JPH0982620A - Method for detection of best focus position - Google Patents

Abstract

PROBLEM TO BE SOLVED: To compute the size of mask for measurement of focal point in a highly precise manner. SOLUTION: When computing the size (L) of the measurement pattern image (M) which is exposed to a photosensitive substrate for the purpose of determining the best focus position of a projection optical system (PL), first, the output signal of the pattern detecting system, with which the measurement pattern image on the substrate is detected, is sliced by a plurality of slice levels (S108), and the size of the measurement pattern image of each slice level is computed (S109). Then, the size of the obtained measurement pattern image is approximated by the function having the slice level as a variable (S110), and the size of the measurement pattern image is computed based on said approximate function (S111). As a result, the affection of the noise of the detection signal can be mitigated, and the size of the measurement pattern image can be computed in a highly precise manner.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for calculating the best focus position of a projection optical system in a projection exposure apparatus used for manufacturing a semiconductor device, a liquid crystal display device, a thin film magnetic head or the like in a lithography process. .

[0002]

2. Description of the Related Art For example, when manufacturing a semiconductor device, a liquid crystal display device, a thin film magnetic head or the like in a linography process,
2. Description of the Related Art A projection exposure apparatus is used that exposes a pattern image of a photomask or a reticle (hereinafter collectively referred to as “reticle”) onto a photosensitive substrate via a projection optical system. In order to expose the pattern image of the reticle on the photosensitive substrate with high resolution by using such a projection exposure apparatus, the exposure surface of the photosensitive substrate is located at the position (best position) where the best image plane is obtained in the optical axis direction of the projection optical system. It is necessary to adjust the focus position). For that purpose, it is necessary to obtain the best focus position of the projection optical system in advance by some method.

As a conventional best focus position measuring method, for example, as disclosed in JP-A-1-187817, the focus measuring mark m (FIG. 2) formed on the reticle is changed in focus position. A method is known in which the best focus position is obtained by so-called test printing in which a photosensitive substrate is exposed. Specifically, for example, as shown in FIG. 2, a focus measurement mark m is formed by arranging a plurality of pattern groups in which four elongated rhombic patterns are arranged at a predetermined pitch P1 in the width direction at a further predetermined pitch P2.
Are exposed on a photosensitive substrate (for example, a wafer coated with a photosensitive material) while changing the focus position (FIG. 3A).

Thereafter, the wafer is developed to convert the mark image M for focus measurement into a resist pattern having irregularities, and the length of each mark image M (mark length) is measured. In this case, when the exposure surface of the wafer is at the best focus position, the length L of the mark image M for focus measurement becomes maximum, so that the best focus position can be determined by measuring the mark L of the mark image M. You can ask.

In order to measure the mark length L, conventionally, for example, a He-Ne laser is irradiated in the vicinity of each mark image M as a vertically long slit-shaped beam SB as shown in FIG. 3 (a). To do. Then, while observing the position of the wafer stage on which the wafer is placed by the laser interferometer, the wafer stage is driven to relatively scan each mark image M and its laser beam SB. At this time, when the laser beam SB is on the mark image M, the mark image M
Since diffracted light or scattered light is generated in a predetermined direction from FIG.
As shown in (b), a waveform (profile) of the stage position (X) on the horizontal axis and the signal intensity (S) on the vertical axis can be obtained. Then, the distance between the intersections when the obtained signal waveform is sliced at a slice level close to the bottom is measured as the mark length L of the mark image M at the focus position.

In this way, conventionally, the mark length L of the measurement mark image M is obtained at each focus position, and the mark length L of these measurement mark images M is used as a function of the focus position, for example, curve approximation by the least square method. To do. Then, the focus position of the mark length L in which the obtained approximate curve shows the maximum value within a predetermined range is obtained, and the approximate curve is sliced at a slice level that is smaller than the maximum value by a predetermined value. The midpoint was the best focus position.

[0007]

However, in the conventional method, since the He--Ne laser having a short wavelength is used as the laser beam for scanning the measuring mark, the diffracted light from the measuring mark image M is used. Alternatively, scattered light is easily affected by interference, and an error is included in the measurement value of the mark length L of the measurement mark image M. Therefore, the calculated best focus position is the line and space mark (hereinafter referred to as L / S mark). There is a case in which there is an offset error with respect to the best focus position of (.

