Alignment systems and methods for lithographic systems
An alignment system for a lithographic apparatus has a source of alignment radiation; a detection system that has a first detector channel and a second detector channel; and a position determining unit in communication with the detection system. The position determining unit is constructed to process information from said first and second detector channels in a combination to determine a position of an alignment mark on a work piece, the combination taking into account a manufacturing process of the work piece. A lithographic apparatus has the above mentioned alignment system. Methods of alignment and manufacturing devices with a lithographic apparatus use the above alignment system and lithographic apparatus, respectively.
Latest ASML NETHERLANDS B.V. Patents:
- Method of using a dual stage lithographic apparatus and lithographic apparatus
- DIFFRACTION GRATING FOR MEASUREMENTS IN EUV-EXPOSURE APPARATUSES
- METROLOGY METHOD AND APPARATUS, COMPUTER PROGRAM AND LITHOGRAPHIC SYSTEM
- STRUCTURES FOR USE ON A SUBSTRATE HOLDER, SUBSTRATE HOLDER, LITHOGRAPHIC APPARATUS AND METHOD
- ETCHING SYSTEMS, MODELS, AND MANUFACTURING PROCESSES
This application is a divisional of U.S. application Ser. No. 10/665,404, filed Sep. 22, 2003, which claims priority to U.S. Provisional Application No.60/411,861, filed Sep. 20, 2002, U.S. Provisional Application No. 60/413,601, filed Sep. 26, 2002, European Application No. 03075954.2, filed Apr. 1, 2003 and European Application No. 03076422.9, filed May 12, 2003. The entire contents of these applications are incorporated herein by reference.
BACKGROUND OF THE INVENTION1. Field of Invention
This invention relates to an alignment system for a lithographic projection apparatus, and a lithographic projection apparatus having such an alignment system, and more particularly to an alignment system that can detect the position of an alignment mark using at least two separate signals detected substantially in parallel and/or detect the position of a multi-target mark.
2. Discussion of Related Art
Lithographic projection apparatuses are essential components for the manufacture of integrated circuits and/or other microdevices. With the aid of such an apparatus, a number of masks having different mask patterns are successively imaged at a precisely aligned position onto a substrate such as a semiconductor wafer or an LCD panel. The substrate must undergo the desired physical and chemical changes between the successive images that have been aligned with each other. The substrate is removed from the apparatus after it has been exposed with a mask pattern, and, after it has undergone the desired process steps, the substrate is replaced in order to expose it with an image of a second mask pattern, and so forth, while it must be ensured that the images of the second mask pattern and the subsequent mask patterns are positioned accurately with respect to the substrate. To this end, the lithographic projection apparatus is provided with an optical alignment system with which alignment marks on the substrate are aligned with respect to alignment marks on the mask.
A lithographic apparatus may not only be used for the manufacture of ICs but also for the manufacture of other structures having detailed dimensions of the order of 1 micrometer, or smaller. Examples are structures of integrated, or plenary, optical systems or guiding and detection patterns of magnetic domain memories, micro-electromechanical systems (MEMS), and structures of liquid crystal display panels. Also in the manufacture of these structures, images of mask patterns must be aligned very accurately with respect to a substrate.
The lithographic projection apparatus may be a stepping apparatus or a step-and-scan apparatus. In a stepping apparatus, the mask pattern is imaged in one shot on an IC area of the substrate. Subsequently, the substrate is moved with respect to the mask in such a way that a subsequent IC area will be situated under the mask pattern and the projection lens system and the mask pattern is imaged on the subsequent IC area. This process is repeated until all IC areas of the substrate are provided with a mask pattern image. In a step-and-scan apparatus, the above-mentioned stepping procedure is also followed, but the mask pattern is not imaged in one shot, but via scanning movement. During imaging of the mask pattern, the substrate is moved synchronously with the mask with respect to the projection system and the projection beam, taking the magnification of the projection system into account. A series of juxtaposed partial images of consecutively exposed parts of the mask pattern is imaged in an IC area. After the mask pattern has been completely imaged in an IC area, a step is made to a subsequent IC area. A possible scanning procedure is described in the article: “Sub-micron 1:1 Optical Lithography” by D. A. Markle in the magazine “Semiconductors International” of May 1986, pp. 137-142.
U.S. Pat. No. 5,243,195 discloses an optical lithographic projection apparatus provided with an alignment system and intended for the manufacture of ICs. This alignment system comprises an off-axis alignment unit for aligning a substrate alignment mark with respect to this alignment unit. In addition, this alignment system comprises a second alignment unit for aligning a substrate mark with respect to a mask mark via the projection lens (TTL). Alignment via the projection lens (on-axis alignment) is frequently used in many current generation of optical lithographic projection apparatuses and provides the advantage that the substrate and the mask can be aligned directly with respect to each other. When the off-axis alignment method is used, the baseline offset as described in U.S. Pat. No. 5,243,195 must be taken into account. However, with the continued decrease in the size of components on ICs and the increase in complexity, on-axis alignment systems have proven to be difficult to improve sufficiently to achieve the require precision and accuracy.
In connection with the increasing number of electronic components per unit of surface area of the substrate and the resultant smaller dimensions of these components, increasingly stricter requirements are imposed on the accuracy with which integrated circuits are made. The positions where the successive masks are imaged on the substrate must therefore be fixed more and more accurately. In the manufacture of new-generation ICs with smaller line widths, the alignment accuracy will have to be improved or, in other words, it must be possible to detect smaller deviations so that the resolving power of the alignment system must be increased. On the other hand, stricter requirements must also be imposed on the flatness of the substrate due to the required higher numerical aperture (NA) of the projection lens system in the case of decreasing line widths. The depth of focus of this system decreases as the NA increases. Since some image field curvature occurs at the desired relatively large image field of the projection lens system, there is hardly any room left for unevenness of the substrate. To obtain the desired flatness of the substrate, it has been proposed to polish the substrate by the chemical mechanical polishing (CMP) process between two consecutive exposures with different mask patterns in the projection apparatus. However, this polishing process affects the accuracy of the on-axis alignment method. In this method, a grating is used as a substrate alignment mark and the sub-beams diffracted in the first order by this grating are used for imaging the substrate mark on the reticle mark. In this process, it is assumed that the substrate is aligned correctly with respect to the reticle when the point of gravity of the substrate grating mark is aligned with respect to the point of gravity of the reticle alignment mark. In that case it has been assumed that the point of gravity for each grating mark coincides with the geometrical center of the grating. However, the CMP process renders the substrate grating mark asymmetrical so that this alignment method is no longer reliable. In addition, the various processing steps contribute to changes in the alignment marks including introducing asymmetries and changes in the effective depth of the grooves of the substrate grating marks. Other processing steps and/or methods often introduce different types of errors. For example, the Cu-damascene process tends to introduce alignment errors in a random distribution across an IC surface. With the decrease in size and increase in complexity of structures built by lithographic techniques there is a continued demand to improve alignment accuracy. Without the improvement of alignment accuracy, improvements in resolution cannot be utilized. In addition, the increased complexity of the micro-devices places greater demand for techniques to control and minimize the number of substrates that have to be discarded during the manufacturing process due to alignment errors.
SUMMARYIt is thus an object of the present invention to provide an alignment system for a lithographic projection apparatus that has improved alignment accuracy and/or robustness.
