JPH0897123A - Alignment method - Google Patents

Abstract

PURPOSE: To lessen the influence of jump shot when performing alignment by EGA method. CONSTITUTION: The coordinates of the arrangement of n (the initial value of n is N) sample shots are measured, and the result of measurement is processed to find the value of three times the standard deviation of nonlinear error component, and this is divided by a specified function to compute an evaluation value An . As to (n-1) sample shots being left over after removal of the sample shot the largest in nonlinear error component out of these sample shots, the evaluation value An-1 is found from the three times the standard deviation of the nonlinear error component, hereafter the sample shots larger in nonlinear error components are removed in order to compute evaluation values An-2 , An-3 ,.... The sample shot the largest in nonlinear error component within the range where the evaluation value An is larger than the average value <An > of these evaluation values is removed as a jump shot.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is used, for example, in manufacturing a semiconductor device or a liquid crystal display device by a photolithography process, and a wafer (or a reticle (or photomask, etc.) pattern coated with a photosensitive material (or It is suitable for use in an exposure apparatus such as a stepper that transfers and exposes each shot area on a glass plate or the like) when sequentially aligning each shot area on a wafer with an exposure position based on array coordinates calculated by statistical processing. Regarding the alignment method.

[0002]

2. Description of the Related Art For example, since a semiconductor element is formed by exposing a plurality of layers of circuit patterns on a wafer and exposing the same, when projecting and exposing the circuit patterns of the second and subsequent layers onto the wafer, It is necessary to accurately align each shot area where a circuit pattern is already formed with the pattern of the reticle, that is, align the wafer and the reticle. Wafer alignment in an exposure apparatus such as a conventional stepper is performed by the following enhanced global alignment (hereinafter referred to as “EG
A method) (for example, JP-A-61-161)
44429).

That is, a plurality of shot areas (chip patterns) each provided with a positioning mark called a wafer mark are formed on the wafer, and these shot areas are set in advance on the wafer. They are regularly arranged based on the arrangement coordinates. However,
Even if the wafer is stepped based on the designed array coordinate values (shot arrays) of a plurality of shot areas on the wafer, the wafer is not always accurately aligned due to the following factors.

(1) Remaining rotation error of wafer (rotation θ) (2) Orthogonality error of stage coordinate system (or shot array) w (3) Linear expansion and contraction of wafer (scaling Rx, Ry) (4) Wafer (center Position) offset (translation) O
x, Oy

At this time, the coordinate transformation of the wafer based on these four error amounts (six parameters) can be described by a linear transformation equation. Therefore, for a wafer in which a plurality of shot areas with wafer marks are regularly arranged, the coordinate system (x, y) on the wafer is used as a stationary coordinate system and the coordinate system (X, Y) on the stage is used. The first-order transformation model to be transformed into can be expressed as follows using those six parameters. However, the approximation is performed assuming that the absolute values of the angles θ and w are small.

[0006]

[Equation 1]

Six parameters R in this conversion formula
x, Ry, θ, w, Ox, Oy are as follows in EGA
It can be determined by the method. In this case, a plurality of shot areas (chip patterns) to be exposed on the wafer
Some singled out shot area (hereinafter, "sample shots" hereinafter) from the each concomitant coordinate system (x, y) on the coordinate on the design, each (x _1,
The wafer marks y ₁ ), (x ₂ , y ₂ ), ..., (x _N , y _N ) are aligned with a predetermined reference position. Then, the actual coordinate values (XM ₁ , YM ₁ ) of the wafer mark at that time in the coordinate system (X, Y) on the stage, (XM ₂ , YM ₂ ) ,.
M _N , YM _N ) is measured.

Also, in designing the selected wafer mark.
Array coordinates of (x_i, Y_i) (I = 1, ..., N)
Calculated array coordinates obtained by substituting into the linear transformation model of
(X _i, Y_i) And the coordinates (X
M_i, YM_i) And the difference (△ x_i, △ y_i)
Think of it as an error. And then align as
The error sum of squares is regarded as the residual error component, and this residual error
Set the values of those 6 parameters to minimize the minutes
Meru.

[0009]

[Equation 2]

Specifically, the residual error component is partially differentiated sequentially with 6 parameters, an equation is set so that the value becomes 0, and if these 6 simultaneous equations are solved, 6 parameters of 6 parameters are obtained. Value is required. The calculation for obtaining the six conversion parameters of (Equation 1) by the method of least squares is called "EGA calculation". After this, the alignment of each shot area of the wafer can be performed based on the array coordinates calculated using the linear transformation equation (Equation 1) using those parameters as coefficients.

For example, FIG. 11A shows eight sample shots SA _{1 to} SA ₈ on the wafer and alignment error vectors V _{1 to} V ₈ in these sample shots.
The respective vectors V _{1 to} V ₈ are vectors obtained by subtracting design coordinates from the measured coordinates. Also, FIG. 11B shows the number of sample shots measured, the average value of the X component and the Y component of the alignment error, and three times the standard deviation of the X component and the Y component (3.
σ), the error from 1 of the scaling Rx and Ry representing the linear error, the orthogonality error w, and the value of the rotation θ.

Further, FIG. 11 (c) shows those parameters R
The value of 3 times (3σ) the standard deviation of the linear error in the X direction and the Y direction obtained for each sample shot from x, Ry, w, θ, Ox, Oy is shown. 3 times (3σ) the standard deviation of the non-linear error components in the X and Y directions. The linear error in FIG. 11C is the coordinate (X _i , Y _i ) calculated from (Formula 1) to the coordinate (x
_i , y _i ) is the error vector obtained by subtracting
The non-linear error in FIG. 11D is calculated coordinates (X _i , Y _i ) calculated from (Formula 1) from coordinates (XM _i , YM _i ) actually measured on the stage coordinate system. It is an error vector obtained by subtraction, that is, an error vector obtained by subtracting a linear error from an alignment error.