Further, the actual length L of the measuring mark M obtained at each focus position is close to the distance between the intersections when the signal waveform (signal profile) is sliced at a slice level close to the bottom. However, since the bottom portion is easily affected by noise or the like, the signal waveform cannot be sliced at a very low slice level in measurement, and therefore the actual length L of the measurement mark image M cannot be obtained. There was a problem.

A method of measuring the mark length of the measurement mark by using an image processing technique has also been proposed. However, as the measurement mark, as shown in FIG. When using, it was difficult to measure the length up to the narrow parts at both ends with high accuracy. The present invention has been made in view of the above problems, and occurs when measuring the mark length of the measurement mark image,
The measurement error caused by the interference of He-Ne laser and the like is minimized to improve the measurement accuracy, and the difference between the best focus position of the L / S mark and the best focus position of the measurement mark caused by the interference is reduced. It is an object of the present invention to provide a method of detecting the best focus position that can be solved and focus management can be performed easily.

[0010]

In order to solve the above problems, the present invention provides an optical axis direction of a projection optical system (PL) for projecting an image (m) of a pattern on a mask (R) onto a predetermined surface. Of the plurality of measurement pattern images (M) formed on the base hill (W) by projecting the measurement pattern image (M) by disposing the photosensitive substrate (W) at each of the plurality of positions. An output signal of a pattern detection system (120) for detecting the measurement pattern image (M) on the base slope (W) when the best focus position of the projection optical system (PL) is detected based on the size (L). Is sliced at a plurality of slice levels and the size (L) of the measurement pattern image (M) at each slice level is obtained, and the size of the measurement pattern image (M) obtained at the first process (L) is approximated by a function with the slice level as a variable, It is characterized by performing a second step of calculating the size of the (L) of the measurement pattern image based on the approximated function (M). Therefore, according to the present invention, since the size of the measurement pattern image is indirectly measured based on the approximated function, it is less likely to be affected by interference,
More accurate measurement can be performed.

In this case, at least one measurement pattern is used.
It is also possible to determine the position in the optical axis direction, which is composed of three wedge-shaped marks and has the maximum length of the measurement pattern image calculated in the second step, as the best focus position. In addition, the measurement pattern is configured as a diffraction grating including a plurality of wedge-shaped marks, and the pattern detection system (120) receives diffracted light generated from the measurement pattern image by relative scanning between the measurement pattern image and the light beam. Alternatively, the photoelectric signal may be output according to the intensity of the diffracted light.

Further, in the second step, it is preferable to calculate the size of the image of the measurement pattern when the slice level is substantially zero using the approximated function. Further, in the first step, the output signal of the pattern detection system may be smoothed. In the second step, the square of the difference between the size of the measurement pattern image obtained by the approximated function and the size of the measurement pattern image obtained in the first step, and the weight corresponding to the slice level The function approximated by the least squares method that minimizes the sum of the products for each slice level may be optimized. At this time, a lower slice level may be given a greater weight.

[0013]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of a method for detecting a best focus position according to the present invention will be described in detail below with reference to FIGS. In the present embodiment, the present invention is applied when obtaining the best focus position of the projection optical system in the reduction projection type exposure apparatus (stepper). FIG. 1 shows the configuration of a stepper to which this embodiment is applied. In FIG. 1, light (g line, i line, etc.) from a mercury discharge lamp 102 as a light source for exposure is condensed by an elliptical mirror 104. After that, the light passes through the exposure amount control shutter 106. The light passing through the shutter 106 is made uniform in illuminance by the optical integrator 108, and then the reticle R is illuminated via the main condenser lens 110. From the discharge lamp 102 to the main condenser lens 110, a so-called exposure illumination optical system is configured, and assuming that the emission intensity of the discharge lamp 102 is substantially constant, the shutter controller 112 controls the opening time of the shutter 106. A constant exposure amount can always be obtained. The reticle R is a reticle stage R that is finely moved in the two-dimensional plane (in the X, Y and rotation directions).
The light held on S and transmitted through various patterns formed in the pattern area of the reticle R is image-projected onto the wafer W by the image-side (or both-side) telecentric projection optical system PL. The wafer W is placed on the wafer stage ST.

