JP2007027263A - Exposure apparatus and method, position detecting apparatus and method, and device manufacturing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a signal processing technique having high accuracy of position detection. <P>SOLUTION: An exposure apparatus for exposing a substrate via an original plate has a detecting means for detecting a reflection light from an object, and a signal processing means for obtaining the position of the object on the basis of the detection signal detected by the detecting means. The detection signal comprises N pieces of constituents. The signal processing means uses M pieces of orthogonal base vector previously obtained, the M being smaller than the number N, to obtain similarities with respect to a plurality of different relative positions between the detection signal and the orthogonal base vector, and signal processing means obtains the position of the object on the basis of the plurality of obtained similarities. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an exposure apparatus and method, a position detection apparatus and method, and a device manufacturing method for obtaining the position of an object such as an original or a substrate in an exposure apparatus that exposes a substrate through the original.

  A fine semiconductor element such as a semiconductor memory or a logic circuit or a liquid crystal display element is manufactured by using a photolithography technique. At this time, a projection exposure apparatus is used in which a circuit pattern drawn on a reticle (mask) is projected onto a wafer or the like by a projection optical system to transfer the circuit pattern.

In a projection exposure apparatus, as the integration of semiconductor elements increases, it is required to project and expose a reticle circuit pattern onto a wafer with higher resolution. The minimum dimension (resolution) that can be transferred by the projection exposure apparatus is proportional to the wavelength of light used for exposure and inversely proportional to the numerical aperture (NA) of the projection optical system. Therefore, the shorter the wavelength, the better the resolution. For this reason, recent light sources emit light having a shorter wavelength than ultra high pressure mercury lamps (g-line (wavelength: about 436 nm), i-line (wavelength: about 365 nm)). That is, a KrF excimer laser (wavelength of about 248 nm) or an ArF excimer laser (wavelength of about 193 nm) is used, and the practical use of F ₂ laser (wavelength of about 157 nm) is also progressing. Furthermore, further expansion of the exposure area is also required.

  Conventionally, a step-and-repeat type exposure apparatus (also referred to as a “stepper”) that performs batch exposure by reducing a substantially square exposure area to a wafer has been the mainstream. However, step-and-scan type exposure apparatuses (also referred to as “scanners”) are becoming mainstream in order to achieve the above requirements. The scanner exposes a large screen with high accuracy by relatively scanning the reticle and wafer at a high speed with the exposure area as a rectangular slit.

  In the scanner, the surface position at a predetermined position of the wafer is measured by the surface position detecting means of the oblique incidence system before the predetermined position of the wafer reaches the exposure slit region during the exposure. When the predetermined position is exposed, correction is performed so that the wafer surface is aligned with the image forming position of the pattern on the reticle.

In particular, the longitudinal direction of the exposure slit (that is, the direction perpendicular to the scanning direction) has a plurality of measurement points for measuring not only the height of the surface position of the wafer but also the tilt of the surface. Yes. Many methods for measuring the height and tilt have been proposed (see Patent Document 1).
JP-A-6-260391

  However, in recent years, the wavelength of exposure light has been shortened and the NA of the projection optical system has been increased, and the depth of focus has become extremely small. For this reason, the accuracy of aligning (focusing) the wafer surface to be exposed with the best imaging plane, so-called focus accuracy, is becoming increasingly severe. In particular, measurement errors of the surface position detecting means due to the influence of the pattern on the wafer and the unevenness of the thickness of the resist applied to the wafer cannot be ignored.

  For example, due to uneven thickness of the resist, a large step is generated in the vicinity of the peripheral circuit pattern and the scribe line, although it is smaller than the focal depth, for focus measurement. For this reason, the inclination angle of the resist surface increases, and the reflected light detected by the surface position detecting means deviates from the regular reflection angle. Further, due to the difference in density of the pattern on the wafer, a difference occurs in the reflectance of the wafer between the dense pattern area and the rough area. In this way, the reflection angle and reflection intensity of the reflected light detected by the surface position detecting means change. For this reason, asymmetry occurs in the signal waveform in which reflected light is detected as shown in FIG. 11, and an error occurs in the position detection of the wafer in the Z direction in signal processing such as center of gravity processing.