In order to achieve these and other objectives of this invention an alignment system for a lithographic apparatus is provided with a source of alignment radiation; a detection system comprising a first detector channel and a second detector channel; and a position determining unit in communication with said detection system. The position determining unit processes information from the first and second detector channels in combination to determine a position of an alignment mark on a first object relative to a reference position on a second object based on the combined information.
According to another embodiment of this invention, a lithographic projection apparatus has a source of illumination radiation; a substrate stage assembly arranged in a radiation path of the source of illumination radiation; a reticle stage assembly arranged in the radiation path of the source of illumination radiation between the source and the substrate stage assembly; a projection system arranged between the reticle stage assembly and the substrate stage assembly; and an off-axial alignment system arranged adjacent to the projection system and proximate the substrate stage assembly. The off-axial alignment system has a source of alignment radiation; a detection system comprising a first detector channel and a second detector channel; and a position determining unit in communication with the detection system. The position determining unit processes information from the first and second detector channels in combination to determine a position of an alignment mark on a first object relative to a reference position on a second object based on the combined information. The alignment system may be located away from said radiation path of illumination radiation. All that is required is that alignment radiation from the alignment system is able to reach the substrate stage assembly.
According to another embodiment of this invention, a method of aligning a workpiece for the manufacture of a microdevice includes forming a multi-target alignment mark on the workpiece; scanning the multi-target alignment mark with an alignment system having a plurality of detectors, wherein a first target of the multi-target alignment mark is adapted to be detected by a first detector of the plurality of detectors of the alignment system, and a second target of the multi-target alignment mark is adapted to be detected from a second detector of the plurality of detectors of the alignment system; receiving a first signal from the first detector of the alignment system in response to the first target of the multi-target alignment mark; receiving a second signal from the second detector of the alignment system in response to the second target of the multi-target alignment mark; and determining a position of the alignment mark on the workpiece relative to a predetermined reference position based on information from the first and second signals.
According to another embodiment of this invention a method of capturing an alignment mark on a workpiece for the manufacture of a microdevice within a measurement region includes forming a multi-grating alignment mark on the workpiece; scanning the multi-grating alignment mark with an alignment system having a plurality of detectors; selecting first and second gratings from said multi-grating alignment mark; comparing a first substantially periodic signal from the first grating of the multi-grating alignment mark from a first one of the plurality of detectors to a second substantially periodic signal from the second grating of the multi-grating alignment mark from a second one of the plurality of detectors; and determining a capture range based on the comparing. The first grating can be a diffraction-order-enhancing grating that enhances an order greater than first order and the second grating can be a diffraction-order-enhancing grating that enhances an order greater than first order.
According to another embodiment of this invention, an alignment mark for use in the manufacture of a micro-device has a first target that has a first detection pattern; and a second target that has a second detection pattern. The first target is adapted to be detected by a first detector and the second target is adapted to be detected by a second detector.
According to another embodiment of this invention, a diffraction-order-enhancing alignment mark for use in the manufacture of a micro-device has a diffraction grating pattern that has a periodic pattern and a sub-pattern. The diffraction-order-enhancing grating enhances a strength of a diffracted beam of non-zero order.
According to another embodiment of this invention, an alignment mark for use in the manufacture of a microdevice has a target that has a detection pattern and a processing pattern. The processing pattern has a structure that changes under micro-device processing in correspondence with changes to said micro-device during manufacture.
BRIEF DESCRIPTION OF THE DRAWINGS
Methods and devices according to this invention will now be described with reference to particular embodiments by way of example. The broad concepts of this invention are not limited to only these specifically described embodiments. The invention will be described with reference to an alignment system for a photolithography system that includes both an on-axis (also referred to as “axial”) and an off-axis (“off-axial”) alignment system that can be used in combination to obtain the eventual alignment of a mask with respect to a substrate (“workpiece”). The axial alignment system may have a separate source of radiation to illuminate alignment marks, such as in through-the-lens (TTL) or through-the-reticle (TTR) systems, or it may employ the same radiation as the exposure radiation. The following example will describe a TTL system in combination with an off-axial system (OAS) as an embodiment of this invention. Furthermore, the invention envisions application to photolithography systems that have refraction projection systems as well as to other types of lithography systems that use shorter wavelengths of electromagnetic radiation than currently employed, systems which use reflective and/or diffraction imaging optics, and/or systems which use other types of radiation such as charged-particle beams, e.g., electron beams that are imaged with magnetic, electromagnetic, and/or electrostatic imaging optics. Embodiments of this invention also envision integrating the alignment process of the lithography systems with other components of an Automated Process Control (APC) system such as a metrology tool that is used to measure the accuracy of an exposure of a photo-resist prior to further processing.
We now describe an optical lithographic projection apparatus that has an on-axis alignment unit and other measuring systems as an example of a system that may incorporate an embodiment of the instant invention.
The apparatus further comprises an illumination system which is provided with a radiation source LA, for example a Krypton-Fluoride Excimer laser or a mercury lamp, a lens system LS, a mirror RE and a condenser lens CO. The projection beam PB supplied by the illumination system illuminates the mask pattern C. This pattern is imaged by the projection lens system PL on an IC area of the substrate W. The projection lens system has, for example, a magnification M=¼, a numerical aperture NA=0.6 and a diffraction-limited image field with a diameter of 22 mm.
The apparatus is further provided with a number of measuring systems, namely a system for aligning the mask MA and the substrate W with respect to each other in the XY plane, an interferometer system for determining the position and orientation of the substrate holder and hence of the substrate, and a focus error detection system for determining a difference between the focal or image plane of the projection lens system PL and the surface of the substrate W. These measuring systems are parts of servo systems which comprise electronic signal-processing and control circuits and drivers, or actuators, with which the position and orientation of the substrate and the focusing can be corrected with reference to the signals supplied by the measuring systems.
The alignment system makes use of two alignment marks M1 and M2 in the mask MA shown in the top right part of
The second alignment beam b′ is reflected by a mirror 31 towards a reflector 29 in the projection lens system PL. This reflector sends the beam b′ to a second reflecting surface 28 of the prism 26, which surface directs the beam b′ onto the substrate alignment mark P1. This mark reflects a part of the radiation of the beams b′ as beam b′1 to the mask alignment mark M1 where an image of the mark P1 is formed. The radiation of the beam b′1 passing through the mark M1 is directed by a reflector 11′ towards a radiation sensitive detector 13′.
The structure of the system AS1 will now be described, as well as the way in which the mutual position of the mask mark M2 and, for example, the substrate mark P2 is determined with this system.
The alignment system AS1 comprises a radiation source 1, for example a Helium-Neon laser emitting an alignment beam b. This beam is reflected to the substrate W by a beam splitter 2. The beam splitter may consist of a semi-transparent mirror or a semitransparent prism but is preferably constituted by a polarization-sensitive splitting prism 2 preceding a λ/4 plate 3, in which λ is the wavelength of the beam b. The projection lens system PL focuses the beam b in a small radiation spot V, having a diameter of the order of 1 mm, on the substrate W. This substrate reflects a part of the beam as beam b1 towards the mask MA. The beam b1 traverses the projection lens system PL, which system images the radiation spot V on the mask. Before the substrate is arranged in the projection apparatus, it has been pre-aligned in a pre-alignment station coupled to the apparatus, for example the station described in EP Patent Application 0 164 165, such that the radiation spot V is located on the substrate mark P2. This mark is then imaged on the mask mark M2 by the beam b1. With the magnification M of the projection lens system being taken into account, the dimension of the mask mark M2 is adapted to that of the substrate mark P2 so that the image of the mark P2 coincides accurately with the mark M2 when the two marks are mutually positioned correctly.