Initialization of the reticle R is performed by finely moving the reticle stage RS based on a mark detection signal from a reticle alignment system 114 that photoelectrically detects an alignment mark around the reticle R. On the other hand, the wafer stage ST mounts the wafer W and steps it in a fixed amount in the X and Y directions to print an image of the pattern area of the reticle R for each shot area (partial area) on the wafer W. Move to. The wafer stage ST also moves when photoelectrically detecting various patterns (alignment marks and resist patterns) on the wafer W. The operation of the wafer stage ST is controlled by the stage controller 116. The stage controller 116 is provided with a driving motor (not shown) and a laser interferometer (not shown) for sequentially measuring the coordinate values of the wafer stage ST.

Further, this type of stepper includes a wafer W
A wafer alignment system for detecting the above various patterns (alignment marks, etc.) is provided. In the present embodiment, slit-shaped spot light is projected onto the wafer W via the projection optical system PL, and optical information, particularly diffracted light or scattered light, from the mark pattern irradiated on the spot light is projected again. A TTL (through-the-lens) type wafer alignment system 120 for detecting through the system PL is provided.

In the wafer alignment system 120, H
A laser beam from a short wavelength laser light source 121 such as an e-Ne laser light source or an Ar ion laser light source is bent by a mirror 125 via a lens system 122 including a cylindrical lens, a beam splitter 123 and an objective lens 124. The light is transmitted so as to pass through the center of the entrance pupil of the projection optical system PL. The laser beam is vertically irradiated onto the wafer W from the off-axis portion of the projection optical system PL, and becomes a slit-shaped spot light that extends in one direction on the wafer W by the action of the lens system 126.

The return light from the wafer W travels backward in the projection optical system PL, is reflected by the beam splitter 123 via the objective lens 124, and passes through the pupil relay system 126, the spatial filter 127 and the condenser lens 128. The light is received by the photoelectric conversion element 129. The spatial filter 127 is arranged in a substantially conjugate relationship with the pupil plane of the projection optical system PL, and blocks specularly reflected light of the return light from the wafer W and allows diffracted light and scattered light to pass through. The signal photoelectrically converted by the photoelectric conversion element 129 is input to the signal processing system 130, where the position of the mark image is detected based on the waveform corresponding to the profile of the mark image (pattern image). At this time, the signal processing system 130 is obtained when the spot light and the wafer W are relatively moved using the position measurement pulse (for example, one pulse every 0.02 μm) from the laser interferometer in the stage controller 116. The signal waveform from the photoelectric conversion element 129 is sampled. In the present embodiment, such T
It is assumed that the position of the resist pattern IR formed on the wafer W is automatically measured using the TL type wafer alignment system 120 and the signal processing system 130.

When the pattern of the reticle R is printed on the wafer W, the best image plane of the projection optical system PL, that is, the plane on which the pattern image of the reticle R is imaged with the highest contrast, and the resist surface of the wafer W. And must match exactly. Therefore, in the present embodiment, the light from the light source 140 which is non-photosensitive to the resist layer is formed into an image-forming light beam by the projection optical system 141 and obliquely onto the wafer W (5 ° to 20 ° with respect to the wafer surface). An oblique incidence type focus detection system (AF sensor) is provided in which the reflected light is projected and received by the photoelectric detector 144 via the light receiving optical system 142 and the slit 143. In this AF sensor, when the best image plane of the projection optical system PL and the surface of the wafer W coincide with each other, the photoelectric detector 144 outputs a focus signal indicating that it is in focus, so that the best image plane is formed. When the wafer surface is vertically displaced (optical axis), a focus signal corresponding to the amount of displacement (for example, within ± several μm) is output. The focus signal indicating the focus and the defocus is the focus control unit (hereinafter referred to as “AF unit”).
145).

