JPH0845813A - Scanning exposure method and device - Google Patents

Abstract

PURPOSE:To accurately follow up a target surface on the surface of a wafer without using the low-pass filter of a variable cut-off frequency and regardless of the condition of the surface of the wafer and scanning velocity when auto- focus is performed in a scanning exposure device. CONSTITUTION:A focus signal is supplied to a filter part 51 from an AF sensor 25 to obtain the estimate of the true value of the focus signal in the filter part 51. The focus position of the approximate surface of the surface of a wafer and a tilt angle are computed by using the estimate in least square approximate computing part 52, and a deviation obtained by subtracting the computation from a target value in a subtrating part 54 is supplied to a control part 62 to drive an actuator 63 by a control signal from the control part 62. The gain of the filter part 51 is optimized by using the dispersion of a focus signal when the wafer is at a standstill or moving.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is suitable for a scanning exposure method and a method for controlling an autofocus mechanism or an autoleveling mechanism of a scanning type exposure apparatus for manufacturing a semiconductor device, a liquid crystal display device or the like. Regarding the device.

[0002]

2. Description of the Related Art In a projection exposure apparatus used in a lithography process for manufacturing a semiconductor device, etc.
In a batch exposure method (static type) projection exposure apparatus such as an and repeat method (stepper, etc.), the wafer (or the substrate such as a glass plate) to be focused is stationary, and therefore, auto focus or auto focus is performed. Leveling is performed by servo control for a constant target value.

FIG. 11 is a simplified view of a conventional auto-focusing and auto-leveling mechanism. In FIG. 11, the focus position of the wafer (the position in the optical axis direction of the projection optical system) is shown in the first input section of the adder / subtractor section 74. ), And the target values of the tilt angles in the two directions are supplied. The actual focus positions at N (N is an integer of 3 or more) measurement points on the target surface (wafer surface) are detected by the focus position detection system (AF sensor) 72, and the detected N positions are detected. Is supplied to the least-squares approximation calculation unit 73, and the least-squares approximation calculation unit 73 uses these N focus positions to focus on the plane of the plane that is approximated to the surface of the wafer by the method of least squares, and Find the tilt angles in two directions.

FIG. 12 (a) shows an example of an AF sensor for one measurement point. In FIG. 12 (a), the detection light passing through the slit plate 77 passes through the light-transmitting side objective lens 78 and the wafer. A slit image is obliquely formed on the surface of 79, and the reflected light from the wafer 79 re-images the slit image on the photoelectric detector 81 via the light-receiving side objective lens 80. Wafer 79
When the focus position of the wafer changes to, for example, the position 79A, the focus position of the wafer 79 is detected from the detection signal of the photoelectric detector 81 because the image formation position on the photoelectric detector 81 is laterally displaced.

Returning to FIG. 11, the focus position and the tilt angles in the two directions measured in this way are determined by the second addition / subtraction unit 74.
The position deviation obtained by subtracting the measured value from the target value in the adder / subtractor 74 is output to the controller 75. An actuator 96 for changing the attitude of the wafer is driven based on a control signal obtained by the control unit 75 multiplying the positional deviation by a predetermined gain, and the changed focus position of the wafer is detected by the AF sensor. By being fed back by 72, a normal position servo system is constructed.

In the projection exposure apparatus of the batch exposure system (static type), since the stationary wafer is automatically focused, the disturbance w (k) on the target surface does not exist, and the entire servo system is used. The dominant noise among the generated noise is the fluctuation v (k) of the AF sensor 72. This fluctuation v (k) changes the optical path length due to the flow of gas on the optical path of the detection light of the AF sensor, etc.
This is a phenomenon in which the surface of 9 appears to be displaced. In FIG. 11, an adder 71 is virtually provided between the actuator 76 and the AF sensor 72, and the fluctuation v (k) of the AF sensor is mixed from the adder 71. In the conventional device, in order to reduce the influence of the fluctuation v (k), the output signal of the AF sensor 72 is averaged over time. In this case, since the wafer moves a minute amount even when it is stationary, the time average can be regarded as a moving average.

On the other hand, in recent years, one chip pattern of a semiconductor element or the like tends to be large, and thus an exposure apparatus for exposing a larger reticle pattern onto a wafer is required. Therefore, as an exposure apparatus that exposes a large-area pattern without imposing a heavy burden on the projection optical system, while scanning the reticle, the reticle is scanned in a first direction across the optical axis of the projection optical system. In sync,
A scanning type exposure apparatus such as a slit scan method or a step-and-scan method that sequentially exposes a pattern of a reticle to each shot area of the wafer by scanning the wafer in a second direction corresponding to the first direction is provided. Attention has been paid. In the case of such a scanning type exposure apparatus, FIG.
As shown in (b), since the exposure is performed by scanning the wafer 79 with respect to the predetermined exposure area 82 in the exposure field of the projection optical system, a larger shot is obtained by using the projection optical system having the same diameter. The area can be exposed.

In the scanning type exposure apparatus, when the wafer moves in the scanning direction, irregularities or steps on the surface of the wafer are observed by the AF sensor as a time series signal. Of these, Figure 1
As shown in FIG. 2 (b), since it is impossible in principle to follow the uneven area 79a having a pitch (wavelength) shorter than the exposure area 82 on the wafer 79 in the scanning direction by autofocusing, Is the disturbance w on the target surface
It can be regarded as noise due to (k). When FIG. 11 is applied to the scanning type exposure apparatus, the addition / subtraction unit 7 is virtually
In step 4, the disturbance w (k) of the focusing target surface is introduced.

[0009]

As described above, in the auto-focusing and auto-leveling mechanism of the scanning type exposure apparatus, the disturbance w on the focusing target surface which does not appear in the collective exposure method.
(K) becomes a new noise source, and the focusing operation becomes unstable. Usually, when performing exposure by the scanning exposure method, the scanning speed is variable over a wide range for controlling the exposure amount of the wafer. Therefore, the frequency distribution of the noise component due to the disturbance on the focusing target surface varies widely according to the scanning speed, and therefore, when considering the averaging, that is, the noise removal by the low-pass filter, the cutoff frequency is changed. It must be variable according to the scanning speed. However, it is difficult to obtain the optimum cutoff frequency by trial and error according to the scanning speed.