On its path to and from the substrate W, the beams b and b1 have traversed the λ/4 plate 4 twice, whose optical axis is at an angle of 45° to the direction of polarization of the linearly polarized beam b coming from the source 1. The beam b1 passing through the λ/4 plate then has a direction of polarization which is rotated 90° with respect to the beam b so that the beam b1 is passed by the polarization splitting prism. The use of the polarization splitting prism in combination with the λ/4 plate provides the advantage of a minimal radiation loss when coupling the alignment beam into the radiation path of the alignment system.
The beam b1 passed by the alignment mark M2 is reflected by a prism 11 and directed, for example by a further reflecting prism 12 towards a radiation-sensitive detector 13. This detector is, for example a composite photodiode having, for example four separate radiation-sensitive areas in conformity with the number of sub-gratings shown in
A beam splitter 14 in the form of, for example a partially transparent prism splitting a portion of the beam b1 as beam b2 may be arranged between the prism 11 and the detector 13. The split beam is incident via, for example two lenses 15 and 16 on a television camera 17 which is coupled to a monitor (not shown) on which the alignment marks P2 and M2 are visible to an operator of the projection apparatus. This operator can then ascertain whether the two marks coincide and move the substrate W by means of manipulators so as to cause the marks to coincide.
Analogously as described hereinbefore for the marks M2 and P2, the marks M1 and P2 and the marks M1 and P1 can be aligned with respect to each other. The alignment system AS2 is used for the last-mentioned two alignments.
For further particulars about the construction and the alignment procedure of the double alignment unit, reference is made to U.S. Pat. No. 4,778,275, which is incorporated herein by reference.
The embodiment of the on-axis alignment unit shown in
Since the projection lens system is designed for the wavelength of the projection beam PB, differences occur when this system PL is used for imaging the alignment marks P1, P2 and M1 and M2 on each other by means of the alignment beam. For example, the substrate alignment marks will not be situated in the plane of the mask pattern in which the mask alignment marks are situated, but will be imaged at a given distance therefrom, which distance depends on the difference between the wavelength of the projection beam and the alignment beam and the difference between the refractive indices of the material of the projection lens elements for the two wavelengths. If the projection beam has a wavelength of, for example 248 nm and the alignment beam has a wavelength of 633 nm, this distance may be 2 mm. Moreover, due to said wavelength difference, a substrate alignment mark is imaged on a mask alignment mark with a magnification which differs from the desired magnification and increases with an increasing wavelength difference.
To correct for said differences, the projection lens system PL incorporates an extra lens, a correction lens, 25. The correction lens is arranged at such a height in the projection lens that, on the one hand, in the plane of the correction lens the sub-beams of the different diffraction orders of the alignment beam, which sub-beams originate from, and are also generated by, a substrate alignment mark, are sufficiently separated to be able to influence these sub-beams separately and, on the other hand, this correction lens has a negligible influence on the projection beam and the mask pattern image formed with it. The correction lens is preferably situated in the Fourier plane of the projection lens system. If, as is shown in
If desired, a wedge or a different deflection element such as a diffraction element may be arranged in the path of the alignment beam(s) proximate to an alignment mark. With such a deflection element, not shown in
In addition to the global alignment marks P1 and P2, shown in
The projection apparatus further comprises a focus error detection system for determining a deviation between the focal plane of the projection lens system PL and the surface of the substrate W, so that this deviation can be corrected, for example by moving the projection lens system along its axis, the Z axis. This system may be constituted by the elements 40, 41, 42, 43, 44, 45 and 46 which are arranged in a holder (not shown) which is fixedly connected to the projection lens system. The reference numeral 40 denotes a radiation source, for example a diode laser, which emits a focusing beam b2. This beam is directed at a small angle onto the substrate by a reflecting prism 42. The beam reflected by the substrate is directed towards a retroreflector 44 by the prism 43. The element 44 reflects the beam in itself so that the beam (b3′) once again traverses the same path via reflections on the prism 43, the substrate W and the prism 42.
The beam b3′ reaches a radiation-sensitive detection system 46 via a partially reflecting element 41 and a reflecting element 45. This detection system consists of, for example a position-dependent detector, or of two separate detectors. The position of the radiation spot formed by the beam b3′ on this system is dependent on the extent to which the focal plane of the projection lens system coincides with the plane of the substrate W. For an extensive description of the focus error detection system, reference is made to U.S. Pat. No. 4,356,392.
For accurately determining the X and Y positions of the substrate holder, a stepping projection apparatus is provided with a multi-axis interferometer system. U.S. Pat. No. 4,251,160 describes a system with two measuring axes and U.S. Pat. No. 4,737,283 describes a system with three measuring axes. In
As is diagrammatically shown in
With an interferometer system, which has not only the X measuring axis shown in
If, as shown in
If the positions in the X and Y directions of the mask are represented by Xr, Yr and those of the substrate by Xw, Yw and the rotation about the Z axis by Φr,r and φz,w, then the following conditions are satisfied when the mask and the substrate are correctly positioned with respect to each other:
Xx−M·Xr=0 (1)
Yw−M·Yr=0 (2)
Φz,w−φz,r=0 (3)
in which M is the magnification of the projection lens system. It has been assumed that the mask and the substrate move in opposite directions. If these elements move in the same direction, the minus sign preceding M in the above conditions should be replaced by a plus sign.
To ascertain whether these conditions have been met, it is sufficient that both the interferometer system for the substrate and that for the mask have three measuring axes.
However, the substrate interferometer system preferably has five measuring axes. Then, not only Xw, Yw and φz,w but also φx,w and φy,w, i.e. the tilts about the X axis and the Y axis can be measured.
To be able to measure such tilts of the mask, a five-axis mask interferometer system may be used, or a combination of a three-axis interferometer system for determining Xr, Yr and φz,r and other sensors such as capacitive sensors for the φx,r and φy,r measurements.
If Xw, Yw, φx,w, and φy,w and Xr, Yr, φz,r, φx,r, φy,r and, with the aid of the focus error detection system, Zw and Zr, i.e. the positions along the Z axis of the substrate and the mask can be measured, it can be ascertained whether not only the conditions (1), (2) and (3) are met, but also the conditions:
M2·Zw−Zr=0 (4)
M·φx,w−φx,r=0 (5)
M·φy,w−φy,r=0 (6)
The on-axis alignment unit, described with reference to
All of these problems can be solved by making use of an off-axis alignment unit for aligning the substrate mark and by using higher-order sub-beams, i.e. sub-beams having a diffraction order which is higher than 1, in the alignment. Since the alignment of the substrate mark no longer takes place through the projection lens system, there will be greater freedom to use more sub-beams, particularly higher-order sub-beams. Since the resolving power of the alignment unit increases with an increasing order number of the sub-beams, the accuracy of the alignment can be enhanced considerably. Since notably the higher-order sub-beams are determined by the edges of the substrate grating mark and, as compared with the center of the grating, these edges are less influenced by said CMP process and other measures affecting the symmetry of the grating, the problem of asymmetry in the grating mark is largely eliminated. Moreover, it is also possible to use alignment radiation with more than one wavelength so that the requirements imposed on the depth of the grating grooves can be alleviated considerably.