  In general, the wafer process causes non-uniformity in the pattern step in the wafer surface and uneven thickness of the resist, and the reproducibility within the wafer or between the wafers is poor, so that the offset process is difficult. Therefore, the wafer surface position cannot be measured during exposure, and the exposure operation is stopped, or as a result of large defocusing, chip failure occurs and yield decreases.

  Accordingly, an object of the present invention is to provide a signal processing technique with high position detection accuracy.

A first invention is an exposure apparatus that exposes a substrate through an original,
Detection means for detecting reflected light from the object;
Signal processing means for determining the position of the object based on the detection signal detected by the detection means;
The detection signal is composed of N elements,
The signal processing means calculates a similarity when approximating the detection signal using M pre-determined orthogonal basis vectors smaller than the N, a plurality of different degrees between the detection signal and the orthogonal basis vectors. An exposure apparatus characterized in that a position of the object is obtained based on a plurality of similarities obtained with respect to a relative position.

A second invention is a position detection device for determining the position of an object,
Detection means for detecting reflected light from the object;
Signal processing means for determining the position of the object based on the detection signal detected by the detection means;
The detection signal is composed of N elements,
The signal processing means calculates a similarity when approximating the detection signal using M pre-determined orthogonal basis vectors smaller than the N, a plurality of different degrees between the detection signal and the orthogonal basis vectors. The position detection device is characterized in that the position of the object is obtained based on a plurality of similarities obtained with respect to a relative position.

The third invention is an exposure method for exposing a substrate through an original,
A detection step for detecting reflected light from the object;
A signal processing step for determining the position of the object based on the detection signal detected in the detection step;
The detection signal is composed of N elements,
In the signal processing step, the similarity when the detection signal is approximated using M pre-determined orthogonal basis vectors less than N is calculated as a plurality of different degrees of difference between the detection signal and the orthogonal basis vectors. The exposure method is characterized in that a position of the object is obtained based on a plurality of similarities obtained with respect to a relative position.

Moreover, 4th invention is a position detection method which calculates | requires the position of a target object,
A detection step for detecting reflected light from the object;
A signal processing step for determining the position of the object based on the detection signal detected in the detection step;
The detection signal is composed of N elements,
In the signal processing step, the similarity when the detection signal is approximated using M pre-determined orthogonal basis vectors less than N is calculated as a plurality of different degrees of difference between the detection signal and the orthogonal basis vectors. In this position detection method, the relative position is obtained, and the position of the object is obtained based on the obtained plurality of similarities.

  Further, a fifth invention is a device manufacturing method characterized by having a step of exposing a substrate through an original plate using the exposure apparatus of the first invention.

  Other objects, features, and effects of the present invention will become apparent from the following description made with reference to the accompanying drawings. In the drawings, the same or similar reference numerals represent the same or similar components throughout the drawings.

  According to the present invention, it is possible to provide a signal processing technique with high position detection accuracy.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

  FIG. 10 is a schematic block diagram showing the configuration of the exposure apparatus 1 as one aspect of the present invention.

  The exposure apparatus 1 is a projection exposure apparatus that exposes a wafer 40 with a circuit pattern formed on a reticle 20 by a step-and-scan method. Such an exposure apparatus is suitable for a lithography process of submicron or quarter micron or less. As shown in FIG. 10, the exposure apparatus 1 includes an illumination apparatus 10, a reticle stage 25 on which a reticle 20 is placed, a projection optical system 30, a wafer stage 45 on which a wafer 40 is placed, and a focus tilt detection system 50. And a control unit 60. The control unit 60 includes a CPU and a memory, and is electrically connected to the illumination device 10, the reticle stage 25, the wafer stage 45, and the focus tilt detection system 50, and controls the operation of the exposure apparatus 1. The control unit 60 also functions as a processing unit that executes signal processing described later in the present embodiment. Note that at least a part of the signal processing may be executed by an apparatus outside the exposure apparatus (for example, a dedicated signal processing apparatus, a suitable computer, a pattern inspection apparatus, etc.). In particular, the learning process described later may be executed by a computer outside the exposure apparatus. The external apparatus and the exposure apparatus are preferably connected by communication means.