In view of the above point, the present invention does not use a low-pass filter having a variable cutoff frequency when performing exposure by a scanning exposure method, and the surface condition of an object to be focused (photosensitive substrate). It is an object of the present invention to provide a scanning exposure method capable of accurately causing the surface of an in-focus object to follow a target surface by autofocusing or autoleveling regardless of the scanning speed. Another object of the present invention is to provide a scanning exposure apparatus capable of carrying out such a scanning exposure method.

[0011]

In a scanning exposure method according to the present invention, a mask (7) on which a transfer pattern is formed is illuminated with exposure illumination light, and the pattern of the mask (7) is projected onto a projection optical system ( 11) through the photosensitive substrate (12) and scan the mask (7) in a predetermined direction (+ X direction), the substrate (12) is moved in a predetermined direction (-X direction). )
By scanning on the mask (7) on the substrate (12)
In the scanning exposure method for sequentially exposing the pattern of (2), the focus position detecting means (25, 3) for outputting a detection signal according to the position of the surface of the substrate (12) in the optical axis direction of the projection optical system
4) and filter means (5) for increasing / decreasing the detection signal output from the focus position detecting means (25, 34) with a predetermined gain.
1A to 51I) and height adjusting means (16A to 16C) for adjusting the position of the substrate (12) in the optical axis direction of the projection optical system based on the detection signal output from the filter means.
And a variation of the detection signal from the focus position detection means (25, 34) when the substrate (12) is stationary,
And the predetermined gain in the filter means (51A to 51I) based on the variation of the detection signal from the focus position detection means (25, 34) when the substrate (12) is being scanned.

In this case, when the substrate (12) is stationary, the variation of the detection signal from the focus position detecting means is previously obtained as the first variation, and the mask (7) and the substrate (12) are synchronized. In each running period from the start of scanning to the start of exposure in the case of scanning, the variation of the detection signal from the focus position detecting means is obtained as the second variation, and the first variation and the first variation thereof are obtained. The predetermined gain determined based on the variation of 2 may be used to operate the filter means during the exposure period following the run-up period.

Further, the scanning exposure apparatus according to the present invention illuminates the mask (7) on which the transfer pattern is formed with the illumination light for exposure, and projects the pattern of the mask (7) onto the projection optical system (1).
1) via the substrate stage (15X, 15Y) onto the photosensitive substrate (12) and scan the mask (7) in a predetermined direction in synchronization with the substrate stage via the substrate stage. In a scanning exposure apparatus that sequentially exposes the pattern of the mask on the substrate by scanning (12) in a predetermined direction, it depends on the position of the surface of the substrate (12) in the optical axis direction of the projection optical system. Focus position detecting means (25, 34) for outputting a detection signal, filter means (51A to 51I) for increasing / decreasing the detection signal output from the focus position detecting means with a predetermined gain, and output from this filter means. Based on the detection signal, the substrate stage (15X, 15X
Height adjusting means (16A to 16C) for adjusting the position of the substrate (12) in the optical axis direction of the projection optical system with respect to Y);
Based on the variation of the detection signal from the focus position detection means when the substrate (12) is stationary and the variation of the detection signal from the focus position detection means when the substrate (12) is scanned, And a calculating means (53) for obtaining the predetermined gain in the filter means.

[0014]

According to such a scanning exposure method of the present invention, focusing on the statistical amount of the noise component, a means for measuring the statistical amount of noise is incorporated in the servo system for applying autofocus or autoleveling, The gain of the filter means (51A to 51I) is optimized based on the measured noise statistic, and filtering is executed using this optimized filter means.

The principle of the present invention will be described with reference to FIG. FIG. 11 shows a conventional basic model of an auto-focusing and auto-leveling mechanism. In FIG. 11, AF sensor 72 as a focus position detecting means is used to measure N measurement points on the surface of a wafer as a photosensitive substrate. The focus position of the plane is measured, the focus position of the plane that best matches the N points, for example, in the least squares method, and the tilt angle are obtained in the least square approximation calculation unit 73, and the measured values are calculated from the target values in the addition / subtraction unit 74. The deviation is obtained by subtracting the deviation, and this deviation is supplied to the actuator 76 as height adjusting means via the control unit 75. The focus position measured by the AF sensor 72 is the deviation amount (focus position deviation) of the position of the surface of the wafer with respect to the position of the reference plane (image plane of the projection optical system). Further, in the virtual adding unit 71, the fluctuation v (k) of the AF sensor 72 is the first
The noise w of the focusing target surface is virtually mixed in as the second noise in the adder / subtractor 74.

The system shown in FIG. 11 can be described by the state equation at time k as follows.

[0017]

## EQU1 ## x (k + 1) = x (k) + w (k), y (k) = x (k) + v (k) In this expression, x (k) is the focus position deviation at the time k on the wafer surface. , X (k + 1) is the focus position deviation at the time (k + 1) on the wafer surface, and y (k) is the observed amount of the focus position deviation at the time k.

Here, it is assumed that the fluctuation v (k) of the AF sensor and the disturbance w (k) of the focusing target surface are both stationary processes, and their optimum estimation amounts are E {v (k)} and E, respectively.
Assuming {w (k)}, the following equation holds. E {v (k)} = 0, E {w (k)} = 0 Next, when performing exposure by the scanning exposure method, for example, as shown in FIG. The wafer 12 as a photosensitive substrate is scanned in the + X direction.
At this time, the scanning speed V _R on the mask side changes, for example, as shown in FIG. 10, the speed of the wafer 12 in the X direction also changes with substantially the same tendency, and as shown in FIG. 9 corresponding to the acceleration period T1. The run-up section 67A is provided in front of the shot area 64A. Consider that auto-focusing or auto-leveling is performed when the exposure area 13 passes through the approach area 67A and reaches the shot area 64A.