In this case, the linear error vectors VL _{1 to} VL ₈ in the sample shot of FIG. 11 are as shown in FIG. That is, when the alignment error vector does not include one having a larger absolute value than other error vectors as shown in FIG. 11, an accurate linear error vector is obtained.

[0014]

DISCLOSURE OF THE INVENTION Problems to be Solved by the Invention
In the A-type alignment method, the plurality of sample shots may include a so-called “jump shot” in which the alignment error is particularly large compared to other sample shots. Since such a jump shot is caused by a measurement error caused by a collapse of a wafer mark attached to the sample shot on the wafer, or a local non-linear distortion caused by a foreign substance on the back surface of the wafer, When calculating the array coordinates of other shot areas, it is desirable to exclude such jump shots.

For example, FIG. 13A shows eight sample shots SA ₁ -S including jump shots on the wafer.
A ₈ and other examples of the alignment error vectors V _{1 to} V ₈ in these sample shots are shown in FIG.
12A to 12D respectively correspond to FIGS. 12A to 12D. Further, in FIG. 13A, the sample shot SA ₃ is a jump shot, and the absolute value of the vector V ₃ of the alignment error in this jump shot is particularly large. In this case, from the calculated coordinates (X _i , Y _i ) calculated from (Equation 1) to the designed coordinates (x _i , y)
The vector of the linear error obtained by subtracting _i ) is
Vector VL _{1 to} VL ₈ of FIG. 13, and the error vector V _{3 of} FIG. 13 becomes two linear error vectors VL ₃ and VL.
Looks like it's mutated to ₆ . This means that if the EGA calculation is performed with the jump shot included, the linear error in the position of the shot area other than the jump shot increases.

When actually performing the alignment,
Depending on the type of wafer to be processed, the operator needs to input the number and position of sample shots into the control unit of the alignment system. In this case, if the number of sample shots is small, the averaging effect is small, but conversely, if the number of sample shots is large, the measurement time becomes long, and as a result, the throughput (productivity) of the exposure process is reduced. The issue is how many sample shots are set.

In view of the above point, the present invention, when performing alignment by applying the EGA method, accurately determines each shot region (processed region) of a wafer (substrate) without being affected by a jump shot. It is an object of the present invention to provide a positioning method capable of setting the exposure position of the. The present invention further comprises
An object of the present invention is to provide a positioning method capable of automatically determining an evaluation standard for removing a jump shot.

[0018]

A first position according to the present invention
The alignment method is a plurality of workpieces set on the substrate (4).
Area (ES₁, ES₂, ..., ES_M) Each of its substrate
Predetermined machining position in the coordinate system (X, Y) that defines the movement position
When aligning the
Within the work area, a predetermined number of preselected sample areas
Area (SA₁, SA ₂, ...) on the coordinate system (X, Y)
A plurality of coordinate positions measured in this way by measuring the gauge position
By statistically calculating
The array coordinates on each coordinate system (X, Y) of the work area
Substrate according to the array coordinates calculated in this way
By controlling the moving position in (4)
Align each of the machined areas with the machining position
On how to do.

[0019] Then, the present invention is, among the plurality of the processing region, the preselected N (N is an integer of 4 or more) coordinate system of the sample area _{(SA 1 ~SA N) (X} , Y) The first step of measuring the coordinate position on the above, and the non-linear error component from the design position of the coordinate position measured in the first step for each of the N sample areas, and the non-linear error components are calculated. Variation in components (NLE in Fig. 7
Divided by rated value (3 [sigma])) the function of the N (F _n and G _n) (second step and determining the A _n) of FIG. 10; to the number of sequential the sample area is a predetermined number, the first A third step of obtaining an evaluation value (A _n ) corresponding to the number of remaining sample areas by removing the sample area having the largest non-linear error component in the two steps and repeating the second step, and the third step A fourth step of removing a sample area having the largest non-linear error component in the second step within a range in which the individual evaluation values obtained in the second step and the third step are larger than a predetermined threshold value based on the plurality of evaluation values; , The coordinate position measured in the first step is statistically processed for the sample area left in the fourth step and the coordinate system (X, Y) of each of the plurality of processed areas on the substrate (4) is processed. Calculate array coordinates in A step, and has a.

[0020] In this case, an example of the predetermined threshold, the average value of the plurality of evaluation values (<A _n> in FIG. 10), or the average value of the evaluation value <A _n> of the plurality of evaluation values It is a value obtained by adding variation. In addition, the second alignment method according to the present invention uses the same number of preselected N (N is an integer of 4 or more) preselected among the plurality of regions to be processed in the same premise as the first alignment method. Sample area (SA ₁
~ SA _N ) from the designed position of the coordinate position measured in the first step and the first step of measuring the coordinate position on the coordinate system (X, Y) of each of the N sample areas. Of the non-linear error components and the variation (NLE (3σ) in FIG. 7) of these plural non-linear error components, and sequentially in the second step until the number of sample areas reaches a predetermined number. A third non-linear error component variation (NLE (3σ)) corresponding to the number of remaining sample regions is obtained by excluding the sample region having the largest non-linear error component and repeating the second step.
A predetermined function (C _{n in} FIG. 7) in which the steps and the variations of the individual nonlinear error components obtained in the second step and the third step monotonically increase according to the number of sample areas to be calculated In a larger range, a statistical process is performed on the fourth step of removing the sample area having the largest non-linear error component in the second step and the coordinate position measured in the first step for the sample area remaining in the fourth step. And a fifth step of calculating array coordinates on the coordinate system (X, Y) of each of the plurality of processed regions on the substrate.

