited by said system or process based on said frequency components, said bandwidth is adjusted so that the repetitive dynamic attributes are incorporated into said learned feed-forward signal at appropriate time instants.

12. A computer-readable storage medium storing a program which, when run on a computer, controls the computer to perform the method of controlling a system or process using a learned feed-forward signal, comprising: executing a feed-forward control of said system or process according based on a first setpoint signal that is indicative of a first setpoint trajectory; determining an error signal between said first setpoint signal and an output signal of said system or process; performing a time-frequency analysis of said error signal; identifying frequency components and corresponding levels of said error signal as a function of time based on said time-frequency analysis; determining a time-varying bandwidth for a filter based on the identified frequency components as a function of time, such that repetitive dynamic attributes exhibited by said system or process are incorporated into said learned feed-forward signal at the appropriate time instants; applying said filter to said learned feed-forward signal; and generating a control signal in response to applying said filter.

13. The computer-readable storage medium of claim 12, wherein said filter is a low-pass filter and said bandwidth function is determined by a cut-off frequency which varies in accordance with said frequency content and level of said error signal.

14. The computer-readable storage medium of claim 12, wherein in said determining a time-varying bandwidth, said the bandwidth is lowered when said system or process does not exhibit said repetitive dynamic attributes.

15. The computer-readable storage medium of claim 12, wherein said learned feed-forward signal has a zero value for those time instants corresponding to points along said first setpoint trajectory where the system or process does not exhibit said repetitive dynamic attributes.

16. The computer-readable storage medium of claim 12, further including: controlling said system or process over a second setpoint trajectory, indicated by a second setpoint signal, said second setpoint trajectory differing from said first setpoint trajectory only with respect those parts of said first setpoint trajectory which do not cause said system or process to exhibit said repetitive dynamics, wherein said feed-forward signal corresponding to said second setpoint signal is developed from said learned feed-forward signal corresponding to said first setpoint signal by inserting or removing values for said part of said second setpoint trajectory that differs from that of said first setpoint trajectory.

17. The computer-readable storage medium of claim 16, wherein said first setpoint trajectory causes a scanning movement along a first scan length and said second setpoint trajectory causes a scanning movement along a second scan length, which is different than said first scan length.

18. The computer-readable storage medium of claim 12, wherein said system is a lithographic scanning system, in which a beam of radiation is scanned via a patterning device onto a substrate having a partial resist layer, said control method being used to control a scanning position of either the substrate or the patterning device according to a scanning trajectory.

19. A control system for controlling a system or process, comprising: an input configured to receive a first setpoint signal; an iterative learning control element configured to generate a learned feed-forward signal based on said setpoint signal; an operational element configured to determine an error signal between said setpoint signal and an output signal; an analysis element configured to perform a time-frequency analysis of said error signal to identify frequency components and corresponding levels of said error signal as a function of time; and an adaptive filter having a time varying bandwidth based on said frequency components of said error signal as a function of time, said filter being applied to said learned feed-forward signal to generate a control signal to control said system or process, wherein, in response to identifying repetitive dynamic attributes exhibited by said system or process based on said frequency components, said bandwidth is increased to accommodate said repetitive dynamic attributes into said learned feed-forward signal at appropriate time instants, and wherein, in response to not identifying repetitive dynamic attributes, said bandwidth is reduced.

20. The control system of claim 19, further including: a second setpoint signal that differs from said first setpoint trajectory only with respect those parts of said first setpoint which do not cause said system or process to exhibit said repetitive dynamics, wherein said feed-forward signal corresponding to said second setpoint signal is developed from said learned feed-forward signal corresponding to said first setpoint signal by inserting or removing values for said part of said second setpoint that differs from that of said first setpoint.

---

Description

---

BACKGROUND OF THE INVENTION

1. Priority Information

This application claims priority from European Patent Application No. 03077461.6, filed Aug. 6, 2003, herein incorporated by reference in its entirety.

2. Field of the Invention

The invention concerns a method of controlling a system or process using a learned feed-forward signal, a control system and a lithographic projection apparatus.

3. Description of the Related Art

The essential steps in the manufacturing process of integrated circuits (ICs) are performed by lithographic machines called wafer scanners or wafer steppers. Important modules of these machines are the wafer and reticle stages. According to current state of the art, a wafer stage is normally a motion system, which positions the silicon wafer with respect to the illumination optics with high precision. As ICs are becoming smaller, the required precision increases proportionally. In order to meet positioning specifications that are currently in the order of nanometers, they require careful design. This is not only the case for the mechanical construction, but also for the design of actuators, electronics, software, measurement and control systems, etc.

In control system design, such specifications could so far be met using basic, well known concepts such as conventional feedback with PID (position plus integral plus derivative control action) controllers and conventional feed-forward based on rigid body acceleration. In the future, these concepts may prove insufficient to achieve the desired tracking performance, for which more advanced designs will be needed. As, in the present case, the control objective is to move an object along a predefined trajectory, the control problem is said to be one of "tracking". The "tracking performance," as understood in the art, is the accuracy with which the predefined trajectory is followed.