First, the variance is obtained as a statistic representing the amount of fluctuation v (k) of the AF sensor and the amount of disturbance w (k) on the focusing surface. Variance E of fluctuation v (k) of AF sensor
{V (k) ² } can be obtained by observing the output signal of the AF sensor for a certain period of time when the focused object is stationary. Put it like this:

[0020]

[Equation 2] E {v (k) ² } = σ _v ² Next, for example, while the wafer 12 as the focusing object is being moved in the run-up section 67A, its movement speed is as much as possible during exposure. The output signal of the AF sensor is observed when the scanning speed is close, and the variance σ _w ' ² of the output signal is obtained. This is because the disturbance w (k) on the surface to be focused is observed as a time series in the output signal of the AF sensor.
Since it is considered that the fluctuation v (k) of the F sensor itself is added, the dispersion of the disturbance w (k) on the focusing surface is E {w
As (k) ² }, the following equation holds.

[0021]

## EQU3 ## E {w (k) ² } = σ _w ' ² −σ _v ² = σ _w ² where σ _w ² is the variance representing the disturbance on the focusing surface. In this case, since v (k) and w (k) are independent of each other, the estimator E {v (k) · of the product of v (k) and w (k) is obtained.
w (k)} is 0. That is, the following equation is established.

[0022]

## EQU00004 ## E {v (k) .w (k)} = 0 Therefore, the stationary Kalman filter is used as the optimum filter for the servo system in FIG. The filtering algorithm is shown below. (1) First, the estimated amount of the true value x (k) of the observed amount of the AF sensor at time k is set to << x
(K) >>, the provisional estimator of the true value x (k) at time (k + 1) is <x (k + 1)>, and the observed value y of the AF sensor is
Assuming that the provisional estimator of (k) is <y (k + 1)>, the following equation holds as the prediction of <x (k + 1)> and <y (k + 1)> from (Equation 1).

[0023]

<X (k + 1)> = << x (k) >>, <y (k + 1)> = <x (k + 1)> (2) Next, the actual observed value y (k + 1) of the AF sensor.
Is used to correct the provisional estimator <x (k + 1)>, and the true value estimator << x of the observed value of the AF sensor at time (k + 1) << x
(K + 1) >> is obtained. That is, the following is performed using a predetermined gain G.

[0024]

## EQU6 ## << x (k + 1) >> = <x (k + 1)> + G {y (k + 1)-
<Y (k + 1)>} Specifically, the process of (Equation 1) is performed at time k, and again at time (k +
By performing the processing of (Equation 2) in 1) in a pipeline manner, filtering is continuously executed, and << x >> can be obtained as a time series of the estimated amount of the true value of the observed value of the AF sensor. Here, the gain G is expressed by the following equation.

[0025]

Equation 7] G = M / (M + σ v 2) variable M in this formula stationary solution of the following algebraic Riccati equation, i.e.,
Variable M (k) when the integer parameter k is set to infinity
Is the limit value of.

[0026]

(8) M (k + 1) = M (k) −M (k) ² / {M
^{_{(K) + σ v 2}}} + σ w 2, M (0) = 0 the variable M from the initial value M (0) to (8) repeated calculations can be determined from the convergence value. From this variable M, the gain G is obtained from (Equation 7), and (Equation 5) and (Equation 6)
The filtering of is executed in real time. In this case, since v (k) · w (k) is a steady process, the calculation of the gain G is performed at the time of measuring σ _w ² and σ _v ² , that is, at the time of the approach run when scanning the focused object. I'll do it. Further, when exposure is performed with the wafer 12 as an object to be focused stationary, it may be treated as σ _w ² = 0. In this case, the above filter is equivalent to a simple averaging filter.

FIG. 8 shows a block diagram of the autofocus and autoleveling function in which the servo system shown in FIG. 11 has the filters obtained as described above, that is, the autofocus and autoleveling mechanism according to the present invention. In FIG. 8, the output signal of the AF sensor 25 is filtered by the filter unit 51 to obtain the optimum estimated amount of the output signal of the AF sensor 25. Then, the optimum estimation amount is subjected to plane approximation by the least-squares approximation calculation unit 52, the two-direction tilt components θ _x and θ _{y of} this approximation plane and the focus position component z are obtained, and the subtraction unit 54 targets the surface of the wafer. Values (θ _xr , θ _yr ,
z _r ) and the value obtained by approximation are passed to the control unit 62, the control signal is calculated by the control unit 62, and the actuator 63 to be controlled is controlled using the control signal obtained by the control signal. are doing.

[0028]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. In this embodiment, the present invention is applied to a step-and-scan type projection exposure apparatus. FIG. 1 shows a projection exposure apparatus according to the present embodiment. In FIG. 1, the illumination light IL for exposure from a light source system 1 including a light source, an optical integrator, and the like is used as a first relay lens 2 and a reticle blind (variable). The slit-shaped illumination area 8 on the lower surface of the reticle 7 is illuminated with a uniform illuminance via the field diaphragm 3, the second relay lens 4, the mirror 5, and the main condenser lens 6. The arrangement surface of the reticle blind 3 is conjugate with the pattern formation surface of the reticle 7, and the shape of the illumination area 8 on the reticle 7 is set by the position and shape of the opening of the reticle blind 3.