As will be elucidated hereinafter, the diffraction orders are separated from each other by optical elements in the alignment unit according to the invention, not by electronic means and/or associated software. Consequently, it is not necessary to measure signal amplitudes but the phase measurements which are more conventional in these kinds of techniques can be used.
wherein N is the diffraction order number and P the grating period.
The path of the sub-beams reflected by the grating incorporates a lens system L1 which converts the different directions of the sub-beams into different positions un of these sub-beams in a plane 73:
Un=f1·αn (8)
In this plane, means are provided for further separating the different sub-beams. To this end, a plate may be arranged in this plane; which is provided with deflection elements in the form of, for example wedges. In
A second lens system L2 is arranged behind the wedge plate. This lens system images the mark P1 in the plane of reference plate RGP. In the absence of the wedge plate, all sub-beams would be superimposed in the reference plane. Since the different sub-beams through the wedge plate are deflected at different angles, the images formed by the sub-beams reach different positions in the reference plane. These positions Xn are given by
Xn=f2·Yn (9)
in which γ is the angle at which a sub-beam is deflected by the wedge plate.
At these positions, reference gratings G90-G96 can be provided, as is shown in
Hitherto only one set of diffraction orders has been considered. As is known, a diffraction grating forms, in addition to +1, +2, +3 etc order sub-beams, also sub-beams of diffraction orders −1, −2, −3 etc. Both the plus orders and the minus orders sub-beams can be used to form the grating image, i.e. a first image of the grating mark is formed by the +1 and −1 order sub-beams jointly, a second image is formed by the +2 and −2 order sub-beams jointly, and so forth. For the +1 order and the −1 order sub-beams no wedges need to be used, but plane-parallel plates which compensate for path-length differences can be provided at the positions of these sub-beams in the plane of the wedge plate. Thus six wedges, both for the plus orders and for the minus orders, are required for the orders 2-7.
In order that in the embodiment of
To lessen said requirements and to release the tolerances of the alignment unit, preferably use is made of the structure of deflection elements shown in
On its way through the plates a sub-beam will encounter, zero, one or two openings. Only the first order sub-beams encounters zero openings and is not deflected by any of the plates. In
In the embodiment of
In the embodiment of
The minimum number of diffraction orders which has to be detected to be able to align in a sufficiently accurate way at a given asymmetry of the substrate mark P1 can be determined by means of computer simulations. Such simulations have proved that, for example an alignment error of 150 nm which remains when a first-order sub-beam is used can be reduced to 20 nm when a fifth-order sub-beam is used.
In principle, the maximum number of orders which can be detected is determined by the minimum intensity which can still be detected and by the numerical aperture of the lens system L1, L2. As is known, the intensity of the sub-beam formed by a diffraction grating quickly decreases with an increase of the order number of this sub-beam; the intensity of a sub-beam is inversely proportional to the square of the order number of this sub-beam. For a seventh-order sub-beam, the intensity is thus approximately 1/50 of that of a first-order sub-beam. The intensity loss due to reflections undergone by an alignment beam when traversing the off-axis alignment unit is, however, considerably smaller than when it traverses an on-axis alignment unit. In the last-mentioned unit, the alignment beam meets, for example approximately one hundred surfaces on which reflection losses may occur and in the first-mentioned unit it meets, for example only twenty of these surfaces. If the total reflection loss is a factor of four in the off-axis alignment unit, the 7-order alignment sub-beam may have as much intensity as a 1-order alignment beam in the on-axis alignment unit.
The numerical aperture NAn which the optical system L1, L2 must have to pass a sub-beam with a diffraction order of N is given by:
For a 7-order sub-beam and a substrate grating mark with a grating period p=16 μm and a wavelength λ=544 nm, the desired numerical aperture is approximately 0.24, which is a very acceptable number.
To guarantee a sufficiently stable system, the different reference gratings are provided on a single plate RGP which preferably consists of quartz. The dimensions of this plate, hence the image field of the second lens system, are determined by the dimension d1 of the reference gratings and their mutual distance d2. This distance and dimension are, for example, both 0.2 mm so that the dimensions dx and dy in the X and Y directions of the plate RGP are 2.8 mm and the desired field diameter is approximately 3 mm.
The discrete wedges of the embodiment of
In order that not too stringent requirements have to be imposed on the groove depth of the substrate mark, alignment radiation having two wavelengths, for example 633 nm and 532 nm, have been found to be suitable. Use can be made of the fact that the angles at which the alignment grating deflects the sub-beams and the positions which these beams occupy in the rear focal plane of the lens system L1 is dependent on the wavelength, as is apparent from the expressions (7) and (8). In principle, the orders for the different wavelengths can be distinguished from each other. Without further measures, however, a given order, for example the second order of the first wavelength (633 nm) may come between, for example the second and third orders of the second wavelength (532 nm). To separate the orders of the different wavelengths better from each other, it can be ensured that the beams with the different wavelengths are incident at different angles on the substrate grating P1. For the case where seven diffraction orders are used, the situation as shown in
An alternative for the alignment with two wavelengths is illustrated in
The second beam b11 is reflected by the reflector 216 towards the beam-splitting layer 213 which reflects the beam towards the quarter-wave plate 212. After the beam b11 has passed this plate it is reflected by the reflective layer 212 at the backside of this plate, so that it passes for a second time through the plate 211. The beam b12 leaving the plate 211 has a polarization direction which is rotated over 90° with respect to the polarization direction of the original beam b11. The beam b12 can pass the beam splitting layer 213 and reach the grating mark P1. The radiation reflected by this mark is also indicated by a single beam ray b16. This beam passes first the beam-splitting layer 213, then traverses twice the quarter-wave plate 211 and finally is reflected by the layer 213 towards the associated structure of wedges and detectors, not shown in
In the embodiments of
In the different embodiments described above the detectors are arranged directly behind the reference gratings. In practice however, behind the reference gratings a bundle of imaging fibers may be arranged which image each of reference gratings and the superposed image of the substrate grating mark at a detector at a remote location, which is more convenient with respect to the design of the whole apparatus and with respect to the performance of this apparatus. For example, cross-talk between the images formed by the sub-beams of the different diffraction orders can be decreased and heat generated by signal amplifiers and electronic processors can be kept away from the alignment unit and the apparatus. Also the radiation sources may be arranged at positions remote from the alignment unit and their radiation can also be guided to the unit by an illumination bundle of fibers. In this way the heat generated by the radiation sources can be kept away from the alignment unit and the projection apparatus.
Between the prism 216 and the second lens system L2 for one of the beams b15 and b17 a partially transmitting reflector may be arranged to split off a portion of this beam towards a camera which, together with a monitor, provides a visual image of the substrate mark to an operator of the apparatus.
There are different possibilities of using the various detector signals. A start may be made with the alignment by means of the first-order sub-beams by processing the signals of the detectors associated with these sub-beams. Subsequently, the signals of the detectors associated with the second-order sub-beams may be used for finer alignment, then the signals of the detectors associated with the third-order sub-beams may be used for even finer alignment, and so forth. This may continue as long as the sub-beams used still have sufficient intensity to be detected in a reliable manner.
Another possibility is based on the recognition that the intensity of certain diffraction orders is increased at the expense of other diffraction orders when given process layers are provided on the substrate. In that case, a direct choice of the preferred orders may be made for the alignment. Under circumstances, said possibilities may also be combined.