There are different kinds of disturbance acting on the system, including disturbances introduced by the setpoint. That is, although a setpoint is known and deliberate, it will be appreciated that it, nevertheless, "disturbs" the steady state condition. Feedback and feed-forward control deal with the stability and performance (in terms of tracking error), but additional control techniques, for example, iterative learning control (ILC), provides options for significantly improving the tracking performance (compared to conventional feedback and feed-forward) of the processes or systems that execute the same trajectory, motion or operation repetitively.

When only conventional feedback and feed-forward are applied to the considered motion system, the servo error often appears to temporarily exceed the required specifications and, in most cases, it bounds on the allowed amplitude of tracking errors. It then takes a certain settling time for this error to settle within specifications again. Learned feed-forward control can improve the tracking performance and shorten the settling time. For motion systems that repeatedly perform the same movement, a feed-forward technique called Iterative Learning Control or ILC can be applied to improve system performance in terms of elimination of the repeating portion in the tracking error. For a detailed description of iterative learning control, reference is made to Tomizuka M., T. C. Tsao and K. K. Chen "Discrete domain analysis and synthesis of repetitive controllers" Proc. 1988, American Control Conference, June 1988, pp. 860 866.

ILC generates a learned feed-forward signal, effective for providing good tracking control performance. The basic principle behind ILC is that it exploits possibilities to incorporate past repetitive control information, such as tracking errors and control input signals into the synthesis of a new feed-forward signal. Past control information is stored and then used in the control action in order to ensure that the system meets the control specifications such as convergence of the servo error during the learning process. The final result after applying ILC is that the magnitude of the servo errors is relatively small, the periodic disturbances are suppressed or, in other words, the tracking performance is no longer deteriorated by periodic contributions. Such performance improvement allows for shortening settling times and thereby improving productivity of various processes.

SUMMARY OF THE INVENTION

Although ILC leads to a good tracking performance, several issues remain that prevent it from being widely used for industrial applications. Experience with ILC algorithms applied to various motion systems has shown that ILC has limited performance within the context of position-dependent dynamics. Further, understanding and controlling position-dependent dynamics (seen as a non-linear system) while motion tracking on different positions is a problem.

Furthermore, ILC does not account for setpoint trajectory changes. The learned feed-forward signal depends on the reference signal and therefore, for different motion profiles a new feed-forward must be issued each time the profile changes. In other words, a new iterative learning procedure must be carried out for each different profile. Despite a very good tracking error performance, completion of the ILC procedure and the implementation of a suitable feed-forward signal, which corresponds to a certain setpoint profile costs time. Studies conducted towards robustness of the learned feed-forward signal against set-point changes are known from "M. Steinbuch, Repetitive control for systems with uncertain period-time, Automatica, vol. 38(42), 2002", but they have not yet led to a satisfactory solution.

Finally, noise amplification can be a problem. In applying ILC, the periodic or deterministic errors are learned while however the stochastic or random effects (noise) and other non-repetitive disturbances are amplified.

For wafer scanning systems, the fact that the learned feed-forward signal is dependent on the setpoint constitutes the main drawback, since the setpoint is related to the frequently changing size of the die, which is illuminated in the scanning process.

For at least one of the issues identified above, the principles of the present invention, as embodied and broadly described herein, provide for method of controlling a system or process using a learned feed-forward signal and, in particular, an iterative learning control method that is applied to lithographic wafer scanning systems. In one embodiment, the method comprises executing a feed-forward control of the system or process according based on a first setpoint signal that is indicative of a first setpoint trajectory; determining an error signal between the first setpoint signal and an output signal of the system or process; performing a time-frequency analysis of the error signal; identifying frequency components and corresponding levels of the error signal as a function of time based on the time-frequency analysis; determining a time-varying bandwidth for a filter based on the identified frequency components as a function of time, such that repetitive dynamic attributes exhibited by the system or process are incorporated into the learned feed-forward signal at the appropriate time instants; applying the filter to the learned feed-forward signal; and generating a control signal in response to applying the filter.

By applying time-frequency analysis to a given standard ILC insight is gained into the dynamics of the considered system and into the learning process. In particular it shows the distribution of a parameter, such as energy or amplitude, of the error signal (i.e. against time and frequency) and reveals that the high level, for example, high energy levels, of the servo error signal is concentrated very locally in time.

The learning process has therefore less potential for improving tracking performance for the time-instants outside these intervals, where it only leads to amplification of noise, thus, the filter in the learned feed-forward path is adapted accordingly, to the effect that ILC is concentrated only within the areas of the setpoint trajectory where the high energy signals are concentrated.

Using this scheme an adaptive filter is introduced that adapts according to the momentary frequency content of the error signal. This leads to a time-frequency adaptive ILC that features a reduced noise amplification compared to standard ILC, while achieving an equivalent suppression of repetitive errors. This scheme moreover produces one learned feed-forward signal for a first setpoint trajectory that can be used for different setpoint trajectories. Further, the insights gained from this technique can be used to solve the problem of setpoint dependency of ILC.