The image of the pattern in the illumination area 8 on the reticle 7 is projected and exposed through the projection optical system 11 into the slit-shaped exposure area 13 on the wafer 12 coated with the photoresist. The Z axis is taken in parallel with the optical axis of the projection optical system 11, the X axis is in the direction parallel to the paper surface of FIG. 1 and the Y axis is in the direction perpendicular to the paper surface of FIG. 1 within a two-dimensional plane perpendicular to the optical axis. I take the. The reticle 7 is held on the reticle stage 9, and the reticle stage 9 is driven on the reticle base 10 in the X direction which is the scanning direction by, for example, a linear motor. The X coordinate of the reticle 7 is measured by the moving mirror 18 on the reticle stage 9 and the external laser interferometer 19, and this X coordinate is measured.
The coordinates are supplied to the main control system 20 that controls the operation of the entire apparatus, and the main control system 20 controls the position and the moving speed of the reticle 7 via the reticle stage drive system 21 and the reticle stage 9.

On the other hand, the wafer 12 is held on the focus / leveling stage 14 via a wafer holder (not shown), and the focus / leveling stage 14 is connected to the Y stage via three actuators 16A to 16C movable in the Z direction. 15Y mounted on the Y stage 15Y
Is mounted on the X stage 15X by, for example, a feed screw method so as to be moved in the Y direction, and the X stage 15X is mounted on the apparatus base 17 by, for example, a feed screw method so as to be moved in the X direction. ing. 3 actuators 16
By expanding and contracting A to 16C in parallel, the position (focus position) of the focus / leveling stage 14 in the Z direction is adjusted, and the three actuators 16A to 16C
The tilt angle of the focus / leveling stage 14 is adjusted by individually adjusting the expansion / contraction amount of 16C.

Further, the focus / leveling stage 1
The X-coordinate of the wafer 12 is constantly monitored by the X-axis moving mirror 22X fixed to the upper end of the wafer No. 4 and the external laser interferometer 23X, and as shown in FIG.
Y and the external laser interferometer 23Y, the wafer 12
Is always monitored, and the detected X and Y coordinates are supplied to the coordinate detection unit 59 in the main control system 20.

Here, a configuration example of the actuators 16A to 16C will be described. FIG. 7 shows the actuator 16A.
FIG. 9 is a configuration diagram including a partial cross-sectional view of FIG.
The drive mechanism housing 40 is fixed on the Y stage 15Y of FIG. 1, the feed screw 41 is rotatably accommodated in the drive mechanism housing 40, and the rotary motor 43 is connected to the left end of the feed screw 41 via a coupling 42. A rotary encoder 45 for detecting a rotation angle is connected to the right end of the feed screw 41 via a coupling 44. Further, the nut 39 is screwed to the feed screw 41, and the slope portion 36A having the upper end inclined is fixed to the nut 39 through the support column 38.
The rotating body 36B is in contact with the upper end of the inclined surface portion 36A. The rotator 36B is embedded in the focus / leveling stage 14 shown in FIG. 1 so as to be rotatable and immovable in the lateral direction.

The slope 36A is supported so as to be movable along the linear guide 37 in the X direction parallel to the feed screw 41. In this case, the control signal from the wafer stage control system 24 shown in FIG.
3, the rotary motor 43 rotates the feed screw 41 at the instructed drive speed (angular speed). This allows
The nut 39 moves in the X direction along the feed screw 41, and the slope portion 36A also moves along the feed screw 41. Therefore,
The rotating body 36B contacting the upper end of the inclined surface portion 36A is displaced in the vertical direction (Z direction) with respect to the drive mechanism housing 40 while rotating. Further, by measuring the rotational angular velocity of the feed screw 43 by the rotary encoder 45, the moving velocity of the rotating body 36B in the vertical direction is detected. The other actuators 16B and 16C have the same configuration.

The actuators 16A to 16C are
In addition to the method using a rotary motor as shown in Fig. 7,
For example, it may be composed of a piezo element or the like. Returning to FIG. 1,
The main control system 20 controls the operations of the X stage 15X, the Y stage 15Y, and the focus / leveling stage 14 via the wafer stage drive system 24 based on the supplied coordinates. For example, when performing exposure by a scanning exposure method, it is assumed that the projection optical system 11 projects an inverted image at a projection magnification β, and the reticle 7 is moved through the reticle stage 9 in the + X direction (or −X direction) with respect to the illumination area 8. Direction) speed V _R
In synchronism with the scanning in step S1, the wafer 12 is scanned through the X stage 15X in the −X direction (or + X direction) at a speed v (= β · V _R ).

Next, the structure of a multipoint focus position detection system (hereinafter referred to as "multipoint AF sensor") 25 for detecting the position (focus position) of the surface of the wafer 12 in the Z direction will be described. In this multi-point AF sensor 25, detection light that is non-photosensitive to the photoresist emitted from the light source 26 illuminates a large number of slits in the light-sending slit plate 28 via the condenser lens 27, and The image is transmitted through the light-transmitting-side objective lens 29 to the projection optical system 11
Of the exposure area 13 on the wafer 12 and the pre-reading areas 35A and 35B (see FIG. 2) before and after the exposure area 13 obliquely to the optical axis of
It is projected on the five measurement points P _{11 to} P ₅₁ .

FIG. 2 shows those measurement points P ₁₁ on the wafer 12.
2 to P _{51. In} FIG. 2, preread areas 35A and 35B are set on the + X direction side and the −X direction side of the slit-shaped exposure area 13, respectively. Then, in the exposure area 13, measurement points P _{21 to} P ₄₃ of 3 rows × 3 columns are arranged.
Is set, and the three measurement points P ₁₁ to
P ₁₃ is set, and three measurement points P are set in the prefetch area 35A.
_{51 to} P ₅₃ are set. In this embodiment, the exposure area 1
The average focus position in the exposure area 13 and the tilt angle are obtained from the information of the focus positions at the nine measurement points in 3. Further, when the scanning direction of the wafer 12 is the −X direction, the level difference is corrected by using the focus position information at the three measurement points in the prefetch area 35A, and the wafer 12 is scanned as described later. When the direction is the + X direction, the step difference is corrected using the information on the focus positions at the three measurement points in the prefetch area 35B.