  The illumination device 10 illuminates a reticle 20 on which a transfer circuit pattern is formed, and includes a light source unit 12 and an illumination optical system 14.

  The light source unit 12 uses laser light, for example, ArF excimer laser light having a wavelength of about 193 nm, KrF excimer laser light having a wavelength of about 248 nm, or the like. The type of light source is not limited to an excimer laser light source, and an F2 laser light source having a wavelength of about 157 nm or an EUV (Extreme Ultraviolet) light source having a wavelength of 20 nm or less may be used.

  The illumination optical system 14 is an optical system that illuminates a surface to be illuminated using a light beam emitted from the light source unit 12, and in this embodiment, the light beam is formed into an exposure slit having a predetermined shape optimum for exposure, and the reticle 20 is formed. Illuminate. The illumination optical system 14 includes a lens, a mirror, an optical integrator, a diaphragm, and the like. For example, a condenser lens, a fly-eye lens, an aperture diaphragm, a condenser lens, a slit, and an imaging optical system are arranged in this order. The illumination optical system 14 can be used regardless of on-axis light or off-axis light. The optical integrator may include an integrator configured by stacking a fly-eye lens or two sets of cylindrical lens array (or lenticular lens) plates. The optical integrator may be replaced with an optical rod or a diffractive element.

  The reticle 20 is made of, for example, quartz, on which a circuit pattern to be transferred is formed, and is supported and driven by the reticle stage 25. Diffracted light emitted from the reticle 20 passes through the projection optical system 30 and is projected onto the wafer 40. The reticle 20 and the wafer 40 are arranged in an optically conjugate relationship. The pattern of the reticle 20 is transferred onto the wafer 40 by scanning the reticle 20 and the wafer 40 at a speed ratio corresponding to the reduction magnification of the projection optical system. The exposure apparatus 1 is provided with an oblique incidence type reticle detection means 70. The position of the reticle 20 is detected by a reticle detection means (not shown), and is arranged at a predetermined position.

  The reticle stage 25 supports the reticle 20 via a reticle chuck (not shown) and is connected to a moving mechanism (not shown). A moving mechanism (not shown) is configured by a linear motor or the like, and can move the reticle 20 by driving the reticle stage 25 in the X-axis direction, the Y-axis direction, the Z-axis direction, and the rotation direction of each axis.

  The projection optical system 30 has a function of forming an object surface (pattern surface on the reticle) on the image plane, and forms an image on the reticle on the wafer 40 by diffracted light that has passed through the pattern formed on the reticle 20. To do.

  The wafer 40 is an object to be processed, and a photoresist is applied on the substrate. In the present embodiment, the wafer 40 is also an object (detected body) whose position is detected by the focus tilt detection system 50.

  The wafer stage 45 supports the wafer 40 by a wafer chuck (not shown). The wafer stage 45 is moved in the X-axis direction, the Y-axis direction, the Z-axis direction, and the rotation direction of each axis by a linear motor or the like, similarly to the reticle stage 25. The position of the reticle stage 25 and the position of the wafer stage 45 are monitored using a laser interferometer, for example, and both are driven at a constant speed ratio. The wafer stage 45 is provided on a stage surface plate supported on a floor or the like via a damper, for example. The reticle stage 25 and the projection optical system 30 are provided on a lens barrel surface plate (not shown) supported via a damper on a base frame placed on a floor or the like, for example.

  The focus tilt detection system 50 detects the surface position (position in the Z-axis direction) of the wafer 40 during exposure using an optical measurement system. Specifically, a method in which a light beam is incident on the surface of the wafer 40 at a high incident angle, and an image shift of reflected light (a positional deviation of reflected light from a reference position) is detected by a photoelectric conversion element such as a CCD, so-called An oblique incidence + image shift detection method is used. In particular, a light beam is incident on a plurality of measurement points to be measured on the wafer 40, each reflected light beam is guided to an individual sensor, and an amount by which the wafer 40 should be tilted from information (measurement result) on the position of each reflected light beam. Calculated.