In one embodiment, there is provided a method of controlling a system wherein the system or process is controlled over a further setpoint trajectory differing from the predetermined setpoint trajectory only with respect those parts of the predetermined setpoint trajectory which do not cause the system or process to exhibit the repetitive dynamics, wherein the feed-forward signal for the further setpoint trajectory is developed from the learned feed-forward signal for the predetermined setpoint trajectory by inserting or removing values for the part of the further setpoint trajectory differing from that of the predetermined setpoint trajectory.

Thus, for example, the motion system of a lithographic projection apparatus is seen to exhibit repetitive deterministic system dynamics at the beginning and at the end of the setpoint trajectory, that is predominantly during and shortly after acceleration and during deceleration, respectively. Between these acceleration and deceleration intervals, that is during the scan (i.e. when the stage is moved at constant velocity) the feed-forward signal may be given constant, typically zero, values. Using this insight, a feed-forward signal can be developed for further setpoint trajectories differing only from a first setpoint trajectory in terms of its scan length without having to perform learning again. The learned feed-forward signal for a first learned setpoint trajectory is "prolonged" or "shortened" by inserting or removing zeros, respectively, for the corresponding interval intermediate the acceleration and deceleration intervals, to develop a modified feed-forward signal corresponding to the required scan length.

In other words, a modified feed-forward signal is developed by "cutting" values from the first feed-forward signal in the interval of constant velocity or "pasting" in additional values in the interval of constant velocity corresponding to the difference in scan length between a first scan length and a further different scan length, to achieve a feed-forward signal for controlling the system or process over the further, different scan length.

BRIEF DESCRIPTION OF DRAWINGS

Embodiments of the invention will now be described, by way of example only, with reference to the accompanying schematic drawings in which:

FIG. 1a depicts a lithographic projection apparatus embodying the invention;

FIG. 1b depicts a typical setpoint trajectory and error signal to which the present invention may be applied;

FIG. 2 depicts a control block diagram according to an embodiment of the present invention;

FIG. 3 depicts further details of the controller depicted in FIG. 2;

FIGS. 4a and 4b depict a comparison between a standard learned feed-forward signal and a time adaptive learned feed-forward signal (in incremental units);

FIG. 5 depicts the learned feed-forward signal for different scan intervals before and after modifying according to the desired new scan length;

FIGS. 6a and 6b depicts a comparison between a time frequency adaptive ILC error signal and a standard ILC error signal;

FIG. 7a depicts two accelerations associated with two different scanning profiles for implementation in an embodiment of the present invention;

FIG. 7b depicts a graph showing the three dimensional Wigner distribution on the time frequency plane of an error signal;

FIG. 7c depicts a time interval divided into four pieces such that the cross terms are eliminated;

FIG. 7d depicts a graph showing a crossing contour when the Wigner distribution results shown in FIG. 4b are intersected in the three dimensional energy distribution plot with a c.sub.e height plane, where c.sub.e is a predetermined value; and

FIG. 8 depicts a flow charts showing the steps carried out according to an embodiment of the invention.

DETAILED DESCRIPTION OF THE INVENTION

FIG. 1a depicts a lithographic projection apparatus embodying the invention. The invention is not however limited in this respect. The present invention has application to any process or system, which is to be controlled over a setpoint trajectory.

In the following detailed description, which references lithographic systems as an example of an application of the present invention, the following remarks should be kept in mind: the term "patterning device" as here employed should be broadly interpreted as referring to a device that can be used to impart an incoming radiation beam with a patterned cross-section, corresponding to a pattern that is to be created in a target portion of the substrate; the term "light valve" can also be used in this context. Generally, the pattern will correspond to a particular functional layer in a device being created in the target portion, such as an integrated circuit or other device (see below). Examples of such patterning devices include: a mask: the concept of a mask is well known in lithography, and it includes mask types such as binary, alternating phase-shift, and attenuated phase-shift, as well as various hybrid mask types. Placement of such a mask in the radiation beam causes selective transmission (in the case of a transmission mask) or reflection (in the case of a reflective mask) of the radiation impinging on the mask, according to the pattern on the mask. In the case of a mask, the support structure will generally be a mask table/holder, which ensures that the mask can be held at a desired position in the incoming radiation beam, and that it can be moved relative to the beam if so desired; a programmable mirror array: one example of such a device is a matrix-addressable surface having a visco-elastic control layer and a reflective surface. The basic principle behind such an apparatus is that (for example) addressed areas of the reflective surface reflect incident light as diffracted light, whereas unaddressed areas reflect incident light as non-diffracted light. Using an appropriate filter, the non-diffracted light can be filtered out of the reflected beam, leaving only the diffracted light behind; in this manner, the beam becomes patterned according to the addressing pattern of the matrix-addressable surface. An alternative embodiment of a programmable mirror array employs a matrix arrangement of tiny mirrors, each of which can be individually tilted about an axis by applying a suitable localized electric field, or by employing piezoelectric actuation mechanism. Once again, the mirrors are matrix-addressable, such that addressed mirrors will reflect an incoming radiation beam in a different direction to unaddressed mirrors; in this manner, the reflected beam is patterned according to the addressing pattern of the matrix-addressable mirrors. The required matrix addressing can be performed using suitable electronic means. In both of the situations described here above, the patterning device can comprise one or more programmable mirror arrays. More information on mirror arrays as here referred to can be gleaned, for example, from U.S. Pat. Nos. 5,296,891 and 5,523,193, and PCT patent applications WO 98/38597 and WO 98/33096, which are incorporated herein by reference. In the case of a programmable mirror array, the support structure may be embodied as a frame or table, for example, which may be fixed or movable as required; and a programmable LCD array: an example of such a construction is given in U.S. Pat. No. 5,229,872, which is incorporated herein by reference. As above, the support structure in this case may be embodied as a frame or table, for example, which may be fixed or movable as required.