Returning to FIG. 1, the reflected light from these measurement points passes through the light-receiving side objective lens 30 and the vibrating slit plate 31.
The slit images which are focused on and projected on the measurement points on the vibrating slit plate 31 are re-imaged. The vibrating slit plate 31 is vibrating in a predetermined direction by the vibrator 32 driven by the drive signal DS from the main control system 20. The light passing through the large number of slits of the vibration slit plate 31 is photoelectrically converted by the large number of photoelectric conversion elements on the photoelectric detector 33, and these photoelectric conversion signals are supplied to the signal processing system 34.

FIG. 3 shows the light-sending slit plate 28 shown in FIG. 1. In FIG. 3, the light-sending slit plate 28 is located at positions corresponding to the measurement points P _{11 to} P ₅₃ on the wafer shown in FIG. Slits 28 _{11 to} 28 ₅₃ are formed. Also, FIG.
Also on the vibrating slit plate 31 inside, as shown in FIG.
Slits 31 _{11 to} 31 ₅₃ are formed on the wafer at positions corresponding to the measurement points P _{11 to} P ₅₃ , respectively, and the vibrating slit plate 31 is vibrated by the vibrator 32 in the measurement direction orthogonal to the longitudinal direction of each slit. ing.

Next, FIG. 5 shows the photoelectric detector 33 in FIG.
5 shows a signal processing system 34. In FIG. 5, the photoelectric conversion elements 33 _{11 to} 33 ₁₃ in the first row on the photoelectric detector 33 are reflected from the measurement points P _{11 to} P _{13 in} FIG.
And the light is incident having passed through the corresponding slits in the vibrating slit plate 31, 2 to fourth rows of the photoelectric conversion elements 33 ₂₁ to
Lights reflected from the measurement points P _{21 to} P _{43 in} FIG. 2 and passing through the corresponding slits in the vibration slit plate 31 are incident on the reference numeral 33 ₄₃ , and the photoelectric conversion elements 33 ₅₁ to
Lights reflected from the measurement points P _{51 to} P _{53 in} FIG. 2 and passed through the corresponding slits in the vibration slit plate 31 are incident on the reference numeral 33 ₅₃ . Then, the detection signal from the photoelectric conversion element 33 _{21-33 43} 2 fourth rows are amplifier 4
It is supplied to the synchronous rectifiers 47A to 47I via 6A to 46I. The synchronous rectifiers 47A to 47I are the vibrators 3 respectively.
By synchronously rectifying the input detection signal using the drive signal DS for 2, the focus signal that is substantially proportional to the focus position of the corresponding measurement point in the predetermined range is generated.
In the present embodiment, the focus signals output from the synchronous rectifiers 47A to 47I have the corresponding measurement points in FIG. 1 at the image forming plane (best focus plane) of the projection optical system 11.
The calibration is performed so that the value becomes 0 when it matches.

The focus signals output from the synchronous rectifiers 47A to 47I are analog / digital (A /
D) Main converter 20 of FIG. 6 via converters 49A-49I
Is supplied to. Further, the first row of the photoelectric conversion element 33 _21-3
Detection signals from the three _43, three sets of amplifiers are fed to the partial processing system 48A including a synchronous rectifier, and an A / D converter, 5
The detection signals from the photoelectric conversion elements 33 _{51 to} 33 _{53 in the} row are
It is supplied to a partial processing system 48B including three sets of amplifiers, a synchronous rectifier, and an A / D converter. These partial processing systems 4
Focus signals from 8A and 48B are also supplied to the main control system 20.

FIG. 6 shows the three actuators 16 of FIG.
6 shows drive systems A to 16C. In FIG. 6, the focus signals output from the A / D converters 49A to 49I in the signal processing system 34 are supplied to the filter units 51A to 51I in the main control system 20, respectively. , Filter units 51A to 51
The focus signals input by I are subjected to the filtering of (Equation 5) and (Equation 6), and the true values of the focus signals are estimated from the filter units 51A to 51I to the least-squares approximation calculation unit 52. The quantity << x >> is sequentially output in near real time.

A specific example of the filtering process in the filter unit 51I will be described. In this case, (Equation 6)
It is assumed that the optimum value of the gain G used in step 1 is previously obtained and set in the filter unit 51I. Furthermore, A / D
Time is standardized by the sampling cycle of the digital focus signal output from the converter 49I, and time 0, ...,
At k and k + 1, the focus signal output from the A / D converter 49I, that is, the actual observation amount is y
(0), ..., Y (k), y (k + 1). Further, the true values of the focus signals are respectively set to x (0), ..., x
(K) and x (k + 1), the filter unit 5 of the present embodiment
In 1I, the estimated value of the true value << x (0) >>, ..., << x
(K) >>, << x (k + 1) >> are sequentially obtained by the following procedure.

First, the estimated value << x (0) >> of the true value of the focus signal at time 0 is approximated by, for example, the observed value y (0), and from (Equation 5), the true value of the focus signal at time 1 is obtained. x
The temporary estimated amount <x (1)> of (1) and the temporary estimated amount <y (1)> of the observed amount of the focus signal are calculated. In this case, it becomes as follows.

[0044]

[Formula 9] <y (1)> = <x (1)> = y (0) After that, the provisional estimated amount <x (1)>, <y (1)>, the observed amount y (1) at time 1 ), And the gain G (Equation 6)
Therefore, the estimated amount of the true value of the focus signal at time 1 << x
(1) >> is calculated. In this case, it becomes as follows.

[0045]

[Equation 10] << x (1) >> = <x (1)> + G {y (1)
-<Y (1)>} Hereinafter, by repeatedly applying (Equation 5) and (Equation 6), the estimated value <x of the true value of the focus signal is sequentially obtained.
(2) >>, ..., << x (k) >>. Next, (Equation 5)
At time (k + 1), the provisional estimator <x (k + 1)> of the true value of the focus signal and the provisional estimator <y of the observed value <y
After calculating (k + 1)>, the time (k + 1) is calculated from (Equation 6).
Then, the estimated value << x (k + 1) >> of the true value of the focus signal at is obtained.