In this case, assuming that the number of sample areas to be calculated is n, an example of the predetermined function is a function obtained by multiplying a predetermined constant by {n (n-3)} ^1/2 . In addition, the third alignment method of the present invention uses the same preconditions as the first alignment method described above, and preselects N (N is an integer of 4 or more) of the plurality of regions to be processed.
Detects the condition of the surface of the substrate (4) in the sample area of
Based on the detection result, the first step of excluding the sample area whose surface condition is determined to be relatively poor, and the coordinate position of the sample area left in this first step on the coordinate system (X, Y) are calculated. The second step of measurement and the coordinate positions measured in the second step of the sample area remaining in the first step are statistically processed to coordinate system of each of the plurality of processed areas on the substrate (4) ( Third, calculating array coordinates on (X, Y)
And a process.

[0022]

According to the first alignment method of the present invention, after the coordinate positions of the N sample areas are measured, the nonlinear error component of the alignment error of the coordinate positions is first obtained for the N sample areas. . Then, for example, three times the standard deviation (NLE (3σ) in FIG. 7) is obtained as the variation of these N nonlinear error components. The maximum value of these non-linear components may be used instead of three times the standard deviation.

In this case, assuming that the number of sample areas to be measured is generally n, the probability of the nonlinear error being regarded as a linear error increases as the value of n decreases, and therefore the reliability of the linear error is determined by the function F of FIG. _n (F ₀ · (n−
3) As represented by ^1/2 ), it becomes smaller as the value of n becomes smaller. In other words, as the value of n becomes smaller, the value of the nonlinear error component measured becomes smaller. Further, as the number n of sample areas increases, the averaging effect increases, so that the accuracy of the linear error increases as the value of n increases as shown by the function G _n (n ^{1/2 in} Gaussian distribution) in FIG. It will increase monotonically. Therefore, 3 of the standard deviation of the nonlinear error component
Multiplying the by dividing (NLE in FIG 7 (3 [sigma])) of the product of the function F _n and G _n obtains the evaluation value A _n, the evaluation value A _n is 1
As shown in 0, the value is almost constant.

Next, among the N non-linear error components, the (N-1) sample regions excluding the sample region having the largest non-linear error component are three times the standard deviation of the non-linear error components of the alignment error. And the evaluation value A _N-1 . Thereafter, in the same manner, the sample area having the largest nonlinear error component is sequentially excluded and left (N
-2), (N-3), ... Evaluated values A _N-2 , A _N-3 , ... are obtained for the sample areas, and as one example, a series of these evaluated values A _N-1 , A _N-2 , ... the average value <A _n> the threshold of. Then, performed, the alignment in the EGA method using a sample region evaluation value A _Ni is left as it becomes a threshold value <A _n> or less. Further, instead of the average value <A _n>, for example, may be used a central value, etc. between the maximum value and the minimum value.

If the number n of each nonlinear error amount is small,
The average value <A_n>less than
Excluding the sample area until
There is. Therefore, is it three times the standard deviation of the nonlinear error component?
Average of evaluation values obtained from <A _nThe variation of
Three times the standard deviation for the number n of sample areas, that is, gau
In the distribution, <A_n> / (N-3)^1/2Ask for. That
Evaluation value A_NiIs the threshold (<A_n> + <A_n> / (N-
3)^1/2) Remaining sample when
EGA calculation may be performed using the region.

Next, according to the second alignment method, three times the standard deviation of the nonlinear error component (NLE (3
σ)) itself is compared with a threshold value C _n that increases monotonically with respect to the number n of sample areas as shown in FIG.
When LE (3σ) is larger than the threshold value C _n , the sample area is excluded. As a result, the same result as the first alignment method is obtained. In the Gaussian distribution, the threshold value C _n is K ₁ {n using a predetermined constant K _1.
(N-3)} ^1/2 .

Next, according to the third alignment method of the present invention, the surface condition of the substrate (4), for example, the condition of unevenness, is detected in order to find a jump shot. Then, for example, a sample area in which the unevenness exceeds a predetermined allowable value is regarded as a jump shot and is eliminated. This allows
Jump shots can be excluded without measuring the coordinate values of the sample area.

Further, when the projection exposure is performed, the positions of all the processed regions (shot regions) are measured on the first substrate (wafer) of the lot, and the jump is performed by the above-described first to third alignment methods. Shots may be excluded. In this case, the sample areas for the alignment of the second and subsequent substrates can be automatically selected with an appropriate number and distribution from the processed region left on the first substrate.

[0029]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the alignment method according to the present invention will be described below with reference to the drawings. FIG. 1 shows a projection exposure apparatus in which the alignment method of this embodiment is used. In FIG. 1, the illumination light IL for exposure from the illumination optical system 1 is shown.
Illuminates the pattern on the reticle 2 with a uniform illuminance distribution,
A reduced image of the pattern through the projection optical system 3 is exposed on each shot area on the wafer 4 coated with the photoresist. Here, the Z axis is taken parallel to the optical axis AX of the projection optical system 3, the X axis is taken parallel to the paper surface of FIG. 1 in the plane perpendicular to the Z axis, and the Y axis is taken perpendicular to the paper surface of FIG. .

The wafer 4 is held on a wafer stage 6 via a wafer holder 5, and the wafer stage 6 is an XY stage for positioning the wafer 4 in the X and Y directions.
It is composed of a Z stage for moving the wafer 4 in the Z direction, a θ stage for rotating, a leveling stage for correcting the tilt angle of the wafer 4, and the like. A laser beam from an external biaxial laser interferometer 8 is reflected by a biaxial moving mirror 7 (only the X axis is shown in FIG. 1) fixed to the upper surface of the wafer stage 6, and the laser beam is reflected. The X coordinate and the Y coordinate of the wafer stage 6 are constantly measured by the interferometer 8. The coordinate system determined based on the coordinates (X, Y) measured by the laser interferometer 8 in this way is called a stage coordinate system or a stationary coordinate system. The measured coordinates (X, Y) are supplied to the main control system 9 that controls the operation of the entire apparatus, and the main control system 9 based on the measured coordinates,
The positioning operation of the wafer stage 6 is controlled via the wafer stage drive system 10.