For purposes of simplicity, the rest of this text may, at certain locations, specifically direct itself to examples involving a mask and mask table/holder; however, the general principles discussed in such instances should be seen in the broader context of the patterning device as set forth here above.

For the sake of simplicity, the projection system may hereinafter be referred to as the "lens"; however, this term should be broadly interpreted as encompassing various types of projection system, including refractive optics, reflective optics, and catadioptric systems, for example. The radiation system may also include components operating according to any of these design types for directing, shaping or controlling the projection beam of radiation, and such components may also be referred to below, collectively or singularly, as a "lens". Further, the lithographic apparatus may be of a type having two or more substrate table/holders (and/or two or more mask table/holders). In such "multiple stage" devices the additional tables may be used in parallel, or preparatory steps may be carried out on one or more tables while one or more other tables are being used for exposures. Twin stage lithographic apparatus are described, for example, in U.S. Pat. No. 5,969,441 and U.S. Ser. No. 09/180,011, filed 27 Feb. 1998 (WO 98/40791), incorporated herein by reference.

In general, a single wafer will contain a whole network of adjacent target portions that are successively irradiated via the projection system, one at a time. In current apparatus, employing patterning by a mask on a mask table/holder, a distinction can be made between two different types of machine. In one type of lithographic projection apparatus, each target portion is irradiated by exposing the entire mask pattern onto the target portion in one go; such an apparatus is commonly referred to as a wafer stepper. In an alternative apparatus--commonly referred to as a step-and-scan apparatus--each target portion is irradiated by progressively scanning the mask pattern under the projection beam in a given reference direction (the "scanning" direction) while synchronously scanning the substrate table/holder parallel or anti-parallel to this direction. Since, in general, the projection system will have a magnification factor M (generally <1), the speed V at which the substrate table/holder is scanned will be a factor M times that at which the mask table/holder is scanned. More information with regard to lithographic devices as here described can be gleaned, for example, from U.S. Pat. No. 6,046,792, incorporated herein by reference.

In a manufacturing process using a lithographic projection apparatus according to the invention a pattern (e.g. in a mask) is imaged onto a substrate that is at least partially covered by a layer of energy-sensitive material (resist). Prior to this imaging step, the substrate may undergo various procedures, such as priming, resist coating and a soft bake. After exposure, the substrate may be subjected to other procedures, such as a post-exposure bake (PEB), development, a hard bake and measurement/inspection of the imaged features. This array of procedures is used as a basis to pattern an individual layer of a device, e.g. an IC. Such a patterned layer may then undergo various processes such as etching, ion-implantation (doping), metallization, oxidation, chemo-mechanical polishing, etc., all intended to finish off an individual layer.

If several layers are required, then the whole procedure, or a variant thereof, will have to be repeated for each new layer. Eventually, an array of devices will be present on the substrate (wafer). These devices are then separated from one another by a technique such as dicing or sawing, whence the individual devices can be mounted on a carrier, connected to pins, etc. Further information regarding such processes can be obtained, for example, from the book "Microchip Fabrication: A Practical Guide to Semiconductor Processing", Third Edition, by Peter van Zant, McGraw Hill Publishing Co., 1997, ISBN 0-07-067250-4, incorporated herein by reference.

For the sake of simplicity, the projection system may hereinafter be referred to as the "lens"; however, this term should be broadly interpreted as encompassing various types of projection system, including refractive optics, reflective optics, and catadioptric systems, for example. The illumination system may also include components operating according to any of these design types for directing, shaping or controlling the projection beam of radiation, and such components may also be referred to below, collectively or singularly, as a "lens". Further, the lithographic apparatus may be of a type having two or more substrate table/holders (and/or two or more mask tables). In such "multiple stage" devices the additional tables may be used in parallel, or preparatory steps may be carried out on one or more tables while one or more other tables are being used for exposures. Dual stage lithographic apparatus are described, for example, in U.S. Pat. No. 5,969,441 and WO 98/40791, both incorporated herein by reference.