Next, a method of calculating the gain G set in the filter section 51I will be described. In this embodiment, the focus signal y (k) output from the A / D converter 49A is also supplied to the correction amount calculation unit 53. First, the wafer 12
The correction amount calculation unit 53 calculates the variance σ _v ² of the focus signal y (k) by observing the focus signal y (k) for a predetermined time while is stationary. afterwards,
When the wafer 12 is moving in the scanning direction (X direction) at a speed substantially close to the scanning speed during scanning exposure, the correction amount calculation unit 5
3 observes the focus signal y (k) for a predetermined time to calculate the variance σ _w ' ² of the focus signal y (k). After that, the correction amount calculation unit 53 is calculated based on (Equation 3).
Calculates the variance σ _w ² representing the disturbance on the focusing surface.

Then, the correction amount calculation unit 53 determines the variable M as the convergence value when the integer parameter k is increased in (Equation 8), and uses this variable M and the variance σ _v ² (Equation 7). A more optimal gain G is determined. Similarly, the correction amount calculation unit 53 independently obtains an optimum gain for each of the filter units 51A to 51H and sets the obtained gain. Note that, for example, the gain obtained for one filter unit 51I may be set for the other filter units 51A to 51H.

Next, in the least-squares approximation calculation unit 52, as shown in FIG.
Measurement points P in the exposure area 13 of _{twenty one}~ P₄₃Corresponding to
Based on the estimated value of the true value of nine focus signals << x (k) >>
Then, the exposure area 13 is approximated by the method of least squares.
Z plane at the center of the determined plane
Inclination angle θ in z, XZ plane_x, And tilt in the YZ plane
Angle θ_yAnd the Z coordinate z and the tilt angle θ_x, And lean
Angle θ_yTo subtraction units 54A, 54C, and 54B, respectively.
Supply to.

Further, the subtraction units 54A, 54C and 54B
From the target value setting unit (not shown) to the projection optical system 1 respectively.
Z coordinate z at the center of the image plane of 1_r, Inclination in the XZ plane
Angle θ _xr, And the tilt angle Δθ in the YZ plane_yr, Ie wafer
Target values for the surface position of are also provided. Subtraction unit 54A,
From 54C and 54B to the control unit 55,
Subtract the approximate plane value from the target value as shown in the following equation.
Obtained Z coordinate deviation Δz, inclination angle deviation Δθ_x, And tilt
Angular deviation Δθ_yIs supplied.

[0050]

Δθ _x = θ _xr −θ _x , Δθ _y = θ _yr −θ _y , Δz = z _r −z Next, in the control unit 55, these Z coordinate deviations Δz,
Tilt angle deviation [Delta] [theta] _x, and the inclination angle deviation [Delta] [theta] _y respectively suitable gain k _z, k _x and k _y is multiplied. And
The velocity Δ in the Z direction from the control unit 55 to the correction unit 56
Control signals corresponding to V _z , angular velocity ΔVθ _x in the XZ plane, and angular velocity ΔVθ _y in the YZ plane are output.
In addition, the correction amount calculation unit 53 receives from the focus signals from the partial processing systems 48A and 48B in the signal processing system 34, the Z coordinate at the center of the approximation plane from the least square approximation calculation unit 52, and the coordinate detection unit 59. The information such as the scanning speed of the X stage 15X is also supplied, and the correction amount calculation unit 53 calculates the correction value such as the speed v _{z in the} Z direction based on the information of the focus position in the prefetch area 35A or 35B in FIG. The correction value obtained is supplied to the correction unit 56. The partial processing systems 48A, 48
Also for the focus signal from B, the filter unit 51A
Optimal filtering may be performed by providing a filter unit such as ~ 51I.

Then, the velocity ΔV _z obtained by adding the correction value from the correction amount calculation unit 53 to the input value from the correction unit 56 to the coordinate conversion unit 58 (the same display as before correction is used, and so on. ), Angular velocity ΔVθ _x , and angular velocity ΔVθ _y
Is supplied. The coordinate conversion unit 58 includes three actuators 16A and 16A at the time k from the coordinate detection unit 59.
(X, Y) coordinates of B and 16C (X
₁ , ₁ , Y ₁ ), (X ₂ , Y ₂ ), and (X ₃ , Y ₃ ) are supplied. These (X, Y) coordinates are the X
It is a value that changes as it moves in the direction. Then, the coordinate conversion unit 58 causes the actuators 16A and 16A to obtain the angular velocity ΔVθ _x , the angular velocity ΔVθ _y , and the velocity ΔV _z.
Driving speeds Δv ₁ and Δ of B and 16C in the Z direction
v ₂ and Δv ₃ are calculated by the following equations. The correction amount from the correction amount calculation unit 53 is ignored.

[0052]

(Equation 12)

The coordinate conversion 58 determines the driving speeds Δv ₁ and Δ.
The control signals corresponding to v ₂ and Δv ₃ are supplied to the velocity loop controller 60. Speed loop controller 60
Drives the actuators 16A to 16C via the power amplifiers 61A to 61C. Also, the actuator 1
6A to 16C are provided with speed sensors 45A to 45C (same as the rotary encoder 45 of FIG. 7) for detecting the moving speed, and speed detection signals from these speed sensors 45A to 45C are sent to the speed loop controller 60. Feedback has been given. As a result, the actuators 16A to 16C each have a driving speed Δv ₁ to
It is driven in the Z direction by Δv ₃ . Then, the position and tilt angle of the surface of the wafer 12 after being driven by the actuators 16A to 16C are measured by the multipoint AF sensor 25 in FIG. 1, and the measurement result is fed back to the control system in FIG. The servo system of FIG. 6 operates so that the drive speeds Δv _{1 to} Δv ₃ finally calculated by the coordinate conversion unit 58 become 0, respectively.