On the side surface of the projection optical system 3, a grazing incidence type focus position detection system including a projection optical system 11 and a light receiving optical system 13 is arranged. From the projection optical system 11, the slit image 12 (or the dot pattern image) is obliquely provided to the optical axis AX of the projection optical system 3 and to a plurality of (three or more) measurement points on the wafer 4 near the optical axis AX, for example. Etc. are projected, and the reflected light from the plurality of slit images 12 re-images the slit images on, for example, a photoelectric microscope in the light receiving optical system 13. In this case, the position of the surface of the wafer 4 is Z
In the light receiving optical system 13, a plurality of focus signals corresponding to Z coordinates (focus positions) of a plurality of measurement points on the wafer 4 are used in the light receiving optical system 13 by utilizing the fact that the re-formed slit image is laterally displaced when displaced in the direction. Are generated, and these focus signals are supplied to the focus signal processing system 14.

The focus signal processing system 14 is a Z coordinate of a plane approximated to a surface including a plurality of measurement points on the wafer 4,
And the tilt angle is calculated and supplied to the wafer stage drive system 10, and the Z coordinate of each measurement point is supplied to the main control system 9. The wafer stage drive system 10 controls the wafer stage 6 so that the Z coordinate and the tilt angle supplied during the actual exposure match the image plane of the projection optical system 3. on the other hand,
At the time of alignment and when measuring the unevenness distribution on the surface of the wafer 4, the main control system 6 moves the wafer 4 in the XY plane via the wafer stage 6 while the focus signal processing system moves, as will be described later. The focus signal from 14 is processed to check the concavo-convex state of each sample shot on the wafer 4.

In general, a semiconductor device or the like is manufactured by repeating the step of projecting and exposing the reticle pattern on each shot area on the wafer to perform development and the like 10 to 20 times. Therefore, the reticle to be exposed from now on. It is necessary to accurately align the pattern with the circuit pattern formed in each shot area of the wafer 4 in the steps up to that point. Therefore, in the projection exposure apparatus of FIG. 1, the TTL (through the lens) method for detecting the coordinates of the wafer mark attached to each shot area on the wafer 4 and the laser step alignment method are used. (LSA
System) alignment optical system 15 is provided. As the alignment optical system 15, an imaging method, a so-called two-beam interference method, or the like can be used, but in this embodiment, the LSA method is used as an example.

In this case, the laser beam for detection emitted from the alignment optical system 15 enters the projection optical system 3 via the mirror 16 for bending the optical path, and the laser beam passing through the projection optical system 3 is shown in FIG. As shown in (b), Y
Wafer 4 as a slit-shaped light spot 17 long in the direction
Focused on top. By driving the wafer stage 6 of FIG.
The X-axis wafer mark Mx _i to be detected on the wafer 4 is moved in the X direction with respect to the light spot 17. The wafer mark Mx _i is formed by arranging a plurality of dot patterns arranged in the Y direction at a predetermined pitch in a row in the X direction. When the wafer mark Mx _i crosses the light spot 17, the wafer mark Mx _i is diffracted in the predetermined direction. Since the light is emitted, the X coordinate of the wafer mark Mx _i is detected.

Returning to FIG. 1, a light spot 17 on the wafer 4 is obtained.
The diffracted light from the wafer mark returns to the alignment optical system 15 via the projection optical system 3 and the mirror 16, and the alignment signal obtained from the alignment optical system 15 to the alignment signal processing system 18 by photoelectrically converting the diffracted light. Is supplied. The coordinates (X, Y) measured by the laser interferometer 8 are also supplied to the alignment signal processing system 18, and when the light spot 17 is at the center position of the wafer mark for the X axis, the alignment signal processing system 18 is supplied. The X coordinate of is detected and supplied to the main control system 8. An Y-axis alignment optical system (not shown) is also provided, and the alignment optical system and the alignment signal processing system 18 detect the Y-coordinate corresponding to the Y-axis wafer mark. It is supplied to the main control system 8.

As a basic operation, first, when the wafer 4 is loaded on the wafer holder 5, the main control system 9 moves the wafer 4 in the XY plane via the wafer stage drive system 10 and the wafer stage 6. The wafer 4
The wafer mark attached to the upper sample shot is set near the position where the light spot is irradiated from the alignment optical system 15 (or the Y-axis alignment optical system). The positioning (coarse alignment) of the wafer 4 in this case is performed based on the coordinates of each shot area defined on the coordinate system on the wafer 4, for example. After that, by moving the wafer 4 in the X direction or the Y direction,
The alignment signal processing system 18 measures the coordinates of the wafer mark with high accuracy. The main control system 9 calculates the array coordinates in the stage coordinate system (X, Y) of all shot areas on the wafer 4 as described later using the coordinates of the wafer marks of each sample shot thus measured. Then, based on the calculation result, the pattern image of the reticle 2 is exposed in each shot area by the step-and-repeat method.

Next, various examples of the alignment method (positioning method) in this embodiment will be described in detail. (A) Basic Alignment Method FIG. 2A shows an example of a shot array of the wafer 4 of this example. In FIG. 2A, M pieces (FIG. 2) are arranged on the wafer 4.
In (a), M = 68) shot areas ES _{1 to} ES _M are arranged, a circuit pattern is formed in each shot area ES _j (j = _{1 to M} ), and the wafer mark Mx _j for the X axis is formed. , And a wafer mark My _j for the Y axis are attached. In this case, the x coordinate x of the center of each wafer mark Mx _j on the sample coordinate system (x, y) set on the wafer 4.
_j and the y coordinate y of each wafer mark My _j for the Y axis
_j is previously stored as a design coordinate in the storage device of the main control system 9 in FIG. In the following, the x coordinate of the center of the wafer mark Mx _j , and the wafer mark Myj. The y coordinate of the center of the shot area ES _j is the x coordinate in the sample coordinate system of the center of the shot area ES _j ,
And the y coordinate, and the coordinates of the wafer mark are regarded as the coordinates of the shot area. Actually, there is generally an offset between the center coordinate of the wafer mark and the center coordinate of the corresponding shot area, but the offset is ignored here for simplicity.