Although specific reference may be made in this text to the use of the apparatus according to the invention in the manufacture of ICs, it should be explicitly understood that such an apparatus has many other possible applications. For example, it may be employed in the manufacture of integrated optical systems, guidance and detection patterns for magnetic domain memories, liquid-crystal display panels, thin-film magnetic heads, etc. The skilled artisan will appreciate that, in the context of such alternative applications, any use of the terms "reticle", "wafer" or "die" in this text should be considered as being replaced by the more general terms "mask", "substrate" and "target portion", respectively.

In the present document, the terms "radiation" and "beam" are used to encompass all types of electromagnetic radiation, including ultraviolet (UV) radiation (e.g. with a wavelength of 365, 248, 193, 157 or 126 nm) and extreme ultra-violet (EUV) radiation (e.g. having a wavelength in the range 5 20 nm), as well as particle beams, such as ion beams or electron beams.

The lithographic apparatus shown in FIG. 1a comprises: an illumination system Ex, IL: for supplying a projection beam PB of radiation (e.g. EUV, DUV, or UV radiation). In this particular case, the radiation system also comprises a radiation source LA; a first object table (mask table/holder/holder) MT: provided with a mask holder for holding a mask MA (e.g. a reticle), and connected to first positioning mechanism for accurately positioning the mask with respect to item PL; a second object table (substrate table/holder/holder) WT: provided with a substrate holder for holding a substrate W (e.g. a resist-coated silicon wafer), and connected to second positioning mechanism for accurately positioning the substrate with respect to item PL; a projection system ("lens") PL: for example, a mirror or refractive lens system that images an irradiated portion of the mask MA onto a target portion C (comprising one or more dies) of the substrate W; and a control mechanism CM: for controlling the position of either the first or second object table with respect to the projection beam or patterned beam, respectively. The control mechanism includes an iterative learning control mechanism ILC. The control mechanism is described in further detail hereinbelow with reference to FIGS. 2 to 8.

As depicted in FIG. 1, the apparatus is of a transmissive type (i.e. has a transmissive mask). However, as depicted in FIG. 2, it may also be of a reflective type, for example (with a reflective mask). Alternatively, the apparatus may employ another kind of patterning device, such as a programmable mirror array of a type as referred to above.

The source LA (e.g. a mercury lamp, a Krypton-Fluoride excimer laser or a plasma source) that produces a beam of radiation. This beam is fed into an illumination system (illuminator) IL, either directly or after being passed through conditioning mechanism, such as a beam expander Ex, for example. The illuminator IL may comprise adjusting mechanism AM for setting the outer and/or inner radial extent (commonly referred to as .sigma.-outer and .sigma.-inner, respectively) of the intensity distribution in the beam. In addition, it will generally comprise various other components, such as a condenser CO and an integrator IN. The integrator IN projects the incoming light into the condensor CO. The integrator IN may, for example, be formed of a quartz rod, and is used to improve the intensity distribution of the beam to be projected over the cross section of the beam. The integrator thus improves the illumination uniformity of the projection beam PB. In this way, the beam PB impinging on the mask MA has a desired uniformity and intensity distribution in its cross-section.

It should be noted with regard to FIG. 1a that the source LA may be within the housing of the lithographic projection apparatus (as is often the case when the source LA is a mercury lamp, for example), but that it may also be remote from the lithographic projection apparatus, the radiation beam which it produces being led into the apparatus (e.g. with the aid of suitable directing mirrors); this latter scenario is often the case when the source LA is an excimer laser. The current invention and claims encompass both of these scenarios.

The beam PB subsequently intercepts the mask MA, which is held on a mask table MT. Having traversed the mask MA, the beam PB passes through the lens PL, which focuses the beam PB onto a target portion C of the substrate W. With the aid of the second positioning mechanism PW (and interferometric measuring mechanism IF), the substrate table/holder WT can be moved accurately, e.g. so as to position different target portions C in the path of the beam PB. Similarly, the first positioning mechanism PM can be used to accurately position the mask MA with respect to the path of the beam PB, e.g. after mechanical retrieval of the mask MA from a mask library, or during a scan. In general, movement of the object tables MT, WT will be realized with the aid of a long-stroke module and a short-stroke module, which are not explicitly depicted in FIG. 1. However, in the case of a wafer stepper (as opposed to a step-and-scan apparatus) the mask table MT may just be connected to a short stroke actuator, or may be fixed. Mask MA and substrate W may be aligned using mask alignment marks M1, M2 and substrate alignment marks P1, P2.