Further, the wafer W is mounted on the wafer stage ST.
There is also provided a Z stage that makes a minute movement in the optical axis direction of the projection optical system PL, and a θ stage that makes a minute rotation of the wafer W in the XY plane. Therefore, by driving the Z stage under the control of the stage controller 116 in response to the focus signal, the autofocus operation can be performed. A part of the optical system of the AF sensor or A
The F unit 145 is also provided with an offset section that shifts the surface position of the wafer W when the focus signal indicates the focus in the optical axis direction of the projection optical system PL. An arbitrary shift amount can be set from the system 134.

Hereinafter, with reference to the flowchart of FIG.
The operation of obtaining the best focus position with respect to the projection optical system PL by the method according to the present embodiment will be described. A mark m for focus measurement is formed on the pattern forming surface of the reticle R of the present embodiment. The focus measurement mark m is a pattern group in which a plurality of (for example, four) elongated rhombic patterns are arranged at a predetermined pitch P1 in the width direction, as in the case described in the conventional example with reference to FIG. Are arranged at another predetermined pitch P2. Then, FIG. 3A shows a mark image M obtained by projecting the mark m for the focus measurement on the wafer W. According to the present embodiment, when the exposure surface of the wafer W is at the best focus position in the direction along the optical axis of the projection optical system PL, the length L of the mark image M in the X direction becomes maximum. Using this, the best focus position can be obtained as follows.

First, in step S101 of FIG.
The focus position of the wafer W as a photosensitive substrate is set to an initial value. For example, the initial value is set to the lower limit position or the upper limit position of the range in which the exposure surface of the wafer W is moved in the test print this time around the focus position which has been the best focus position. Then, in step S102, the image M of the focus measurement mark m of the reticle R is exposed on the wafer W as a photosensitive substrate. At this time, the wafer stage ST is stepped in a direction perpendicular to the optical axis of the projection optical system PL to expose the pattern of the reticle R, so that, for example, at the same focus position, for example, 3
The image M of the mark m for focus measurement is exposed in each shot area.

Then, from step S103 to step S
In step 104, the focus position on the exposure surface of the wafer W is changed by a predetermined step amount. When the initial value of the focus position on the exposure surface of the wafer W is the lower limit of the moving range, the focus position gradually rises, and when the initial value is the upper limit of the moving range, the focus position gradually lowers. Then, the process proceeds to step S102 again, and the images M of the marks m for focus measurement are exposed in shot areas at a plurality of locations (for example, three locations) in the unexposed area on the exposed surface of the wafer W as the photosensitive substrate. . Then, steps S104 and S102 are repeated until the focus position reaches the lower limit or the upper limit of the moving range, and when the focus position reaches the lower limit or the upper limit of the moving range, the operation shifts from step S103 to step S105. ,
The wafer W is developed. When performing exposure, the exposure time can be averaged by slightly changing the exposure time for each exposure process. By thus averaging the exposure times, the influence of interference peculiar to a certain exposure time (for example, when the diffracted light emitted from the tip of the rhombus mark becomes extremely small and the measurement accuracy deteriorates at a certain exposure time) is removed. This makes it possible to improve the measurement accuracy of step S106 and thereafter.

Next, in step S106, the mark length which is the length of the mark image for focus measurement on the wafer W after development is measured. FIG. 3A shows a mark image M for focus measurement of the measurement target. In FIG. 3A, the laser beam emitted from, for example, the He-Ne laser light source 121a of the wafer alignment system 120 of FIG. The scanning beam SB having a slit shape that is long in the Y direction is irradiated to the vicinity of the focus measurement mark image M on the wafer W via the projection optical system PL. In such a state, the wafer stage S of FIG.
When T is moved in the −X direction, while the slit-shaped scanning beam SB irradiates the mark image M for focus measurement, diffracted light is emitted from the mark image M functioning as a diffraction grating in a predetermined direction. . Therefore, when the diffracted light is detected by the photoelectric conversion element 129 of FIG. 1 and the detection signal S output from the photoelectric conversion element 129 is plotted with respect to the X coordinate of the wafer stage ST, as shown in FIG. 3B. In addition, the value of the detection signal S shows a profile that increases in the region where the mark image M exists in the X direction. Therefore, the length in the X direction of the range in which the detection signal S exceeds the predetermined threshold value can be set as the mark length of the mark image M for focus measurement.