  FIG. 1 is a block diagram for explaining the outline of this embodiment.

  In order to explain the signal processing method of the present embodiment, first consider the case where a one-dimensional signal as shown in FIG. 2 is input. Even if this signal is distorted by a semiconductor process or the like, it is an object to provide a signal processing method capable of accurately obtaining the center position of this signal. This one-dimensional signal is received by, for example, a one-dimensional CCD sensor or the like, and is given as a sample value sequence sampled at a sampling interval determined by optical and electrical conditions of the measurement system. In view of the current level of technology, it is essential to obtain the center position of this signal from the sample sequence with an accuracy equal to or less than the sampling interval of the sample sequence.

  The signal processing method in the present embodiment is a method for obtaining a measurement position by approximating an input signal using various types of learning signals whose correct positions are known, and learning a signal for performing this approximation in advance. This is the basic idea.

  A certain existing method, for example, the slice method, cuts an input signal to be processed by a certain slice value and uses the center position as a measurement position. In the existing method, as shown in FIG. 3, the input signal is cut by a slice value (for example, 40% of the maximum value of the signal), and the center (average) position of two signal positions taking the slice value is measured. It is what.

  On the other hand, in the signal processing method of this embodiment, the input signal is approximated by a signal for approximating the input signal, and the position of the approximate signal having the highest similarity (also referred to as approximation) is used as the measurement position. is there.

  This approximate signal is expressed as a linear sum of several principal component signals obtained from various types of learning signal groups. At the time of this approximation, the position of the approximate signal (= linear sum of principal component signals) with respect to the input signal is not known. Therefore, the degree of overlap (relative position) between the input signal and the approximate signal is changed. Specifically, for example, one of the signals is shifted toward the pixel on the sensor. In that way, the similarity is evaluated. The position of the approximate signal with the highest similarity (= the position of the linear sum of the principal component signals that can be approximated most accurately) is taken as the measurement position.

  The actual measurement position is the correct known position of the approximate signal at the position with the highest similarity. This is the basic idea of this embodiment.

  Next, the derivation of the principal component signal from the learning signal group in the signal processing method of the present embodiment will be described. In the present embodiment, as described below, Karhunen-Loeve expansion (hereinafter referred to as KL expansion) is used to derive the principal component signal.

  First, KL expansion will be described.

  This KL expansion is a method for obtaining a subspace that best approximates the distribution of feature vectors in a linear space, and is a technique for principal component analysis in multivariate analysis.

  Assume that there is a signal FL1 given by N pixels. The number of dimensions of the feature vector is said to be N-dimensional from N of the number of pixels. This number of pixels is also called a sampling number. To completely represent the signal FL1, N dimensions, that is, N pixels are required. A total of m signals FL1, 2,. . . . Suppose there is m.

  If each signal is represented by N pixels, mN pieces of information are required. KL expansion is a technique for reducing the amount of information and expressing the original signal. For this purpose, m signals FL1,2. . . . Each vector indicating a feature of m (a space formed by this is called a subspace) is obtained by multivariate analysis.

  If the KL expansion is used, the original signal can be expressed by the sum of products of each base vector as an axis of the subspace and a value (coefficient) obtained by projecting the signal to the base vector, that is, a linear sum of each base vector. it can.

  Furthermore, the basis vectors obtained by KL expansion are orthogonal. To be orthogonal means that the inner product of different basis vectors becomes zero. Using this, taking the inner product of each basis vector and the signal to be expressed, the coefficient of the basis vector is obtained.

  The basis vectors can be calculated up to N if the learning signal is N-dimensional (number of pixels N), and the order is given in the order of the ratio that can represent the original signal, and is called a primary basis vector and a secondary basis vector.

Detailed information on further KL expansion can be obtained by referring to the following documents, for example.
Easy-to-understand pattern recognition, Kenichiro Ishii, Ohm, ISBN 4274131491, or Basic theory of image processing algorithm, page 98, Kohei Arai, Academic Publisher In addition, in this specification, principal component signals are orthogonal basis vectors or simply Also called a basis vector.

  Next, an example of expressing the one-dimensional signal f (x) using KL expansion will be shown.