The depicted apparatus can be used in different modes: step mode: the mask table MT is kept essentially stationary, and an entire mask image is projected in one go (i.e. a single "flash") onto a target portion C. The substrate table/holder WT is then shifted in the X and/or Y directions so that a different target portion C can be irradiated by the beam PB; scan mode; essentially the same scenario applies, except that a given target portion C is not exposed in a single "flash". Instead, the mask table MT is movable in a given direction (the so-called "scan direction", e.g. the Y-direction) with a speed .nu., so that the projection beam PB is caused to scan over a mask image; concurrently, the substrate table/holder WT is simultaneously moved in the same or opposite direction at a speed V=M .nu., in which M is the magnification of the lens PL (typically, M=1/4 or 1/5). In this manner, a relatively large target portion C can be exposed, without having to compromise on resolution; and other mode: the mask table MT is kept essentially stationary holding a programmable patterning device, and the substrate table/holder WT is moved or scanned while a pattern imparted to the projection beam is projected onto a target portion C. In this mode, generally a pulsed radiation source is employed and the programmable patterning device is updated as required after each movement of the substrate table/holder WT or in between successive radiation pulses during a scan. This mode of operation can be readily applied to maskless lithography that utilizes programmable patterning device, such as a programmable mirror array of a type as referred to above.

Controllers for controlling the position of either the support structure or the substrate table/holder are known. However, conventional controllers suffer various drawbacks. For example, with reference to lithographic apparatuses, the setpoint trajectory includes an exposure interval. During the exposure interval, the stage is controlled to move at a constant velocity for step and scan. Servo errors must lie within certain acceptable bounds for the exposure interval.

Using conventional feedback and feedforward control (i.e. non-iterative learning control techniques), it is known that position errors exceed specifications during and directly after acceleration to constant velocity, as denoted in FIG. 1b as time t1. To accommodate this in conventional apparatuses, a settling time is necessary to let the error settle to within the acceptable bounds before actual exposure starts. Waiting for the settle time limits productivity (throughput) and therefore, elimination of the settle time is a significant productivity benefit. Conventional control techniques fail to eliminate this problem.

Due to the reproducible character of the settle behaviour, iterative learning control (ILC) techniques are effective in eliminating this error contribution and therefore, help to establish a so-called "zero" settling (that is a zero settling time where exposure is started directly at the end of the acceleration phase). In iterative learning control, the learned feed-forward signal depends on the particular setpoint (the scanning profile). Thus, each time the scanning profile is changed, i.e. the setpoint r is changed, for example when the length of the illumination interval varies, the learning process has to be repeated and consequently, the learned feed-forward signal uk has to be redesigned. Thus, if scans with different lengths are to be carried out, but with identical acceleration phase of the trajectory, in conventional iterative learning control, learning must be repeated, which reduces the productivity since learning costs time and expertise during which wafer exposure cannot take place.

According to an aspect of the present invention, a technique for scan length independent control using iterative learning control is proposed, wherein the same learned feed-forward signal can be applied to a range of possible scan lengths provided that the acceleration phase remains unchanged. The present invention has application in a lithography apparatus to control the movement of at least one of the support structure and/or the substrate table/holder along a trajectory. In one embodiment, the movement of at least one of the support structure and/or the substrate table/holder is controlled according to a trajectory determined with respect to a fixed reference. In an alternative embodiment of the present invention, the support structure and the substrate table/holder are controlled to move simultaneously, wherein the trajectory of the support structure and the substrate table/holder is determined on the basis of a relative movement between the support structure and the substrate table/holder, wherein the learned feed-forward signal is used to control the relative movement of the support structure with respect to the substrate table/holder, so as to project the patterned beam onto a target portion of the substrate in a desired manner.

FIG. 1b shows a typical setpoint trajectory to which the present invention may be applied. A plot of a typical acceleration 10 associated with a particular scanning profile, i.e. the second derivative with respect to time of the scanning profile, is shown, wherein acceleration in m/s/s is plotted against time. During operation, the maximum acceleration is typically in the region of 5 m/s/s. FIG. 1b also shows a plot 20 of the associated servo error under conventional feedback and feed-forward control signal plotted against the same time period over which the scanning profile has been applied. In FIG. 1b, the acceleration profile is shown to provide information where the stage is moving with constant velocity, i.e. with zero acceleration. t1 denotes the end of the acceleration phase, the interval [t1,t2] denotes the scanning interval, where t_scan=t2-t1, that is the interval during which illumination takes place. The interval [t2,t3] denotes the deceleration phase, where t_deceleration=t3-t2. t3 represents the end of the set point, r. It can be seen from the plot 20 that the settling behaviour exhibited at the end of the acceleration phase under standard feed-forward control at time, t1, and at the beginning of the deceleration phase at time, t2, is not good.

In contrast, the present invention obtains very good servo errors (tracking errors) during the time interval 30 where the stage moves with constant velocity, that is within the time interval [t1,t2]. The motion system shows a settling behaviour, that is for a certain time after time instant t1, as shown by interval 40, the servo error is still large. For example, an illumination interval starting with t1=0.1 s having very high accuracy, for example, having a servo error smaller than 0.05.times.10.sup.-7 m, that is t=0.1 s to t=0.15 s, it is observed that approximately 0.15 s must be waited in order to be able to scan while achieving a servo error of less than 0.05.times.10.sup.-7 m. It has been found that using ILC in control of a stage in a lithographic apparatus improves the settling behaviour, that is the servo error is smaller in the beginning and end of the illumination interval. The error signal in FIG. 1b was obtained using standard acceleration feedforward control.