It is also possible to calibrate the alignment unit before a batch of substrates is illuminated with a mask pattern or at the beginning of a production day. For a number of positions of the substrate mark the detector signals for each of the diffraction orders are measured. The results of these measurements are stored in graphs or tables showing for each position of the substrate mark the value of the detector signal for each diffraction order. During illumination of the substrates, the alignment measurement can be performed by measuring only the relatively large detector signals of the lower diffraction orders, for example the first three orders. By interpolation the corresponding value for a higher diffraction order, for example the seventh order, can be determined. In this way it is possible to determine alignment errors with high resolution and large signal amplitude.
Hitherto, only the alignment of the substrate with respect to an apparatus reference in the form of reference gratings has been described. With the same alignment unit also the position of the substrate holder or table can be determined. To that end this holder or table is provided with an alignment mark similar to the substrate alignment mark. (See, e.g., the fiducial mark illustrated schematically in
Another possibility of aligning the mask pattern with respect to the substrate table is the use of the image sensor unit described in, for example U.S. Pat. No. 4,540,277. In such a unit, a mask alignment mark is imaged by means of projection radiation on a corresponding and transmissive reference mark in the substrate table. In this table, a detector may be arranged behind the reference mark for converting the radiation passed by the reference mark into an electric signal. In the first instance, this image sensor unit is intended for, for example calibrating an on-axis alignment unit which operates with alignment radiation having a wavelength which is considerably different from that of the projection radiation, or for checking the image quality of the image formed by the projection lens system and for measuring distortions and aberrations which may occur, but it is also eminently suitable for aligning the mask pattern with respect to the substrate table. Instead of the transmission image sensor unit described in U.S. Pat. No. 4,540,277, an image sensor unit operating in reflection may be alternatively used for aligning a mask mark with respect to a substrate table mark. Such a unit, which is described in U.S. Pat. No. 5,144,363, operates with a reflective mark on the table and comprises a relatively large number of detectors which observe the mark at different angles and which, together with the associated optical systems, are provided in a sensor plate which is arranged between the projection lens system and the substrate table. The off-axis alignment unit according to the invention must also be provided in this space. This unit must be arranged as close as possible to the center of the substrate table and requires a building space which is conical with an aperture of, for example 0.3. In practice, the length of the Y side of the substrate table approximately corresponds to the radius of the substrate for which the projection apparatus has been designed, for example 102 mm for an 8-inch substrate, so that there is little room for building in the alignment unit in this direction. The X side of the substrate table is, however, for example 25 mm longer than the Y side, so that an alignment unit which can handle an 8-inch substrate can be placed at a distance of 25 mm from the optical axis of the projection lens system. This is shown very diagrammatically in
As hitherto described the off-axis alignment unit is arranged in the projection column, comprising the mask holder, the projection system and the substrate holder, of the lithographic projection apparatus. With the increasing demand for larger IC's having smaller details, and thus comprising more electronic components, the alignment procedure becomes more and more time-consuming. The throughput of these apparatuses therefore tends to decrease without further measures taken. It has already been proposed to add to such an apparatus a separate measuring station. In this station the position in, for example the X-, Y- and Z-direction, of a substrate is measured before this wafer is brought in the projection column, or projection station. In the measuring station substrate marks can be aligned with respect to alignment marks on the substrate holder or table. After the substrate, together with the holder has been placed in the projection system only a mask alignment mark needs to be aligned with respect to the substrate holder mark, which takes only a short time. As in the apparatus, comprising a separate measuring station and projection station, during the illumination of a first substrate in the projection station a second substrate is being measured in the measurement station, the throughput of this apparatus is considerably larger than in an apparatus without a separate measurement station. The alignment unit used in the measuring station for aligning a substrate mark with respect to a substrate holder mark is preferably an off-axis alignment system as described herein.
The off axis alignment systems described above are examples of alignment systems that have a plurality of sensors which produce a plurality of signals that can be combined to determine the position of an alignment mark. In those examples, there are sensors that produce signals from the alignment mark for separate diffraction orders of light diffracted from the alignment mark. In the particular embodiments described, separate orders from the first order up to the seventh order can be detected. Furthermore, this is done for each of the X and Y directions. In addition, each of the seven orders in the X and Y directions can be detected at two different wavelengths of light illuminating the alignment mark. Consequently, this provides a total of 28 channels that provide signals substantially simultaneously during an alignment scan. One could also receive signals that are switched in time, rather than being simultaneous, without departing from the scope and spirit of this invention, especially if there is a rapid switching. According to the instant invention, information from a plurality of channels of such a multi-sensor alignment system are combined to obtain improved alignments. Improved alignment includes improved precision to align on a smaller scale and/or reduced errors due to systematic effects, such as processing effects, and/or improved reproducibility. It also includes “fallback procedures” in which alternative alignment steps or strategies are available to replace a failed strategy. In an embodiment of this invention, signals from a plurality of different diffraction orders are combined to determine a position of an alignment mark. In this embodiment, detected positions of an alignment mark from a plurality of diffractive-order channels are fit to a curve which can be expressed as a continuous function and then this continuous function is used to predict a position that would be obtained at substantially no systematic error.
An embodiment of such a predictive recipe has been found to be useful in the case where marks are deformed, for example when a semiconductor wafer undergoes the tungsten chemical mechanical polishing process (W-CMP). (The aluminum physical vapor deposition (Al-PVD) process is typically performed in conjunction with W-CMP. Therefore, when we refer to W-CMP one should understand that it can also include Al-PVD.)
The inventors have found good results with reducing errors due to the W-CMP processing by combining signals from the third diffractive order, the fifth diffractive order and the seventh diffractive order in the following linear combination
Xpred=−0.9399x3+0.6329x5+1.307x7, (11)
which is based on a three point least squares fit to the following general equation
xmeas(n)=C/n+xpred, (12)
where C is a constant and n is the order number.
Although the inventors have found good results using only three of the seven diffractive orders, the broader concepts of the invention are not limited to only the above-noted predictive recipe. Two or more diffractive-order channels may be used and measured values may be fit to other functions without departing from the general concept of this invention. Furthermore, this aspect of the invention is not limited to only providing a predictive recipe to correct for effects introduced in the W-CMP or similar processing.
In another example, the copper damascene process tends to introduce errors which appear to be substantially random as observed in the different wavelength and diffractive order channels. The inventors have found that a predictive recipe in which multiple available signal channels are averaged with equal weight tends to lead to good results for the Cu-damascene process. General concepts of this invention include combining information from a plurality of diffractive order channels to obtain an improved accuracy of detecting an alignment mark on a substrate compared to using a single channel alone. Such predictive recipes may take particular forms for particular processes performed on the substrate as described above for the case of the W-CMP process and for the Cu-damascene process. Furthermore, the concepts of this invention are not limited to only combining information from a plurality of wavelength channels and diffractive-order channels to provide predictive recipes to account for processing effects on a substrate. Information from a plurality of wavelength channels and diffractive order channels may be combined to provide predictive recipes to account for other changes to a substrate that may lead to errors in determining the position of an alignment mark.