At this time, using six parameters (scaling Rx in the X direction, scaling Ry in the Y direction, rotation θ, orthogonality error w, offset Ox in the X direction, and offset Oy in the Y direction), The conversion relationship from the sample coordinate system (x, y) to the stage coordinate system (X, Y) is defined by the conversion formula of 1). And those 6
N shot areas (4 ≦ N ≦ M) selected from the M shot areas, that is, N sample shots SA _{1 to} SA _N in order to determine the value of each parameter.
About the X coordinate X of the wafer mark for the X axis attached to each of them using the alignment optical system 15 of FIG.
M _i and the Y coordinate YM _i of the wafer mark for the Y axis (i = 1
~ N) is measured. The coordinates (X
M _i, YM _i) sequence coordinates (x _i in design from, y _i) by subtracting the resulting vector (Δx _i, Δx _i) is a vector of alignment error.

Next, by substituting the designed array coordinates (x _i , y _i ) of each sample shot SA _i as the coordinates (x, y) in (Equation 1), the stage coordinate system ( Calculated array coordinates (X _i , Y _i ) in (X, Y)
Is represented as a function of the six parameters and the designed array coordinates. Then, the values of the six parameters are determined by EGA calculation so that the sum of squares of the alignment errors of the N sample shots represented by (Equation 2), that is, the residual error component is minimized.

Then, the six determined parameters (R
x, Ry, θ, w, Ox, Oy) and each sample show
SA_iDesigned array coordinates (x_i, Y_i) (Equation 1)
To each sample shot SA_iThe most
The final calculated array coordinates (X _i, Y_i). This and
Computed array coordinates (X_i, Y_i) To the design array
Coordinates (x_i, Y_i) Is the vector obtained is linear
The vector of the error component, and the measured coordinates (XM_i，
YM_i) To the calculated array coordinates (X_i, Y_i)
Is the vector of the nonlinear error component
It And, in this example, the non-linear error component vector
For sample shots with large logarithms,
To exclude the sample shots that were left
6 by EGA calculation based on the measured array coordinates
The parameters of. After that, using (Equation 1),
Calculate the array coordinates of all shot areas on the
Position each shot area of the wafer at the exposure position
Arrange and expose. Exclude that jump shot below
A specific example of the method will be described.

(B) First method of excluding jump shots In this case, all the shot areas of the leading wafer of a certain lot are regarded as sample shots for measurement,
A method of identifying a jump shot from among them will be described. However, this method can be similarly applied even when the number of sample shots is smaller than the number of all shot areas.

FIG. 3A shows, by way of example, 32 on a wafer.
All shot areas are sample shots SA _{1 to} SA
₃₂ shows a case where deemed, 3 sample shots SA ₁ -SA (remainder obtained by subtracting the coordinates of the design from the measurement coordinate) ₃₂ alignment error in the vector V ₁ in (a)
~ V ₃₂ . Further, FIG. 3B shows the number of sample shots measured, the average value of the X component of the alignment error, and the average value of the Y component of the alignment error, and these X values.
3 times the standard deviation (3σ) of the component and the Y component, the error from 1 of the scaling Rx and Ry, and the orthogonality error w,
And the value of rotation θ are shown. Further, FIG. 3C shows the X direction and Y obtained for each sample shot from the parameters Rx, Ry, w, θ, Ox, Oy.
The value of 3 times (3σ) of the standard deviation of the linear error component in the direction is shown, and FIG. 3D shows the value of 3 times (3σ) of the standard deviation of the nonlinear error component in the X and Y directions obtained for each sample shot. ) Value.

FIG. 4A shows the vectors VL _{1 to} VL ₃₂ of the linear error components of the alignment error in FIG. 3A, and FIG. 4B shows the alignment error in FIG. 3A. 4A to 4C show nonlinear error component vectors VN _{1 to} VN ₃₂ obtained by subtracting the linear error component of FIG. In this case, as can be seen from FIG. 4B, the sample shot S
Since the absolute value of the nonlinear error vector VN ₈ of A ₈ is particularly large, the sample shot SA ₈ is considered to be a jump shot. Further, even in the alignment error stage of FIG. 3A, the absolute value of the vector V ₈ indicating the error in the sample shot SA ₈ is the maximum, and therefore the jump shot SA ₈ causes the linear error component of the linear error component shown in FIG. The vector is also likely to be affected.

Therefore, from FIG. 3A, the jump shot SA
₈Using the remaining 31 sample shots
The linear error component and the nonlinear error component
This is shown in FIGS. 5 and 6. Figure 5 (a) shows a sample shot
SA₈Excluding 31 sample shots SA₁~ SA₇,
SA₉~ SA₃₂Alignment error vector V₁~ V
₇, V₉~ V₃₂Fig. 5 (b)-(d) shows 31
The same values as in Fig. 3 (b) to (d) for sample shots
The results obtained are shown below. In addition, FIG. 6 (a) corresponds to FIG. 5 (a).
Vector V of the linear error component of the alignment error in
L₁~ VL₃₂FIG. 6 (b) shows the array of FIG. 5 (a).
Subtract the linear error component of Fig. 6 (a) from the
Vector VN of the nonlinear error component obtained by₁~ VN₃₂To
Show.

FIGS. 3B to 3D and FIGS. 5B to 5D.
By comparing with, it can be seen that the nonlinear error component in FIG. 5A is smaller than the nonlinear error component in FIG. 3A. Subsequently, the work of excluding the sample shot having the largest nonlinear error component from the remaining sample shots is repeatedly executed, and the standard deviation 3 of the absolute value of the vector of the nonlinear error component is repeated.
If the multiple is NLE (3σ), NLE (3σ) is represented by a polygonal line 23 in FIG. 7 as a function of the number n of the remaining sample shots. However, the horizontal axis of FIG. 7 represents the number n of remaining sample shots when the number of initial sample shots is N, and the vertical axis represents NLE (3σ). As shown in FIG. 7, NLE (3σ) decreases monotonically as the number n of remaining sample shots decreases, and becomes 0 when the number of remaining sample shots reaches 3.