As noted above ILC is employed as a feed-forward control for motion systems that repeatedly perform the same movement to eliminate repeating or periodic deterministic positioning errors while shortening settling times. However, it has been found that ILC amplifies non-periodic deterministic disturbances and stochastic disturbances, such as noise, in the control signal. Further, the learning process takes up considerable processing time during which the process or system is not operational. For this reason, prior to the present invention, ILC was not considered as a viable option for the control of supports in lithographic apparatuses because in lithography control is required to take place over a variety of setpoints, corresponding to differing die or reticle image sizes. Prior to the present invention, the time it would take for the control system to learn the feed-forward signal for each die or reticle size was prohibitive.

The present invention provides a solution to the above problems. In a lithographic projection apparatus, the present invention has particular application to controlling either the wafer stage or the reticle stage or both. One of the components of the wafer scanner is the six degrees of freedom wafer stage. The wafer stage is an electro mechanical servo system that positions the wafer with respect to the imaging optics. Typically, the wafer has a diameter of 200 300 mm. The control of the wafer stage largely determines the throughput and the accuracy of the products and is subject to severe performance requirements. Typical throughputs are of the order of 80 100 wafers per hour, with approximately 80 200 integrated circuits per wafer. Typical scan speeds and decelerations are 0.6 meters per second and 10 meter per second per second, respectively. The positioning accuracy is in terms of nanometers and microradians. Such high accuracy is needed regarding the fine patterns to be produced. Typical feature dimensions are of the order of 180 250 nanometers. Moreover, various layers, typically 20 layers, of different patterns have to be aligned very accurately with respect to each other.

FIG. 2 depicts a control block diagram according to an embodiment of the present invention. In particular, FIG. 2 shows an iterative learning control system (ILC) including an adaptive filter according to the invention. A conventional iterative learning controller and the iterative learning controller of the present invention comprise two filters: a learning filter L and a robustness filter Q. The learning filter L of the present invention does not have to differ from a conventional learning filter. However, in conventional iterative learning controllers the robustness filter is fixed, in contrast to the present invention, where the robustness filter Q is an adaptive time varying filter as will be described below.

In FIG. 2, P indicates a process or a system to be controlled. The motion system includes the process to be controlled P, which for the lithographic application includes analog to digital and digital to analog converters, an actuator system and a sensor system, or in other words includes mechanics, software and electronics, an input of the plant, which in the lithographic application is a force which is going to move the system, and an output signal, which in the lithographic application gives the position when the system is moved. Therefore, the signal y gives, at each moment, the position of the plant to be controlled. A feedback controller C functions conventionally to provide a feedback control signal to the process P.

The signal y is supplied via a feed back loop to a summer 5. A setpoint signal r1 is supplied to the summer 5. The setpoint signal r may be defined as being the trajectory the motion system has to follow. And, although not essential, the setpoint signal r may be further defined as the motion the system has to follow to complete one movement from stand still to stand still. In the case of a lithographic apparatus, the setpoint signal r is the trajectory the substrate table/holder, that is the wafer or mask stage has to follow to make one exposure, wherein illumination takes place during the interval in which the substrate table/holder, such as the wafer or reticle stage, travels at constant velocity. In other words, the substrate table/holder has to follow the setpoint.

The summer 5 determines an error signal ek indicative of the deviation between the output signal y and the set point signal r1, that is ek=r-y. The error signal ek represents the servo error developed during the "kth" iteration. ILC derives a next value e(k+1), in which the undesired effects observed in the error signal ek are suppressed. Theoretically, when ILC is applied to a process in which only periodic deterministic disturbances occur, the servo error for increasing values of k reduces to zero. In practice, however, a variety of stochastic and non-periodic deterministic effects occur in the system, which cannot be learned due to their unpredictable nature. Thus, after applying the ILC procedure a number of times, it is observed that the servo error ek is not further improved after subsequent iterations. At this point in practice, the ILC procedure is stopped.

The ILC element 6 in FIG. 2 is designed to iteratively provide a feed-forward signal uk (where k is the trial index), which controls the repetitive disturbances in the servo error ek. Thus at iteration k, i.e. when the learned feed-forward signal uk is applied, the error ek is obtained and analysis element 2 analyzes the error signal ek obtained at iteration k.

The error signal ek is supplied to an input of a control circuit including feedback controller C. In addition, the error signal ek is supplied to the iterative learning control element 6, an analysis element 2, and a signal modification unit 8, which modifies the learned feed-forward signal according to a different setpoint signal r2. The feed-forward signal uk generated by the iterative learning control circuit is supplied via the summer 7 to the output of the feedback controller C, to control the process P.

The ILC element 6 in FIG. 2, as mentioned above, comprises two filters: a standard learning filter and an adaptive robustness filter. The learned feed-forward signal uk includes amplified stochastic and non-periodic deterministic disturbances. After iteration k, that is, after implementing the learned feed-forward signal uk, the analysis element 2 analyzes a parameter of the servo error ek. The parameter may be any parameter that yields information about the nature of the servo error, such as, for example, the energy or amplitude. In the examples discussed, the parameter of energy is taken. However, the invention not limited in this respect, and energy is taken by way of example only.