The processing steps may introduce systematic effects that vary from substrate to substrate. For example, an apparent expansion due to a W-CMP and Al-PVD processes is plotted in
The above examples of predictive recipes may be viewed as static recipes in the sense that the information from the plurality of channels are combined with fixed coefficients. The term predictive recipe is intended to include the general concept of obtaining a mathematical representation of multichannel information and then using the mathematical representation to determine the position of the alignment mark. The concepts of this invention also include dynamic recipes, within the general concept of predictive recipes, in which information from the various channels are combined in a way that depends on measured quantities. For example, information from a plurality of diffraction order channels may be combined with coefficients which depend on the measured signal strength. Other measured quantities may also be used in dynamic recipes. For example, the output signal may be fit to an expected functional form, such as a sinusoid. The correlation coefficient in such a fit provides another measured quantity which can be used in a dynamic recipe for combining signals from the plurality of channels. The use of other input parameters, such as “mccr”, “minirepro”, “signal to noise ratio” “signal shape”, “signal envelope”, “focus” “tilt” “order channels position offset” “wavelength channels position offset”, “shift between segments”; and/or “coarse-fine position deviation”, possibly in combination with user input parameters, can enhance the performance.
Many of these parameters are related to the accuracy of the aligned position determination. The parameter “mcc” is the multiple correlation coefficient indicating how well the measured signal resembles the signal expected for a perfect alignment mark; the “minirepro” is the standard deviation of the aligned position of different sections or portions of an alignment measurement, indicating the accuracy of the aligned position; the “signal to noise ratio” is the fitted signal divided by the relative level of noise across the spectrum of the measured signal, while the “signal shape” is the relative level of a few discrete frequencies in this spectrum, generally at multiples of the base frequency; the “signal envelope” is variance of the signal strength during the measurement; the “focus” is the offset in wafer height with respect to the detector; the “tilt” is the angle between the wafer angle and the detector angle during the measurement; “order channels position offset” is the measured difference in aligned position of the various channels of one wavelength; the “wavelength channels position offset” is the measured difference in aligned position of the various wavelength channels; the “shift between segments” is the measured difference in aligned position of the various segments of a multi segmented alignment mark; and the “coarse-fine position deviation” is the difference between the position of the alignment marks in the fine phase with respect to their expected position based on alignment mark measurements in the coarse phase.
The coefficients may also be determined by including the historical data of the process. For example, one can compare the information obtained for the diffraction order channels with the information from previous wafers. If the information of a certain channel differs significantly from the information for that channel on previous wafers, one could give that channel a lower weighted coefficient than when the information from that channel closely resembles the information from previous wafers. Another way of dealing with the information from the plurality of diffraction order channels, is to model the individual channels in terms of wafer grid parameters (translation, rotation, wafer expansion, orthogonality, asymmetric scaling and higher order parameters) for each of the channels. The residuals of the wafer model fit to the individual signals—the so-called grid residuals—are qualifiers for the relative importance of a certain diffraction channel. For example, if a residual for a channel on a certain position on the wafer closely resembles the historical residual distribution at that position, a larger weight factor is assigned than when the residual is way off the average residual on previous wafers. (Also see
Based on experimental data it is possible to determine the correlation between individual detectors. If the correlation is introduced by the process it can be utilized to generate weighting coefficients that are capable of providing a more accurate measurement. One embodiment utilizes a set of data and determines a static predictive recipe based on this information. A second embodiment verifies and adjusts a predictive recipe on the fly. Small adjustments are made to the predictive recipe if the correlation between the detectors measured on the current wafer are not identical to the predictive recipe.
In the examples of static recipes so far, the same information from the plurality of channels is used for all marks. Another type of static recipe can be distinguished that defines which information from the plurality of channels should be used for each target on the wafer. Such a static recipe enables the alignment system to deal with process variations across the surface. Instead of immediately selecting or weighting part of the information from the plurality of diffraction order channels for each mark, it can also be useful to collect all information from all targets first.
Then the wafer grids can be determined for all of the individual channels. Such a scenario offers much more flexibility. For instance, it is now possible to determine the wafer expansion on channels and/or marks that are different from the channels/marks that are used to determine the rotation.
The off-axial alignment system OAS may be used to detect various types of alignment marks.
The capture process is a form of coarse alignment in which the position of the alignment mark is established within a desired range. A fine alignment is performed to determine a more precise position of the subject alignment mark. During the alignment process, the wafer 400 will be moved so that a desired alignment mark is scanned across the detection field of view of the off-axial alignment system OAS in a direction substantially orthogonal to the grooves in the target gratings of the alignment mark. Note that all signals for capture and fine alignment can be obtained substantially in parallel. When the wafer 400 is moved so that the target grating 416 of the alignment mark 410 moves across the field of view of the alignment system OAS along the direction of the scribe line 408, an alignment beam having two wavelength components is reflected and diffracted from the target grating 416 (see also
An order enhancing grating is defined as a grating with a signal strength, in a diffraction order detection channel of the alignment sensor, which is enhanced with respect to the signal strength detected for a base pitch grating.
When the reference mark 422 is scanned across the field of view of the off-axial alignment system OAS along its long dimension, the sensor produces signals in seven diffractive-order channels for each of the two illumination wavelengths. Consequently, a single scan across the alignment mark 422 can provide fourteen color/order signals for each of the targets 424, 426, 428 and 430. For an ordinary diffraction grating, the signal strength for the diffraction orders decreases with increasing diffraction order. The order-enhancing gratings enhance a particular diffraction order over what would normally be obtained with a diffraction grating with a purely constant pitch without any substructure. By selecting targets within the multi-target alignment mark 422 to be particular order enhancing gratings, one may then choose to utilize data of the enhanced order from the corresponding target. For example, when one selects target grating 424 to be a seventh order enhancing grating, target 426 to be a fifth order enhancing grating, and target 428 to be a third order enhancing grating, the inventors have obtained good results utilizing only the signals from the seventh order channels from the target 424, only the fifth order channels from the target 426, and only the third order channels from the target 428. The data illustrated in
The alignment marks 410 and 412 in
A goal of the sacrificial target(s) is to determine process dependencies to improve the determination of the alignment mark position using the non-sacrificial target(s). Therefore the sacrificial target(s) are sensitive to process influences. For example, a first target of a multi-target alignment mark is optimized for position stability, while a second target of this multi target alignment mark is very sensitive to, for example, mark depth, line width or line angle. The sensitivity can be used to correct the process influence on the first target, resulting in a more stable position. While the position of the first target is relatively stable regardless of the process, the second target is optimized to detect with a proper detector the process effects and with this knowledge a better position of the alignment mark is calculated based on the combined information of both targets. An example of a target useful for position stability is a target with a section immediately above another section of the target. For example, the sections may be in different layers of a substrate. The different sections of the target may be gratings with different pitches in one exemplary embodiment. Since the target sections are at a nearly identical position relative to the plane of the substrate, signals from both target sections can be measured simultaneously. The signals from the different sections of the targets are thus not sensitive to errors in the motion of the substrate or the substrate holder.
The goal of a sacrificial target in another embodiment is to prepare or finish an alignment target positioned either above or below the sacrificial target. An example of a sacrificial target placed on top of another target is a structure which removes opaque material from the wafer surface. This enables light to penetrate the substrate and reach an underlying target, enabling alignment on a target which would otherwise not have been possible. An example of a sacrificial target placed beneath another target is an opaque structure. The effective depth of an alignment grating located in a transparent material can be tuned by placing an opaque structure underneath the grating. The scope of this embodiment is not limited to the examples provided. Other structures placed above or below an alignment target to improve the alignment performance are within the scope of this embodiment.