FIG. 8 shows a simplified control system of FIG.
In FIG. 8, N measured by the multipoint AF sensor 25
6 (N is 9 in FIG. 2) focus signals are supplied to the filter units 51 corresponding to the filter units 51A to 51I in FIG. 6, and the filter units 51 perform filtering with optimum gains. The estimated values of the true values of the N focus signals output from the filter unit 51 are supplied to the least squares approximation calculation unit 52, and the least squares approximation calculation unit 52 supplies the wafer input to the first input unit of the subtraction unit 54. The focus position z of the surface approximated to the focusing target surface, the tilt angle θ _x in the XZ plane, and the tilt angle θ _y in the YZ plane are supplied, and the second input unit of the subtraction unit 54 receives the Z coordinate The target value z _r and the tilt angle target values θ _xr and θ _yr are supplied.

Then, the deviation output from the subtraction section 54 is supplied to the control section 62, and the control signal obtained by multiplying the deviation by a predetermined gain in the control section 62 corresponds to the actuators 16A to 16C in FIG. Actuator 63
Is being driven. In this case, according to the present embodiment, the dispersion of the focus signal output from the multipoint AF sensor 25 is measured in the stationary state and the moving state of the wafer 12, and the gain in the filter unit 51 is optimized based on these dispersions. There is. Therefore, the fluctuation of the AF sensor 25 and the influence of the disturbance caused by the movement of the focusing surface are reduced, and the projection optics of the surface of the wafer 12 when the autofocus and the autoleveling are applied when the exposure is performed by the scanning exposure method. The tracking accuracy with respect to the image plane of the system 11 is improved. Therefore, the entire pattern of the reticle 7 is exposed on the wafer 12 with high resolution.

Next, when exposing a large number of shot areas on the wafer 12 by the scanning exposure method, as shown in FIG.
An example of the timing for calculating the optimum gain for the inner filter units 51A to 51I will be described with reference to FIGS. 9 and 10. 9 shows an arrangement of shot areas on the wafer 12. In FIG. 9, a large number of shot areas 64A, 64B, 64C, ... Are arranged on the wafer 12 at predetermined intervals in the X direction (scanning direction) and the Y direction. The circuit pattern images of the reticle 7 are exposed on these shot areas by the scanning exposure method.

First, when the first shot area 64A on the wafer 12 is exposed, the slit-shaped exposure area 13 is formed with the illumination light IL shielded by a variable blind (not shown) in FIG. On the other hand, the shot area 64A is accelerated in the + X direction, and the wafer 12 reaches a predetermined scanning speed within the run-up section 67A. Thereafter, the reticle 7 is irradiated with the illumination light IL shown in FIG. 1, and when the wafer 12 is used as a reference, the exposure region 13 gradually moves from the position 66A to the position 6 along the locus 65A.
6B, the pattern image of the reticle 7 is sequentially exposed on the shot area 64A. Also, the shot area 64A
In a section 67B between the position 66B and the position 66B (this is called a "deceleration section"), the illumination light IL is blocked and the wafer 12 is gradually decelerated.

After that, when the Y stage 15Y of FIG. 1 is stepped in the Y direction, the slit-shaped exposure area 13 is moved from the position 66B to the second shot area 6 as shown in FIG.
Move to position 66C before 4B. Next, the shot area 64B is accelerated in the −X direction with respect to the slit-shaped exposure area 13, and the wafer 12 reaches a predetermined scanning speed within the run-up section 67C. Then, with the wafer 12 as a reference,
The exposure region 13 gradually moves from the position 66D to the outside of the shot region 64B along the locus 65B, and the pattern image of the reticle 7 is sequentially exposed on the shot region 64B.

[0059] Figure 10, in this case shows how the change in the scanning speed V _R of the reticle 7 in, 10, the acceleration period T1, the scanning speed V _R corresponding to the acceleration region 67A in FIG. 9
Gradually increases and reaches a predetermined value, and the scanning speed V _R is maintained at the predetermined value in the subsequent exposure period T2. Further, in the deceleration period T3 corresponding to the deceleration section 67B and the approach section 67C in FIG. 9, the scanning speed V _R of the reticle 7 once becomes 0 and then is accelerated in the negative direction to reach the predetermined value. In the subsequent exposure period T4, the scanning speed V _R is maintained at the predetermined value, and in the subsequent period T5, the scanning speed V _R gradually becomes 0 again. Further, in FIG. 9, the scanning direction of the wafer 12 is determined so that, for example, the scanning direction of the reticle 7 is sequentially reversed when the exposure is performed on the third shot area 32C and thereafter. Then, the change in the moving speed of the wafer 12 in the X direction also has a tendency similar to that in FIG.

In this case, the correction amount calculation unit of this embodiment shown in FIG.
In 53, when the wafer 12 is stationary in advance, A /
Focus signal output from D converters 49A to 49I
Variance σ_v ²Ask for. After that, the correction amount calculation unit 53
Is the A / D converter 49 in the latter half of the run-up section 67A of FIG.
Variance σ of the focus signal output from A to 49I_w' ² 
And (Equation 7) and (Equation 8) based on these results.
A more optimal gain G is determined, and this determined gain G is
The filter units 51A to 51I are set. This allows
When exposing to the hot area 64A, the immediately preceding run-up section 6
Filtering is performed based on the gain determined in 7A.
Will be

Similarly, the gains in the filter portions 51A to 51I during the exposure for the second shot area 64B are set based on the dispersion measured in the latter half of the preceding run-up section 67C. As a result, autofocusing and autoleveling are performed with high accuracy even when the unevenness in each shot area or the scanning speed changes. In addition, since the variance σ _v ² is first obtained and the time for measuring the variance σ _w ' ² is not specially set thereafter, the throughput of the exposure process does not decrease. However, for example, the variance of the focus signal may be measured in the first half of the deceleration section 67B of the first shot area 64A, and the gain in the second shot area 64B may be set based on the measurement result.