Analysis element 2 analyzes the parameter in order to distinguish where in time and frequency the deterministic energy is located in the error signal ek (which may be periodic or non-periodic) and where in time and frequency the stochastic effects are present in the error signal ek. That is, the analysis element 2 carries out time frequency analysis, which analyzes time locality and frequency spectra in a energy plot of the chosen parameter, for example an energy or amplitude plot, of the servo error signal ek, for a given predetermined value k. On the basis of this analysis, the stochastic and non-periodic deterministic disturbances, such as noise, are identified.

It has been found that the characteristics of the deterministic components differ from those characteristics of a stochastic disturbance component. In particular, an energy plot of the analyzed error signal ek, shows that the energy of a deterministic disturbance over a particular time interval differ from the energy of a stochastic disturbance over the same time interval. That is, the energy of the non-deterministic disturbance component has a smaller magnitude than the energy of the deterministic disturbance component. When the system is in operation, for example, when the stage is moving, the high energy deterministic components are concentrated around the jerk moments, which correspond primarily to the acceleration and deceleration intervals, at the beginning and the end, respectively, of the illumination interval, around the time instants t1 and t2 in FIGS. 1a and 7a. Thus, the analysis element 2 identifies where in time, there are deterministic components and where not. The adaptive filter Q filters the current feed-forward signal such that effectively ILC is active only when the deterministic components of the error signal are predominant.

The deterministic components include repetitive dynamic attributes. The expression "repetitive" is well known in the art of control engineering and is understood to mean that if the same task is applied to a system or process, the behavior of that system or process reproduces. The level of the repetitive dynamic may vary from relatively small to relatively large. The present invention has particular application to dealing with significant repetitive dynamics. By "significant" it is meant those dynamic attributes whose level is higher than those of random noise.

The adaptive filter Q effectively changes its bandwidth (cutoff frequency) as a function of time and the dynamics which cause certain deterministic components, including the repetitive dynamic attributes, of the error signal to be incorporated into the learning feed-forward controlled process at the appropriate time instants. That is, the filter opens to allow the repetitive dynamic attributes exhibited by the system or process to be incorporated into the learned feed-forward signal at the appropriate time instants for controlling the system or process.

It is noted that the deterministic, including repetitive dynamics may be high frequency dynamics. However, depending on the particular dynamic they may not be high frequency. The invention is not limited to dealing with high frequency dynamics. The adaptive filter Q is opened sufficiently (i.e. has a high cut off frequency) to allow repetitive dynamics to enter the learning process. In fact, the time frequency analysis reveals how high the deterministic contributions are at which moment in time, so as to open the filter just as much is as needed to learn what can be learned, and thereby minimize the deteriorating impact of unwanted noise amplification.

The bandwidth of the filter is reduced when the system dynamics do not exhibit high levels of deterministic components so that the stochastic components do not enter the learned feed-forward signal, and thus, it avoids noise amplification when applying learning feed-forward control. Therefore, the adaptation of the filter bandwidth will adjust the bandwidth frequency as a function of time (for a given setpoint, time is strongly related or coupled to position) to maximize the tracking performance while still maintaining a good noise performance. In particular, the adaptive filter Q is designed to filter out the stochastic components of the error signal, so that they do not enter the learned feed-forward signal. The adaptive filter Q may be realized in the form of a low pass filter whose cutoff frequency varies according to the current frequency content present in the error signal.

The output from the analysis element 2 is input to the adaptive filter Q. The filter is designed to have a predetermined threshold. In short, the adaptive filter identifies whether there is a something to be learned from the error signal. If, for example, the error signal is analyzed as including periodic deterministic errors, the filter will then "turn on" the learning process, that is the bandwidth of the adaptive filter Q is increased, thus ensuring that those periodic deterministic disturbances are learned in order for them to be corrected for.

If, on the other hand, the error signal is analyzed as including predominant stochastic or non-periodic deterministic disturbances, the filter will then "turn off ", or "turn down", the supply to the iterative learning filter L, since there is nothing to learn from these errors because they are random in the case of stochastic errors, or non repeating, as in the case of the non-periodic deterministic disturbances. It is desired to keep the deterministic components, in particular the periodic disturbance components in the ILC error signal ek, because, as mentioned previously, learned feed-forward signal uk is adapted to suppress or eliminate the periodic disturbances. In particular, the information in the error signal ek aids the construction of an effective feed-forward signal, such that the repetitive errors can be eliminated. The adaptive ILC of the present invention, generates a learned feed-forward signal uk which compensates in an intelligent way for the repeating disturbances: when there are deterministic components in the disturbances to be learned, the feed-forward signal uk accounts for them, when there are mainly stochastic components the feed-forward signal uk does not learn them, and thus, it does not amplify them. By analyzing the components of the disturbances in the error signal ek, a