In another embodiment a first target closely resembles the product structure and thus suffers substantially the same distortion and process effects as the actual product. The target can be built up out of substructures comprising features like contact holes or lines with a resolution more closely resembling the features in the product. Such a target represents the position of the product better, but can also be affected more by the process, making the target hard to detect. For example the CMP process is known to completely remove surface structures at product resolution. A benefit of the multi target alignment mark is that a robust second target is essentially always available for backup purposes.
In another embodiment the overlay is determined using a multi target alignment mark by printing target(s) or part of target(s) in different layers. An example is given using the Moiré technique where in layer 1 pattern a is printed and in layer 2 pattern b is printed overlapping pattern a, both forming a first target of a multi target alignment mark. Using a specific pattern a and pattern b, this generates a beat pattern that can be detected by the alignment sensor and the position shift of the first target with respect to a second target indicates the overlay. Benefits of the multi target mark in this matter are that the alignment targets are placed at substantially the same position on the wafer. Possible measurement errors due to process variations within a die will not occur.
The sacrificial target can also be used for other purposes than position determination, for example focus, energy, dose, line width, contact hole width or critical dimension measurements and can be used in single-target marks or multi-target marks. In this embodiment the sacrificial target is especially sensitive to these effects making them detectable with an alignment or overlay sensor. Product line width, contact hole width or critical dimension generally can be measured using the relative signal strength of the alignment target when this target consists of groups of contact holes or lines with resolution and density similar to the product. An example of an alignment target that is sensitive to focus and dose is shown in WO 02/052350 A1, which is hereby incorporated by reference in its entirety. See also, EP-02253 1766, the entire content of which is hereby incorporated by reference. For these measurements, some edge dies with no yield prospect are printed with a focus offset or dose offset and the focus or dose is detected with an alignment or overlay sensor and adjusted in a feedback loop.
Using three detector channels, it becomes possible to do an online qualification and calibration of the measurement systems. For example, three detector channels measure the position of a multi target alignment mark. With two detector channels agreeing on the measured position and the third sensor measuring a different position this third sensor can be qualified as unreliable when measuring this alignment mark. In another example two detector channels always measure a constant position difference in the aligned position, regardless of the process or alignment mark. This offset can be corrected by automatic calibration or matching. In another example two detector channels have a specific offset in position measurement for alignment marks on a particular process, this offset can be calibrated for said particular process. The calibrated offset can be used to correct the measured value when switching from one sensor to another or when both detector channels are used at the same time.
As noted above, the concept of multi-target alignment marks is not limited to the use of only multi-grating alignment marks. The targets within the multi-target alignment mark may be targets of various types. It may include gratings as targets as well as other targets which are detected by other processes, such as edge detection and/or image recognition techniques. In addition, some targets within a multi-target alignment mark may be optimized for one particular type of sensor, such as gratings are optimized for the off axial alignment system OAS described above, but other targets within the same mark may be suitable for other alignment systems and/or other measurement systems. For example, a target within a multi-target alignment mark may be suitable for detection by an online metrology tool that is used to assess the accuracy of exposure of a photo-resist before further processing.
The capture process may also be improved by the use of new multi-target alignment marks. Under the current capture process, the alignment sensor is scanned across a diffraction grating that has a first pitch. As noted above, the scanning is typically accomplished by moving the substrate through the field of view of the alignment sensor. This results in a sinusoidal signal 500 of a first period, as is illustrated schematically in
The capture process can also be performed using diffraction order sub-beams detected with either of the axial or off-axial alignment systems described above. Furthermore, higher than first order sub-beams may be used for the capture process and/or diffraction order enhancing gratings may be used. In addition, a variety of types of targets may be used.
Other target types may also be included in a multi-target mark for performing capture of the alignment mark. For example, one may include multiple grating pairs in which each pair has a different sub-segmentation, e.g., diffraction-order-enhanced gratings, or different process segmentations, e.g., as discussed previously with reference to
One of the multigrating embodiments is a capture multigrating mark. This mark has two gratings optimized for capturing and two grating optimized for fine alignment. A major benefit of this embodiment is that a compromise is not required to enable both capturing and fine alignment with a single grating. Another embodiment is the process multigrating. (See,
Another embodiment of this invention is a system for an improved overlay strategy. The parameters for wafer alignment are important for an overlay strategy in building up a micro-device. Some of the important parameters are the number of alignment marks used, the alignment recipe, residual thresholds, and location of the marks on the wafer.
One or more targets within a multi-target alignment mark are used for active alignment while other targets are measured in parallel to provide additional information such as the Signal Quality (MCC) which is a correlation coefficient for the fit of the signal to an expected functional form, signal strength (SS) and grid modeling parameters (Translation, Rotation, Magnification, etc).
The embodiment described immediately above is for a feedback loop in an external control system. This is useful for slowly varying parameters such as on a batch-to-batch basis. Another embodiment of this invention provides a “fallback” system and method, and dynamic global alignment strategies on wafer-to-wafer basis. This lithography-tool-based control loop is an embodiment of Automated Equipment Control (AEC). In this invention, an automatic fallback is to one of the alternative strategies in case the active alignment strategy fails. For instance, a predefined fallback strategy or a strategy that performs second best can be used for fallback. In that way, the number of wafers that is rejected during the manufacturing process due to alignment errors can be minimized.
In case of alignment failure, it is clear that a switch to another strategy should take place. However, in cases of dynamic global alignment strategies within a batch, switching to an alternative strategy is based on indirect indicators of overlay performance, since no external overlay data is available. Such indicators are for instance the order-to-order stability, residual analysis or signal quality analysis (SS, MCC). See, “Extended Athena alignment performance and application for the 100 nm technology node” by Ramon Navarro, Stefan Keij, Arie den Boef, Sicco Schets, Frank van Bilsen, Geert Simons, Ron Schuurhuis, Jaap Burghoorn. 26th annual international symposium on microlithography, Feb. 25-Mar. 2, 2001 in Santa Clara, Calif., the entire content of which is hereby incorporated by reference. Order-to-order stability is a measure for the variation on the aligned position as induced by processing only. Residual analysis characterizes how well the modeled wafer grid fits into the measured positions. The decision to switch to an alternative strategy can be implemented within a batch. However, when switching to an alternative strategy, usually new process corrections are required. In both cases (fallback and dynamic global alignment strategies), the determination of the correct process corrections is a problem. Since the process corrections are assumed to be stable over batches, those can be derived in the slow feedback APC system. Thus, the APC system does not determine the parameters of the new alignment strategy here, but determines the process corrections for all segments of the multi-target mark and sends those corrections to the lithography tool. With these data on the lithography tool, a feedback loop can be implemented with switching of alignment strategy on a batch-to-batch basis.