  First, let the one-dimensional signal be f (x). It is assumed that the number of samples sampled at a certain sampling interval is N. Then, f (x) is expressed as an N-dimensional vector f = [f (1), f (2), f (3),..., F (N)]. In general, an N-dimensional vector f is obtained by using M N-dimensional basis vectors pk = [pk (1), pk (2), pk (3),..., Pk (N)] orthogonal to each other. It can be approximated as follows:

  At this time, when N = M, there exists a coefficient sequence ak and a base vector pk that completely match both sides of the equation (1) for any vector f. In the case of N> M, the above two sides do not completely coincide. However, a plurality of N-dimensional vectors f are KL-expanded, and the base vectors are arranged in descending order in which the original signal can be expressed as described above, and are used in order as the pk. By doing so, it is known that the error of both sides of the equation (1) can be minimized with the same M.

  Next, a method for deriving basis vectors from a plurality of learning signal groups using KL expansion will be described.

  That is, M pieces of a plurality of learning signal groups each represented as an N-dimensional vector f = [f (1), f (2), f (3),..., F (N)] are expanded by KL expansion. A method for obtaining the basis vector pk of will be described.

  First, for a group of learning signal vectors f [f (1), f (2), f (3),..., F (N)], a specific sample number corresponding to the position to be obtained (for example, c, this The sample value at that time makes f (c)) coincide with the entire learning signal group. That is, the position of the learning signal is shifted in advance so as to be so. Then, the L given learning signal vectors are arranged in the column direction to form the following matrix A (which has L rows and N columns).

Furthermore, the covariance matrix given by the product of the transposed matrix A ^T and A of the A and B (the N rows and N columns).

  The matrix B is subjected to eigenvalue decomposition, and eigenvector numbers k are assigned in descending order of eigenvalues. This is the basis vector pk.

The covariance matrix, eigenvalues, and basis vectors are described in the following documents.
・ Junkichi Satsuma, Shoji Yotsuya, Key Point Linear Algebra (Key Point 2 of Science and Engineering Mathematics), Iwanami Shoten, ISBN4-00-007862-3.
Masayoshi Yoshida et al., Matrix Eigenvalue Problem for Modern Engineering (Applied Mathematics Guide Series 4), Hyundai Engineering Co., Ltd., ISBN 4-87472-161-3.

  Next, a method for approximating an input signal using a basis vector will be described.

  The input signal f can be approximated by using the signals represented by the basis vectors pk obtained as described above and using the equations (1) and (2). This approximation uses M basis vectors of k = 1,. The larger the value of M, the better the accuracy of approximation. However, even if the value of M is increased to a certain extent, it will only take calculation time, and no improvement in accuracy can be expected. In this embodiment, M = 3 is empirically set. However, this method can be used even when M = 3.

  In the approximation method, first, ak (k = 1,..., M) is obtained for the input signal vector f using equation (2). Then, an approximate signal vector fappropimate is obtained using the following equation.

  In this approximation, since the base vector is obtained by aligning the center position of the learning signal vector with a specific sample position (for example, c), the center position of the base vector is also c. However, the position (center position) where the input signal is to be measured is unknown. Therefore, when the center position is different between the input signal and the basis vector, the similarity of the approximate signal to the input signal is low. However, when the center position of the input signal completely matches the center position of the basis vector, the approximate signal vector fapproximate is considered to be closest to the input signal vector f. Therefore, if the center position of the base vector that gives the approximate signal vector closest to the input signal is estimated, it can be considered as the center position of the input signal.

  Therefore, the search for the approximate signal vector fappropimate having the highest similarity will be described.

If the basis vectors obtained by shifting the center position by b and pk ^b, thereby approximating signal vector fapproximateb approximating the input signal can be determined as follows.

The similarity S between the approximate signal vector fappropimate ^b and the input signal vector f can be calculated using the inner product of both as follows.

Therefore, as shown in FIG. 4, it is considered that the center position of the input signal can be estimated if bmax that gives a fappropimate ^b that maximizes S (b) can be estimated.

  The estimation of bmax that maximizes S can be performed by a method by curve fitting.