The reduction in the value of the nonlinear error component means that the probability that the random nonlinear error component is regarded as the linear error component increases. In other words, when the number of remaining sample shots n is small, the probability that a random non-linear error component mixes with the linear error component is high, and the reliability of the linear error component is low.
As the number of remaining sample shots increases, the reliability of the linear error component increases due to the averaging effect.

FIG. 8 corresponds to the reliability of the linear error component.
Function F_nIndicates. In this case, the remaining sample
The number of shots is n, and those sample shots are
The design coordinates of the wafer mark attached to the
₁, Y₁), (X₂, Y₂), ..., (x_n, Y_n), Araime
The wafer mark measurement accuracy by the_aThen,
This function F_nAre their values (x₁, Y₁, X₂，
y₂, ..., x_n, Y_n, Σ _a) Function.

Specifically, in order to obtain the value of the function F _n , when the linear error component of the array coordinate of n sample shots is 0, the measured value (alignment error) of the array coordinate of each sample shot is randomly selected. Error is given, and least squares approximation is performed based on this data to obtain six parameters. Then, using these parameters, the standard deviation three times (3σ) of the linear error component of the array coordinates of the n sample shots may be obtained and plotted with respect to the number n. Further, using a predetermined constant F ₀ , the function F _n is as follows for a normal Gaussian random error.

[0049]

[Mathematical formula-see original document] F _n = F ₀ · (n−3) ^{1/2 Further} , apart from the function F _n , such a random error is given to obtain the non-linear error component of the array coordinate of the sample shot, The averaging effect increases with the number n of sample shots. The accuracy of the linear error component should increase accordingly.

FIG. 9 shows the accuracy of the linear error component as a function G _n of the number n of sample shots. And the function G _n Is a function that increases monotonically as the number n increases. For the random error of the normal Gaussian distribution, the function G _n Is as follows.

[0051]

(4) G _n = N ^1/2 or more, the function F _{n in} FIG. 8 also represents the degree of reliability of the nonlinear error component, and the function G _{n in} FIG. Also represents the averaging effect on the nonlinear error component, so the value NLE (3σ) that is three times the standard deviation of the nonlinear error component shown in FIG. 7 is divided by the product of the functions F _n and G _n to obtain the evaluation value A. _{Find n} . That is, there is the following relationship.

[0052]

[Equation 5] A _n = NLE (3σ) / (F _n · G _n ). This evaluation value A _n becomes almost constant with respect to the number n of sample shots as shown in FIG. 10, but jump shots are mixed. It will be a large value. Therefore, the number n of sample shots is until the maximum value N to four is the minimum value, after each determined evaluation values A _n, calculates an average value <A _n thereof evaluation value A _n>. As a result, the evaluation value A
_n it can be considered in the n samples available for when the average value <A _n> greater, contains a large jump shot nonlinear error component. Therefore, the be excluded as a shot jump as an example, the sample shots other than the sample shots are left when the evaluation value A _n first comes to the average value <A _n> or less.

In the example of concrete 10, the number of sample shots (N-3) from for evaluation value A _n where became (N-49) is set to the average value <A _n> or less The removal of the jump shot is stopped when the sample shot becomes (N-4), that is, when the four jump shots are removed. However, when the nonlinear error is very large, the evaluation value A _n
There When reducing the sample shots to an average value <A _n> or less, there is a possibility to rather deteriorate the alignment accuracy. Therefore, seeking triple A (3 [sigma]) of the average standard deviation of <A _n> in the number n of sample shots, the threshold B _n evaluation value A _n is formed of _{(<A n> + A (3σ} ))
In the following cases, skip shots may be removed in a range where the evaluation value A _n exceeds the threshold value B _n without removing the sample shots. Assuming a Gaussian distribution, the threshold value B _n can be expressed as follows using a predetermined coefficient k. The coefficient k is usually 1, and is increased or decreased from 1 as necessary.

[0054]

Threshold B _n of Equation 6] _{B n = <A n> + A} (3σ) = <A n> + k · <A n> / (n-3) 1/2 This equation (6) is curved in FIG. 10 As shown by 22, it is a function that decreases monotonically with respect to the number n of sample shots. Further, in the example of FIG. 10, when the number of sample shots becomes (N−1), the evaluation value A _n becomes _{equal to} or less than the threshold value B _n , so that only one sample shot is removed as a jump shot. is there. The jump shot is removed by the above method.

In the above embodiment, the variation of the non-linear error component is N times the standard deviation of the non-linear error component.
Although LE (3σ) is used, a similar result can be obtained by using, for example, the worst value of the nonlinear error component in the remaining sample shots instead. Further, in the above example, since all shot areas on the wafer are regarded as sample shots, it is considered that the number of sample shots is still too large in most cases even if the jump shots are excluded. Therefore, from the remaining sample shots, for example, only the required number of sample shots close to the outer periphery of the wafer are selected. For the remaining wafers in the lot, only the sample shot finally selected on the first wafer is measured in the stage coordinate system, and the EGA method is used to perform alignment using the measurement result. As a result, when the positions where jump shots occur are almost the same in the same lot, alignment can be performed with high accuracy without being affected by jump shots.

[0056] Further, with respect to the number of sample shots finally left, for example may be changed automatically the number of the sample shots according to the average value the value of the <A _n> of the evaluation value A _n shown in FIG. 10 . For example, by when the average value <A _n> is large to increase the number of sample shots, poor device of shot sequences precision alignment is performed by the EGA method using a number of sample shots, good device with the shot array accuracy is small samples The shot can be used to perform the alignment by the EGA method.