Although the measurement points P _{21 to} P ₄₃ for tilt detection are distributed in the exposure area 13 in FIG. 2, even if the measurement points P _{21 to} P ₄₃ are out of the exposure area 13. Good. Further, the number and arrangement of the entire measurement points P _{11 to} P ₅₃ are not limited to those in FIG. 2, and the measurement points may be arranged in different steps in the X direction, for example. Further, in the above embodiment, the wafer 12
The multi-point AF sensor 25 is used to detect the tilt angle of the upper exposure area 13. Instead, for example, the surface of the wafer 12 is obliquely irradiated with a parallel light beam and the converging position of the reflected light is changed. A parallel light beam oblique incidence type leveling sensor that detects the inclination angle of the surface from the lateral shift amount may be used.

The present invention is not limited to the above-mentioned embodiments, and it goes without saying that various configurations can be made without departing from the gist of the present invention.

[0064]

According to the scanning exposure method of the present invention, the filter is based on the variation of the detection signal from the focus position detection means when the substrate as the focusing object is stationary and when the substrate is being scanned. The gain in the instrument is optimized. Therefore, it is not necessary to search for the cutoff frequency of the low-pass filter by trial and error, and even if the surface condition or scanning speed of the focused object changes, it is automatically folded into the characteristics of the filter means. There is an advantage that the surface of the focused object can always be accurately followed by the target surface by autofocus or autoleveling.

Further, when the variation of the detection signal is measured during the run-up period, which is essentially a dead time, no special measurement sequence is provided, so that the throughput of the exposure process is not lowered. There is. Furthermore, according to the scanning exposure apparatus of the present invention, the scanning exposure method can be implemented.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing a scanning exposure type projection exposure apparatus according to an embodiment of the present invention.

FIG. 2 is a plan view showing a distribution of measurement points of focus positions on the wafer 12 of FIG.

FIG. 3 is a diagram showing a light-sending slit plate 28 in FIG.

FIG. 4 is a diagram showing a vibrating slit plate 31 in FIG.

5 is a photoelectric detector 33 and a signal processing system 34 in FIG.
FIG.

6 is a configuration diagram including a partial perspective view showing a focus / leveling mechanism of a wafer 12 in FIG. 1 and a control system thereof.

FIG. 7 is a partial cross-sectional configuration diagram showing a configuration example of an actuator 16A in FIG.

8 is a block diagram showing a simplified control system of the focus / leveling mechanism of FIG.

9 is a partially enlarged plan view showing an example of a shot array on the wafer 12 of FIG.

FIG. 10 is a diagram showing changes in the scanning speed of the reticle when performing exposure by the scanning exposure method.

FIG. 11 is a block diagram showing a conventional auto focus and auto leveling mechanism.

FIG. 12A is a diagram showing an example of an AF sensor, and FIG.
FIG. 4 is a diagram showing a case where the surface of the wafer is uneven in the scanning exposure apparatus.

[Explanation of symbols]

7 reticle 9 reticle stage 11 projection optical system 12 wafer 13 exposure area 14 focus / leveling stage 15Y Y stage 15X X stage 16A to 16C actuator 19, 23X, 23Y laser interferometer 20 main control system 25 multipoint AF sensor P _{11 to} P ₅₃ measurement point 33 photoelectric detector 34 signal processing system 35A, 35B look-ahead region 51A to 51I filter unit 52 least-squares approximation calculation unit 53 correction amount calculation unit 55 control unit 58 coordinate conversion unit 59 coordinate detection unit 60 speed loop controller 63 actuator

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication H01L 21/30 516 A 525 H 526 A

Claims (3)

[Claims] 1. A mask on which a transfer pattern is formed is illuminated with exposure illumination light, the pattern of the mask is projected onto a photosensitive substrate through a projection optical system, and the mask is projected in a predetermined direction. In the scanning exposure method of sequentially exposing the pattern of the mask on the substrate by scanning the substrate in a predetermined direction in synchronism with the scanning, the optical axis of the projection optical system on the surface of the substrate. A focus position detecting means for outputting a detection signal according to a position in the direction, a filter means for increasing / decreasing the detection signal output from the focus position detecting means with a predetermined gain, and a detection signal output from the filter means. Height adjusting means for adjusting the position of the substrate in the optical axis direction of the projection optical system, and a variation in the detection signal from the focus position detecting means when the substrate is stationary, and Scanning exposure method whose serial board and obtains the predetermined gain in the filter means on the basis of the variation of the detection signal from the focus position detecting means when being scanned. 2. A scan for scanning the mask and the substrate in synchronization with the variation of the detection signal from the focus position detecting means being obtained in advance as the first variation when the substrate is stationary. During each run-up period from the start to the start of exposure, the variation of the detection signal from the focus position detection means is obtained as the second variation, and is determined based on the first variation and the second variation. 2. The scanning exposure method according to claim 1, wherein the filter means is operated in an exposure period following the run-up period by using the predetermined gain. 3. An illumination area of a predetermined shape on a mask on which a transfer pattern is formed is illuminated with exposure illumination light, and the pattern of the mask in the illumination area of the predetermined shape is projected through a projection optical system. The substrate is moved in a predetermined direction through the substrate stage in synchronism with projection on a photosensitive substrate on the substrate stage and scanning of the mask in a predetermined direction with respect to the illumination region of the predetermined shape. In a scanning exposure apparatus that sequentially exposes the pattern of the mask on the substrate by scanning, a focus position detection unit that outputs a detection signal corresponding to the position of the surface of the substrate in the optical axis direction of the projection optical system, Filter means for increasing / decreasing a detection signal output from the focus position detection means with a predetermined gain, and front of the substrate with respect to the substrate stage based on the detection signal output from the filter means. Height adjusting means for adjusting the position of the projection optical system in the optical axis direction, dispersion of detection signals from the focus position detecting means when the substrate is stationary, and when the substrate is being scanned A scanning exposure apparatus comprising: a calculation unit that obtains the predetermined gain in the filter unit based on a variation in a detection signal from the focus position detection unit.