  The center position is obtained for the input signal by using the above-described slice processing technique. Then, the value of S (b) given by the equation (6) is obtained at several sample positions in the vicinity (for example, about 10 points). As shown in FIG. 5, a continuous function that best matches this value is obtained by model curve fitting. An even function such as a Gaussian function or a quadratic function can be used as the model curve function at this time.

Thus, bmax is determined as a position that gives the maximum value of the continuous function obtained by the model curve fitting.
Next, the “supervised signal” necessary in the signal processing using the KL expansion described above will be described.

  A learning signal (learning signal whose correct answer is known) including information on the correct answer (position to be obtained (center position)) is referred to as a supervised signal in the field of signal processing and image recognition.

  In the present embodiment as well, a supervised signal is necessary to obtain a basis vector by performing principal component analysis and KL expansion on a large number of signals.

  In this embodiment, an actual signal and inspection data are used to obtain a supervised signal.

  It is easy to get real signals.

  For example, the measurement principle of an alignment detection system that measures the position in the XY direction of a wafer in a semiconductor exposure apparatus includes an image processing method that acquires image data of alignment marks illuminated in bright field and processes the image data. When this method is used, an image signal (video signal) from a photoelectric conversion element such as a CCD is an actual signal.

  For such alignment measurement, an inspection method using an overlay inspection apparatus is actually used, and a supervised signal can be obtained by associating the inspection result with the actual alignment signal (the image data). . That is, the overlay inspection result can be used as teacher data.

  Further, as the measurement principle of the focus detection system for measuring the position of the wafer in the Z direction in the semiconductor exposure apparatus, there is the above-described oblique incidence + image shift detection method (see FIG. 10). In this case, as in the case of the alignment measurement described above, the video signal from the photoelectric conversion element such as a CCD becomes an actual signal.

  For this focus measurement, the amount of focus shift can be inspected by the technique described in Japanese Patent Laid-Open No. 2005-101511 proposed by the present applicant. A supervised signal can be obtained by associating the inspection result with an actual signal. That is, the inspection result can be used as teacher data.

  Next, an embodiment of a device manufacturing method using the above-described exposure apparatus 1 will be described with reference to FIGS. FIG. 8 is a flowchart for explaining a manufacturing process of a device (a semiconductor chip such as an IC or LSI, an LCD, a CCD, or the like). Here, the manufacture of a semiconductor chip will be described as an example. In step 1 (circuit design), the device circuit is designed. In step 2 (mask production), a mask on which the designed circuit pattern is formed is produced. In step 3 (wafer manufacture), a wafer is manufactured using a material such as silicon. Step 4 (wafer process) is called a pre-process, and is a process for forming a semiconductor chip using a mask and a wafer, and includes processes such as an assembly process (dicing and bonding), a packaging process (chip encapsulation), and the like. In step 6 (inspection), inspections such as an operation confirmation test and a durability test of the semiconductor device created in step 5 are performed. Through these steps, the semiconductor device is completed and shipped (step 7).

  FIG. 9 is a detailed flowchart of the wafer process in Step 4. In step 11 (oxidation), the surface of the wafer is oxidized. In step 12 (CVD), an insulating film is formed on the surface of the wafer. In step 13 (electrode formation), an electrode is formed on the wafer by vapor deposition or the like. In step 14 (ion implantation), ions are implanted on the wafer. In step 15 (resist process), a photosensitive agent is applied to the wafer. Step 16 (exposure) uses the exposure apparatus 1 to expose a circuit pattern on the mask onto the wafer. In step 17 (development), the exposed wafer is developed. In step 18 (etching), portions other than the developed resist image are removed. In step 19 (resist stripping), the resist that has become unnecessary after the etching is removed. By repeatedly performing these steps, multiple circuit patterns are formed on the wafer. According to the present embodiment, since the exposure apparatus 1 has means for performing signal processing as described above, a high-quality semiconductor device can be manufactured with high yield.