In another embodiment of this invention, data collected from the alignment systems can be used to improve quality control during manufacture. Quality control is performed on an overlay metrology tool on a few wafers that are usually randomly picked from each batch. Therefore, it is very well possible that non-representative wafers are chosen from a batch. Since the process corrections for the next batch are based on these overlay metrology measurements, this may result in overlay variations from batch-to-batch. In this embodiment, alignment data—that is available for every wafer—is used to identify the wafers to be measured on the overlay metrology tool. To determine which wafers are representative for a batch, one can for instance determine the wafer-to-wafer distribution of the wafer model parameters (translation, expansion, rotation) and identify the wafers that are closest to the batch averaged wafer parameters. In particular, the wafer expansion and non-orthogonality are useful for this purpose. Alternatively, one can look at the grid residuals, i.e., the deviation of each measured position from a best grid fit to the measured positions. If the alignment marks have been exposed on a different machine, a systematic error can occur with the grid residuals. The alignment system will then measure each mark with a different offset. This offset gives a large contribution to the residuals, thereby obscuring true processing effects. By determining the residual distribution per mark location as shown in
In another embodiment of the invention, the alignment data is used to determine the process corrections for the alternative strategies. The process corrections are determined on the overlay metrology tool for the active strategy only. The difference between the active and alternative strategy is measured by the alignment system. The formula to calculate process corrections for a new strategy from the process corrections of an old strategy and alignment data is as follows:
PCnew=<Wnew−Wold>+PCold. (13)
This applies for each wafer model parameter, where PC is offline modeled wafer model parameters and W is the exposed wafer grid parameters. The applicable wafer parameters are: translation X, Y; wafer expansion X, Y; wafer rotation, non-orthogonality. Different scenario's are available for the averaging of alignment data, for example:
-
- 1. all wafers from the batch;
- 2. only the wafers used for offline overlay measurement;
- 3. previous batches (in case of rejects).
In another embodiment of this invention, instead of correcting the modeled overlay metrology data (PC), the raw overlay metrology data are given an offset according to the grid difference Waltenative−Wactive for each overlay target and each wafer measured offline. Now the overlay performance and process corrections for that alternative strategy are calculated for the batch. The operator or APC system can now monitor the trends in overlay (i.e., as a function of time for several batches) so that the operator or APC system determines whether to switch to one of the alternative strategies.
Since the numerous systems described above use coherent alignment radiation sources, the phase modulation techniques described in U.S. Pat. No. 6,384,899 may also be used in combination with the systems described herein. The entire content of the U.S. Pat. No. 6,384,899 is hereby incorporated herein by reference in its entirety. The alignment system according to the current invention may be implemented in a variety of alignment apparatuses. In a specific example, it may be implemented in the alignment system illustrated in
The invention is described with reference to exemplary embodiments. It is not limited to only those embodiments and includes combinations and variations of those embodiments within the scope of the invention as defined by the appended claims. The invention is also described with reference to its use in apparatuses for step-and-scan imaging of a mask pattern on a substrate for manufacturing ICs, but this does not mean that it is limited thereto. The invention may be alternatively used in such an apparatus for manufacturing integrated, or plenary, optical systems, guidance and detection patterns for magnetic domain memories, or liquid crystalline display panels. The projection apparatus may not only be an optical apparatus, in which the projection beam is a beam of electromagnetic radiation and the projection system is an optical projection lens system, but also an apparatus in which the projection beam is a charged-particle beam such as an electron beam, an ion beam or an X-ray beam, in which an associated projection system, for example an electron lens system is used. Generally, the invention may be used in imaging systems with which images having very small details must be formed. Of course, various combinations of the above are also within the scope of this invention.
Claims
1. A metrology system for a lithographic apparatus, comprising: a source of alignment radiation; a detection system comprising a first detector channel and a second detector channel; and a processing unit in communication with said detection system, wherein said processing unit is constructed to process information from said first and second detector channels in a combination to determine at least one of a focus, energy, dose, line width, contact hole width or critical dimension based on detecting an alignment mark on a work piece.
2. A metrology system according to claim 1, wherein said first detector channel is a first non-zero-diffraction-order channel, and said second detector channel is a second non-zero-diffraction-order channel.
3. A metrology system according to claim 2, wherein said processing unit is constructed to process said information from said first and second non-zero-diffraction-order channels by weighting first and second signals from said first and second non-zero-diffraction-order channels with factors that depend on a relative strength of said first signal to said second signal.
4. A metrology system according to claim 3, wherein the weight factor for said second non-zero-diffraction-order channel is set to zero when a strength of said first signal relative to said second signal exceeds a predetermined threshold.
5. A metrology system according to claim 1, wherein said first detector channel is a first wavelength channel, and said second detector channel is a second wavelength channel.
6. A metrology system according to claim 1, wherein said source of alignment radiation comprises a laser providing illumination radiation at a first wavelength.
7. A metrology system according to claim 6, wherein said source of alignment radiation further comprises a second laser providing illumination radiation at a second wavelength.
8. A metrology system according to claim 1, wherein said source of alignment radiation comprises a broadband source adapted to provide illumination radiation at a plurality of wavelengths.
9. A metrology system according to claim 1, wherein said detection system comprises a first detector constructed to provide a first detection signal to said first detector channel, and a second detector constructed to provide a second detection signal to said second detector channel.
10. A metrology system according to claim 1, wherein said detection system is constructed to detect a first target of a multitarget alignment mark and a second target of said multitarget alignment mark.
11. A metrology system according to claim 10, wherein said first target of said multitarget alignment mark includes at least one of a structure above or below a portion of said first target.
12. A metrology system according to claim 10, wherein said detection system is constructed to detect a first target of a multitarget alignment mark to provide said first detection signal, and said detection system is constructed to detect a second target of said multitarget alignment mark to provide said second detection signal.
13. A metrology system according to claim 10, wherein said first target of said multitarget alignment mark is a process target that is constructed to undergo predetermined changes during a manufacturing process of said work piece.
14. A metrology system according to claim 13, wherein said second target of said multitarget alignment mark is constructed to undergo less changes during said manufacturing process of said work piece than said first target to provide a robust backup target to said first target.
15. A metrology system according to claim 10, wherein said first target of said multitarget alignment mark is a sacrificial target that is constructed to be destroyed during a manufacturing process of said work piece.
16. A metrology system according to claim 10, wherein said first target of said multitarget alignment mark is constructed in a first layer of material of said work piece, and said second target of said multitarget alignment mark is constructed in a second layer of material of said work piece.
17. A metrology system according to claim 10, wherein said first target of said multitarget alignment mark is a diffraction order enhancing grating that enhances a first diffraction order of radiation diffracted therefrom compared to a purely periodic grating, and said second target of said multitarget alignment mark is a diffraction order enhancing grating that enhances a second diffraction order of radiation diffracted therefrom compared to a purely periodic grating, said first diffraction order being an integer value different from an integer value of said second diffraction order.
18. A metrology system according to claim 1, wherein said detection system further comprises a third detector channel, and said processing unit is constructed to process information from said first, second and third detector channels in a combination to determine at least one of said focus, energy, dose, line width, contact hole width or critical dimension while performing at least one of a calibration and a qualification.
Type: Application
Filed: Dec 6, 2005
Publication Date: Apr 20, 2006
Applicant: ASML NETHERLANDS B.V. (Veldhoven)
Inventors: Franciscus Van Bilsen (Eindhoven), Jacobus Burghoorn (Haelen), Richard Franciscus Van Haren (Veldhoven), Paul Hinnen (Veldhoven), Hermanus Van Horssen (Maasbracht), Jeroen Huijbregtse (Breda), Andre Jeunink (Bergeyk), Henry Megens (Eindhoven), Ramon Navarro Y Koren (Veldhoven), Hoite Tolsma (Eindhoven), Hubertus Gertrudus Simons (Venlo), Johny Schuurhuis (Eindhoven), Sicco Schets (Eindhoven), Brian Lee (Seoul), Allan Dunbar (London)
Application Number: 11/294,473
International Classification: G01N 21/86 (20060101);