The signal processing system 130 shown in FIG.
As shown in, the gate is set in the range of the width 2G in the X direction centering on the position of the reference intensity with respect to the detection signal S, and only the detection signal S within the range of the gate is processed. The detection signal S is represented by a curve 22 in which the mark length has a length of about 0 μm at the time of defocus and a mark length has a design value at the best focus position. Even the characteristics that change. Therefore, in order to deal with all the states, the width G on one side of the gate in the X direction at the time of signal processing is opened from the position where the signal intensity is maximum to a range of 0 to 1/2 or more of the mark length. Need to be.

The measurement result of each mark length thus obtained by the signal processing system 130 of FIG. 1 is stored in the disk device or the like connected to the main control system 134 of FIG. 1 together with the data of the corresponding focus position. It is stored in a file in the device. At that time, in step S107, if the reliability of the profile of the detection signal S (particularly the profile of the bottom portion) is deteriorated due to the influence of high frequency noise, for example, smoothing such as invalidating skipped data exceeding a predetermined set value is performed. You may give a conversion process.

Next, in step S108, the signal processing system 130 sets a plurality of slice levels Si (i = 1, 2, 3, ...) For the profile of the detection signal S.
The slice processing is performed with. In this slicing process, as shown in FIG. 6, for example, the largest intensity of the detection signal S is used as the reference intensity (100%) and sliced from 10% to 90% in 10% steps. Then, step S1
09, the coordinates X1i and X2i at two positions in the X direction where the detection signal S crosses each slice level Si are obtained, and
The interval | X1i-X2i | of the coordinates at one place is the mark length at each slice level. Although the slice level is set in 10% increments in the present embodiment, it goes without saying that the slice level can be set to any value. Further, it is not necessary to set the slice levels at equal intervals, and for example, the slice levels can be set at narrower intervals near the bottom of the profile of the detection signal.

The mark lengths at each slice level thus obtained are plotted as shown by dots in FIG. 7, and in step S110, the mark lengths are curve-approximated by an nth-order function having the slice level as a variable. . For example, the model function formula of the approximate curve is expressed by the following formula (1):
The next function. y = ax4 + bx3 + cx2 + dx + e (1) Then, when the approximate curve is obtained by the least square method using the above model function formula (1), it can be plotted as the curve shown by the solid line in FIG. 7.

Then, in step S111, the mark length L of the actually exposed image M is obtained from the approximate curve obtained as described above. Specifically, in the model function expression (1), the value of y when x = 0 (that is,
In FIG. 7, the y value at a slice level of 0%) represents the spread (width) of the profile of the detection signal S, so the signal processing system 130 uses this value as the mark of the actually exposed image M. The length L is stored in a predetermined storage device. As described above, in the present embodiment, the mark length of the image of the measurement mark is not directly obtained from the detection signal in which noise is likely to be generated due to the influence of interference or the like as in the conventional case, but the mark length is once determined according to the signal strength. The curve length is approximated by a function that uses the slice level as a variable, and the mark length is calculated based on the approximated function, so it is possible to suppress the influence of noise caused by interference of short wavelength lasers, etc. It becomes possible to measure.

Further, the actual s mark length L of the image of the measurement mark corresponds to the distance between the intersections when the portion close to the bottom is sliced. With the method, it was not possible to slice at a portion close to the bottom. However, according to the present embodiment, it is possible to obtain the inter-intersection distance when slicing at a portion closer to the bottom (for example, slice level 0%), irrespective of noise particularly on the bottom portion. The mark length L close to the value can be calculated.

In this way, the mark length of the image of the measurement mark at a certain focus position is obtained. After that, steps S106 to S111 are repeated, and when it is determined in step S112 that the mark lengths of the images of the measurement marks at all the focus positions have been calculated, the process proceeds to step S113 and subsequent steps, for example, Japanese Patent Laid-Open No. The best focus position is determined by the method disclosed in Japanese Patent Laid-Open No. 216004.

Specifically, in step S113,
Based on the mark length Li (i = 0, 1, 2, ...) At each focus position Fi (i = 0, 1, 2, ...) By the method of least squares, the mark length Li is used as a focus position Fi as a variable. It is approximated by an n-order function (for example, n = 4, 5, 6). In this case, for example, when the number of focus positions Fi is 7 or more, it is approximated by a 6th order function of the focus position F, when it is 6, it is approximated by a 5th order function, and when the number of focus positions is 5, Is approximated by a quartic function, and when the number of focus positions is four or less, the calculation of the best focus position can be stopped because there is a problem with the measurement conditions.