(Other embodiments)
As mentioned above, although preferable embodiment of this invention was described, this invention is not limited to it, A various deformation | transformation and change are possible within the range of the summary. In the above-described embodiment, mainly the position detection in the Z direction of the wafer in the semiconductor exposure apparatus, that is, so-called focus detection has been described. However, the present invention can also be applied to position detection in the XY directions of a wafer in a semiconductor exposure apparatus, so-called alignment detection (measurement for wafer alignment), as mentioned in the description of the above embodiment.

In the above-described embodiment, the supervised signal is generated based on “actual signal and inspection data”. However, instead, one of the following two types of signals can be used to achieve the same purpose.
1) Signal obtained by simulation 2) Result of position recognition by real signal and high-accuracy signal processing Each will be described below.

(Embodiment using signals obtained by simulation)
If a signal is generated by simulation using object information and measurement system information that are actually measured, a supervised signal including a correct position to be recognized can be obtained.

  In the semiconductor process, the measured object information is process material and shape information of a pattern to be produced. For example, the shape after CMP of W or Cu. For that purpose, the shape data obtained by measuring the shape of the pattern obtained in the actual process with a measuring device (profiler) that uses an AFM (Atomic Force Microscope) or the like and that guarantees nanometer-order accuracy is used. Use it.

  When using an optical method for the detection system, include optical conditions such as the NA (numerical aperture) of the optical system and the wavelength of light used, and information on aberrations in the optical system caused by manufacturing and assembly errors. Simulate. By doing so, a highly accurate supervised signal can be obtained.

  FIG. 7 is a block diagram illustrating an outline of an embodiment in which a signal obtained by simulation is used as a supervised signal.

(Embodiment using position recognition result by real signal and high-accuracy signal processing)
The actual signal is the same as that in the above-described embodiment using the above-described “actual signal and inspection data”.

  Here, high-accuracy signal processing cannot be used for a detection system in an actual apparatus (for example, a focus detection system for measuring the position in the Z direction of a wafer of a semiconductor exposure apparatus) because the processing speed is low, for example. Is. Such signal processing is, for example, one that has been confirmed to be highly accurate by inspection data obtained by the method described in Japanese Patent Laid-Open No. 2005-101511.

  Although the processing time is long, the position recognition result obtained by processing the actual signal by the processing method with high accuracy is associated with the actual signal. By doing so, it becomes a supervised signal.

  FIG. 6 is a block diagram for explaining the outline of an embodiment in which the position recognition result by the real signal and high-precision signal processing is used as the supervised signal.

  In the above-described embodiment, the estimation of bmax that maximizes the similarity S is performed by curve fitting. However, the estimation method is not limited to this. For example, as another method, a method called “subsample accuracy search” can be used.

  FIG. 12 is a block diagram illustrating an outline of an embodiment using a subsample accuracy search.

  Hereinafter, a method based on this sub-sample accuracy search will be described.

  For the input signal, a center position is obtained by using a slice processing method that is an existing method, and the position is set to bp. Then, bp + w and bp−w separated from this position by ± w, and similarity S (bp + w), S (bp−w) and S (bp) at bp are calculated.

The three magnitude relationships of S (bp + w), S (bp−w), and S (bp) are checked.
When S (bp) is a single maximum value of three values, that is, when S (bp)> S (bp + w) and S (bp)> S (bp−w),
The new value of w is set to 1/2 of the original value, that is, w ← w / 2.
When S (bp + w) is the maximum of the three values, that is, when S (bp) ≦ S (bp + w) and S (bp−w) ≦ S (bp + w),
Let bp + w be a new bp, that is, bp ← bp + w.
When S (bp−w) is the maximum of the three values, that is, when S (bp) ≦ S (bp−w) and S (bp + w) ≦ S (bp−w),
Let bp-w be a new bp, that is, bp ← bp-w.

  The above process is repeated, and the process ends when the value of w falls below the preset subsample search accuracy th. Then, bp at this time is determined as bmax. FIG. 13 is a diagram showing the flow (algorithm) of this process.

  Moreover, in the above embodiment, the example which projects light on a wafer and detects the reflected light from the said wafer was shown. However, the measurement target is not limited to the wafer, and may be another object. The other object may be a reticle as an original.