Further, in the above-mentioned embodiment, the non-linearity shown in FIG.
Three times the standard deviation of the error component NLE (3σ) is calculated by the function F_n
And G_nEvaluation value A divided by the product of_nI am looking for
3 times the standard deviation of NLE (3σ)
Good. In this case, the function F _nAnd G_nThe product of
Coefficient C₀The product of and is the function C_nAnd this function C_nAnd NL
It will be compared with E (3σ). (Equation 3), and
From (Equation 4), the Gaussian distribution has a function C_nIs as follows
It

[0058]

C _n = C ₀ · F ₀ · (n−3) ^1/2 · n ^1/2 = K ₁ · {n (n−3)} ^1/2 where coefficient K ₁ is C ₀ F ₀ , and the function C of (Equation 7)
_n is a function as shown by the curve 24 in FIG. The coefficient K ₁ is obtained by averaging the results obtained by dividing NLE (3σ) by {n (n-3)} ^1/2 within the range in which the value of n changes from N to 4. Determined by Then, in FIG. 7, the jump shots are removed until NLE (3σ) of the polygonal line 23 becomes equal to or less than the function C _{n of the} curve 23.

Further, in the above-described embodiment, all shot areas of the first wafer of the lot are regarded as sample shots, and the jump shots of the second and subsequent wafers are determined. May be a shot area selected from all shot areas. Furthermore, for all wafers, the alignment error of a predetermined number of sample shots is measured, and each jump shot is determined based on this measurement result.
E using the sample shots left for each wafer
The GA method may be used for alignment. In this example, the function G _n = N ^1/2 , but if the measurement result is not a Gaussian distribution, the function G _n May be a predetermined function of n according to the distribution.

(C) Second Method for Excluding Jump Shots Next, a method for removing jump shots using the focus position detection system including the projection optical system 11 and the light receiving optical system 13 in FIG. 1 will be described. In this example, the wafer to be exposed is the wafer 4 shown in FIG. 2A, and the jump shots are removed from the _N sample shots SA _{1 to} SAN. First, the operation in the Z direction is locked by the wafer stage 6, and the wafer stage 6 is determined based on the array coordinates of each sample shot SA _{i on} the sample coordinate system (x, y).
Is driven, and the formation area of the wafer marks Mx _i and My _i attached to each sample shot SA _i is the projection position (measurement point) of the slit image by the focus position detection system.
To cross. At this time, the focus signal processing system 1
In step 4, the focus signal from the light receiving optical system 13 is used to obtain the unevenness distribution in the formation region of the wafer marks Mx _i and My _i , and the information of this unevenness distribution is supplied to the main control system 9.

The main control system 9 finds the pitch of the wafer mark in the measuring direction, the step of the unevenness, etc. from the unevenness distribution, and, for example, the sample shot to which the wafer mark whose unevenness of the unevenness does not reach a predetermined allowable value belongs to. Is excluded as a jump shot. That is, even if the position of a wafer mark in a poorly formed state is measured by the alignment system, a measurement error, which is a non-linear error component, has a high probability of occurrence, and therefore is regarded as a jump shot in advance and removed. As a result, it is possible to remove a sample shot having a high probability of a jump shot without performing measurement by the alignment system.

In the above embodiment, the TTL alignment system is used as the alignment system.
It goes without saying that an off-axis type alignment system or a TTR (through the reticle) type alignment system may be used. Further, in the above-described embodiment, each sample shot has one wafer mark on the X-axis and one wafer mark on the Y-axis.
Although they are provided one by one, three or more wafer marks may be provided for each sample shot. Further, it is not always necessary to measure the coordinates of all the wafer marks in each sample shot.

As described above, the present invention is not limited to the above-described embodiments, and various configurations can be adopted without departing from the gist of the present invention.

[0064]

According to the first alignment method of the present invention, one sample area (sample shot) having the largest non-linear component is sequentially removed, and the remaining sample areas are evaluated from the variations of the non-linear error components. The value is calculated, and the sample shot (jump shot) with a large non-linear error component is removed until this evaluation value becomes equal to or less than the threshold value according to the average value of the evaluation value. Therefore, there is an advantage that the position can be accurately adjusted.

Further, since the threshold value is obtained from a plurality of evaluation values corresponding to the number of sample areas, there is an advantage that a reference for detecting a jump shot can be automatically set.
Furthermore, for example, by adjusting the number of sample regions to be actually measured according to the size of the threshold value, the arrangement of sample regions according to the state of the substrate (wafer) can be determined.

In this case, if the threshold value is the average value of the plurality of evaluation values, the calculation is easy. Further, when the threshold value is a value obtained by adding the variation of the evaluation values to the average value of the plurality of evaluation values, the probability that a sample area that is not a jump shot is erroneously removed as a jump shot becomes small. According to the second alignment method of the present invention,
Since the non-linearity error component variation obtained for each of the remaining sample areas is compared with a predetermined function to remove the jump shot, the jump shot can be reliably removed as in the first alignment method. Therefore, there is an advantage that the position can be accurately adjusted.

At this time, the predetermined function is {n (n-
3)} ^1/2 Is effective when the error has a Gaussian distribution. Furthermore, according to the second alignment method of the present invention, by detecting the state of the surface of the substrate,
There is an advantage that the jump shot can be removed quickly without actually measuring the array coordinates of the processed region (sample shot).

[Brief description of drawings]

FIG. 1 is a configuration diagram showing a projection exposure apparatus in which an embodiment of an alignment method according to the present invention is executed.

2A is a plan view showing an arrangement example of sample shots on a wafer to be exposed in the embodiment, and FIG. 2B is an explanatory diagram of a wafer mark detecting method.

FIG. 3 is a diagram showing an example of an array of sample shots including jump shots and an alignment error in the shot array in the embodiment.

4A is a diagram showing a vector of a linear error component in the shot array of FIG. 3, and FIG. 4B is a diagram showing a vector of a non-linear error component in the shot array of FIG.

5 is a diagram showing alignment errors and the like in a shot array obtained by removing jump shots from the shot array of FIG.