In FIG. 8, the profile of the approximate function obtained by the method of least squares in step S113 is shown by the curve 2
It is shown as 3. Then, in step S114, the maximum value M1 of the approximate curve 23 within the range of the focus position F having the effective mark length data, and the focus position F when the approximate curve 23 takes the maximum value M1.
Ask for 1. Next, in step S115, when a threshold value (M1-T) smaller than the maximum value M1 by a preset slice value T is set, and the approximated curve crosses the threshold value (M1-T), Focus position F1
From-side focus position F2A and focus position F
The focus position F2A on the + side is calculated from 1 and the intermediate (average) focus position F2 between these focus positions F2A and F2B is set as the best focus position. In this way, according to the present embodiment, the projection optical system P shown in FIG.
The best focus position of L is determined. When determining the best focus position, weighting according to each focus position may be added.

In the above embodiment, the diamond-shaped mark is used as the measurement mark. However, as shown in FIG. 9A, for example, a linear pattern in which the line width changes continuously,
For example, it is possible to use a pattern in which a predetermined number (for example, 3) of wedge-shaped patterns are arranged at regular intervals. By scanning the image P of such a pattern with a slit-shaped laser beam as in the above-described embodiment, a profile of the detection signal as shown in FIG. 9B can be obtained. It can be configured to solicit. Also in this case, the profile of the detection signal is sliced at a plurality of slice levels, and the length of the pattern obtained at each slice level is curve-approximated by a function having the slice level as a variable, so that noise due to interference or the like It is possible to reduce the influence and improve the measurement accuracy of the pattern length.

Further, in the above embodiment, when the actual mark length of the measurement mark is obtained from the approximate curve of the measurement mark size with the slice level as a variable, when x = 0 in the model function, Value of y (slice level 0%
However, the present invention is not limited to such an example. For example, when the slope of the approximate curve changes rapidly near the slice level of 0% and an accurate mark length cannot be obtained, a predetermined value close to x = 0, for example, y at the slice level of 1%. The value of may be set as the mark length.

Further, in the above embodiment, the mark length is obtained by using the approximation curve smoothed in step S107 of FIG. 5, but the approximation curve is weighted to further improve the reliability. You may give a process. Specifically, a weight is set for the mark length obtained in step S111 of FIG. 5 according to each slice level. As described above, in general, the lower the slice level (that is, the closer to the bottom of the profile of the detection signal), the closer the distance between the points of intersection with the approximated curve and the true mark length is. Try to give weight. However, since the lower the slice level is, the more susceptible it is to noise, depending on the S / N ratio of the detection signal, the lower the slice level, the lighter the weight may be.
Using the weights set in this way, the model function can be optimized by the least squares method of weighting the model function. That is, step S1 in FIG.
The sum of the products of the square of the difference between the mark length value calculated according to the approximation function obtained in step 10 and the mark length value actually measured in step S106, and the weight of the corresponding focus position is the minimum. Set the parameters of the model function so that By thus determining the mark length based on the model function optimized by the weighted least squares method, it is possible to reduce the influence of noise particularly near the bottom and further improve the measurement accuracy.

In the above-described embodiment, the output signal of the wafer alignment system using spot light is sliced at a plurality of slice levels. However, for example, the output signal of the wafer alignment system equipped with an image pickup device (CCD or the like). May be sliced at multiple slice levels. The measurement mark does not have to be in the shape of a diffraction grating, and may be at least one mark. Furthermore, the measurement mark does not have to be wedge-shaped, and may be, for example, a linear mark (bar mark or the like).