  According to the embodiment described above, based on the signal distorted by the wafer process, highly accurate position recognition can be performed without reducing the throughput. For example, it is possible to improve the wafer position detection accuracy in the semiconductor exposure apparatus, thereby improving the yield of the semiconductor devices.

Block diagram for explaining the outline of the signal processing system of this embodiment Diagram showing the one-dimensional signal to be processed The figure which shows the signal processing by the existing method, and the flow of the signal processing in this embodiment The figure which shows the estimated amount of b which maximizes similarity Figure showing the result of curve fitting of similarities at several sample positions The block diagram explaining the outline of the signal processing system in other embodiment The block diagram explaining the outline of the signal processing system in other embodiment Flow diagram explaining the device manufacturing process Flow diagram explaining the wafer process The figure explaining the whole structure of a semiconductor exposure apparatus Diagram showing distorted signal The block diagram explaining the outline of the signal processing system in other embodiment Diagram showing search for approximate position by subsample accuracy search

Explanation of symbols

50 Focus tilt detection system 60 Control unit

Claims (12)

An exposure apparatus that exposes a substrate through an original plate,
Detection means for detecting reflected light from the object;
Signal processing means for determining the position of the object based on the detection signal detected by the detection means;
The detection signal is composed of N elements,
The signal processing means calculates a similarity when approximating the detection signal using M pre-determined orthogonal basis vectors smaller than the N, a plurality of different degrees between the detection signal and the orthogonal basis vectors. An exposure apparatus characterized in that a position of the object is obtained based on a plurality of similarities obtained with respect to a relative position.   2. The exposure apparatus according to claim 1, further comprising learning means for obtaining the M orthogonal basis vectors from a group of learning signals corresponding to the detection signals.   The exposure apparatus according to claim 2, wherein the learning signal is one of the detection signal and a signal obtained by simulation.   The exposure apparatus according to claim 2, further comprising a generation unit that generates information about a position to be recognized by the detection signal as a part of the information of the learning signal.   The exposure apparatus according to claim 2, wherein the learning unit obtains the M orthogonal basis vectors by Karhunen-Loeve expansion.   The signal processing means estimates a relative position between the detection signal having the maximum similarity and the orthogonal basis vector based on the plurality of similarities, and based on the estimated relative position, the target 4. The exposure apparatus according to claim 1, wherein the position of an object is obtained.   The exposure apparatus according to claim 1, wherein the object is one of the original plate and the substrate.   The exposure apparatus according to claim 1, wherein the object is the substrate, and the signal processing unit obtains a position of the substrate for either focusing or alignment. . A position detection device for determining the position of an object,
Detection means for detecting reflected light from the object;
Signal processing means for determining the position of the object based on the detection signal detected by the detection means;
The detection signal is composed of N elements,
The signal processing means calculates a similarity when approximating the detection signal using M pre-determined orthogonal basis vectors smaller than the N, a plurality of different degrees between the detection signal and the orthogonal basis vectors. A position detection device characterized in that a position of the object is obtained based on a plurality of similarities obtained with respect to a relative position. An exposure method for exposing a substrate through an original plate,
A detection step for detecting reflected light from the object;
A signal processing step for determining the position of the object based on the detection signal detected in the detection step;
The detection signal is composed of N elements,
In the signal processing step, the similarity when the detection signal is approximated using M pre-determined orthogonal basis vectors less than N is calculated as a plurality of different degrees of difference between the detection signal and the orthogonal basis vectors. An exposure method characterized in that a position of the object is obtained based on a plurality of similarities obtained with respect to a relative position. A position detection method for determining the position of an object,
A detection step for detecting reflected light from the object;
A signal processing step for determining the position of the object based on the detection signal detected in the detection step;
The detection signal is composed of N elements,
In the signal processing step, the similarity when the detection signal is approximated using M pre-determined orthogonal basis vectors less than N is calculated as a plurality of different degrees of difference between the detection signal and the orthogonal basis vectors. A position detection method characterized by obtaining a position of the object based on a plurality of similarities obtained with respect to a relative position. A device manufacturing method comprising the step of exposing a substrate through an original plate using the exposure apparatus according to claim 1.

Images