6A is a diagram showing a vector of a linear error component in the shot array of FIG. 5, and FIG. 6B is a diagram showing a vector of a non-linear error component in the shot array of FIG.

FIG. 7 is a diagram showing a relationship of NLE (3σ), which is three times the standard deviation of the nonlinear error component, with respect to the number n of sample shots.

FIG. 8 is a diagram showing a relationship of a reliability function F _n of a linear error component with respect to the number n of sample shots.

FIG. 9 is a diagram showing the relationship of the function G _n of the averaging effect with respect to the number n of sample shots.

10 is three times the standard deviation NL of the non-linear error component of FIG.
Evaluation value A obtained by dividing E (3σ) by the function F _n · G _n
of _n, it is a diagram illustrating a relationship to the number n of sample shots.

FIG. 11 is a diagram showing an array of sample shots when there is no jump shot and an alignment error in the array in the conventional example.

12 is a diagram showing a vector of a linear error component in the array of sample shots of FIG.

FIG. 13 is a diagram showing an array of sample shots including a jump shot, an alignment error in the array, and the like in the conventional example.

14 is a diagram showing a vector of a linear error component in the array of sample shots of FIG.

[Explanation of symbols]

2 reticle 3 projection optical system 4 wafer 6 wafer stage 8 laser interferometer 9 main control system 11 projection optical system of focus position detection system 13 light receiving optical system of focus position detection system 14 focus signal processing system 15 alignment optical system 18 alignment signal processing systems ES ₁ ~ES _M shot areas SA ₁ -SA _N sample shot Mx _i, My _i wafer mark

Claims (6)

[Claims] 1. When aligning each of a plurality of processing regions set on a substrate with a predetermined processing position in a coordinate system that defines a moving position of the substrate, the plurality of processing regions are processed. Among these, the plurality of processed regions on the substrate are obtained by measuring the coordinate positions on the coordinate system of a predetermined number or more of preselected sample regions and statistically calculating the plurality of measured coordinate positions. By calculating array coordinates of each of the above on the coordinate system and controlling the moving position of the substrate according to the calculated array coordinates, each of the plurality of processed regions is aligned with the processing position. In the method, a first step of measuring coordinate positions on the coordinate system of N preselected N (N is an integer of 4 or more) sample regions among the plurality of processed regions; Sun A second non-linear error component from each of the coordinate positions measured in the first step from the designed position, and a variation of the plurality of non-linear error components divided by the N function to obtain an evaluation value. The number of remaining sample areas by repeating the second step by excluding the sample area having the largest non-linear error component in the second step until the number of the sample areas reaches a predetermined number. And a third step of obtaining an evaluation value corresponding to
In a range in which the individual evaluation values obtained in the step are larger than a predetermined threshold value based on the plurality of evaluation values, the sample area having the largest non-linear error component in the second step is removed.
Step; statistical processing of the coordinate positions measured in the first step with respect to the sample area remaining in the fourth step to obtain array coordinates on the coordinate system of each of the plurality of processed areas on the substrate. A fifth step of calculating; and a registration method comprising: 2. The alignment method according to claim 1, wherein the predetermined threshold value is an average value of the plurality of evaluation values. 3. The alignment method according to claim 1, wherein the predetermined threshold value is a value obtained by adding variation of the evaluation values to an average value of the plurality of evaluation values. 4. When aligning each of a plurality of processing regions set on a substrate with a predetermined processing position within a coordinate system that defines a moving position of the substrate, the plurality of processing regions are processed. Among these, the plurality of processed regions on the substrate are obtained by measuring the coordinate positions on the coordinate system of a predetermined number or more of preselected sample regions and statistically calculating the plurality of measured coordinate positions. By calculating array coordinates of each of the above on the coordinate system and controlling the moving position of the substrate according to the calculated array coordinates, each of the plurality of processed regions is aligned with the processing position. In the method, a first step of measuring coordinate positions on the coordinate system of N preselected N (N is an integer of 4 or more) sample regions among the plurality of processed regions; Sun A second step of obtaining a non-linear error component from the designed position of the coordinate position measured in the first step for each region, and a variation of the plurality of non-linear error components; By removing the sample area having the largest non-linear error component in the second step and repeating the second step until a predetermined number is obtained, the variation of the non-linear error component corresponding to the number of remaining sample areas is obtained. A third step; in a range in which the variation of each non-linear error component obtained in the second step and the third step is larger than a predetermined function that monotonically increases according to the number of the sample areas to be calculated. A fourth step of removing the sample area having the largest non-linear error component in the second step; and a total of the sample areas remaining in the fourth step in the first step. A fifth step of statistically processing the measured coordinate positions to calculate array coordinates on the coordinate system of each of the plurality of processed regions on the substrate; 5. The number of sample areas to be calculated is n
As the predetermined function, a predetermined constant is {n (n-
3)} The function obtained by multiplying by ^1/2 is the alignment device according to claim 4. 6. The plurality of processed regions when aligning each of the plurality of processed regions set on the substrate with a predetermined processing position within a coordinate system that defines a moving position of the substrate. Among these, the plurality of processed regions on the substrate are obtained by measuring the coordinate positions on the coordinate system of a predetermined number or more of preselected sample regions and statistically calculating the plurality of measured coordinate positions. By calculating array coordinates of each of the above on the coordinate system and controlling the moving position of the substrate according to the calculated array coordinates, each of the plurality of processed regions is aligned with the processing position. In the method, the state of the surface of the substrate in a preselected N (N is an integer of 4 or more) sample region among the plurality of processed regions is detected, and the surface state is detected based on the detection result. Exclude sample area that is determined to be worse than a predetermined allowable value first
A step; a second step of measuring the coordinate position on the coordinate system of the sample area left in the first step; a coordinate measured in the second step of the sample area left in the first step A third step of statistically processing the positions to calculate array coordinates on the coordinate system of each of the plurality of processed regions on the substrate;