Although the preferred embodiments of the present invention have been described above with reference to the accompanying drawings, it goes without saying that the present invention is not limited to such embodiments. It is clear that those skilled in the art will come up with various changes or modifications without departing from the scope of the invention described in the claims, and it is understood that these also belong to the technical scope of the present invention. It

[0038]

As described above, according to the present invention,
Instead of directly obtaining the size of the measurement pattern image from the detection signal detected by the pattern detection system, the detection signal is sliced at a plurality of slice levels, and then the size of the pattern image at each slice level is obtained. Curve approximation is performed using a model function of the size of the pattern image with each slice level as a variable, and the size of the pattern image actually exposed on the substrate is calculated from the obtained approximation curve. As described above, according to the present invention, the short wavelength H
Since it is possible to reduce the influence of the noise of the detection signal due to the influence of the interference of the e-Ne laser or the like, it is possible to calculate the size of the pattern image with higher accuracy.
Further, since there is no difference between the best focus of the L / S mark (device pattern on the reticle) and the best focus of the rhombus mark caused by the influence of interference, focus management can be efficiently performed.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an example of a stepper to which a method for detecting a best focus position according to the present invention can be applied.

FIG. 2 is a schematic plan view showing an example of a focus measurement mark provided on a reticle.

3A is a schematic plan view showing a focus measurement mark image and a slit-shaped beam projected and exposed on a photosensitive substrate, and FIG. 3B is a detection result obtained from the mark image of FIG. 3A. It is a wave form diagram which shows the profile of a signal.

FIG. 4 is a waveform diagram illustrating how a gate is set in a detection signal obtained from a focus measurement mark image.

FIG. 5 is a flowchart showing an example of a best focus position detection method according to the present invention.

FIG. 6 is a waveform diagram illustrating how a detection signal waveform obtained from a focus measurement mark image is sliced at a plurality of slice levels.

FIG. 7 is a graph illustrating how the length of the mark image at each slice level is curve-fitted by a function having each slice level as a variable.

FIG. 8 is a graph showing how the best focus position is determined based on the mark length at each focus position.

9A and 9B show another focus measurement mark applicable to the present invention, in which FIG. 9A shows a mark shape applied on a reticle, and FIG. 9B shows the mark of FIG. It is a waveform diagram which shows the signal waveform obtained from the mark image which carried out the projection exposure on the board | substrate.

[Explanation of symbols]

 R reticle RS reticle stage PL projection optical system W wafer m focus measurement mark on reticle M focus measurement mark image on photosensitive substrate SB slit beam 120 wafer alignment system 121 He-Ne laser light source 129 photoelectric conversion element 130 signal processing system

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G03F 9/02 G03B 3/00 A H01L 21/30 516A 526A

Claims (7)

[Claims] 1. A photosensitive substrate is arranged at each of a plurality of positions in the optical axis direction of a projection optical system that projects an image of a pattern on a mask onto a predetermined surface, and an image of a measurement pattern is projected. A method of detecting the best focus position of the projection optical system based on the sizes of a plurality of measurement pattern images formed on a base slope, the output of a pattern detection system detecting the measurement pattern image on the base slope. A first step of slicing a signal at a plurality of slice levels and obtaining the size of the measurement pattern image at each slice level; and the slice level of the size of the measurement pattern image obtained in the first step. A second step of approximating with a function as a variable and calculating the size of the measurement pattern image based on the approximated function. . 2. The measurement pattern includes at least one wedge-shaped mark, and the position in the optical axis direction where the length of the measurement pattern image calculated in the second step is maximum is determined as the best focus position. The method according to claim 1, wherein 3. The measurement pattern is a diffraction grating made up of a plurality of wedge-shaped marks, and the pattern detection system causes diffracted light generated from the measurement pattern image by relative scanning between the measurement pattern image and a light beam. 3. The method according to claim 1 or 2, wherein the light is received and a photoelectric signal corresponding to the intensity of the diffracted light is output. 4. The size of the measurement pattern image when the slice level is substantially zero is calculated in the second step by using the approximated function. The method described in either. 5. The method according to claim 1, wherein the first step includes a step of smoothing an output signal of the pattern detection system. 6. The square of the difference between the size of the measurement pattern image obtained by the approximated function in the second step and the size of the measurement pattern image obtained in the first step, and the slice. Of the product with the weight according to the level,
The function is optimized by a least squares method that minimizes the sum for each slice level.
5. The method according to any one of 5 above. 7. The method according to claim 6, wherein the lower the slice level, the greater the weight.