JP3348918B2 - Alignment method, exposure method using the method, and device manufacturing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a projection exposure apparatus for sequentially transferring a reticle pattern image to each shot area on a wafer based on array coordinates predicted using, for example, a statistical method. The present invention relates to a suitable alignment method applied to a case where each shot area is sequentially aligned.

[0002]

2. Description of the Related Art When a semiconductor device or a liquid crystal display device is manufactured by a photolithography process, a pattern image of a photomask or a reticle (hereinafter collectively referred to as a "reticle") is formed on a photosensitive substrate through a projection optical system. A projection exposure apparatus that projects onto a shot area is used. In recent years, as a projection exposure apparatus of this type, a photosensitive substrate is placed on a two-dimensionally movable stage, and the stage is moved (stepped) by the stage, so that a reticle pattern image is formed on the photosensitive substrate. A so-called step-and-repeat type exposure apparatus that repeats an operation of sequentially exposing each shot area,
For example, a reduction projection type exposure apparatus (stepper) is frequently used.

[0003] For example, a semiconductor element is formed by laminating a large number of circuit patterns on a wafer coated with a photosensitive material as a photosensitive substrate. In such a method, it is necessary to accurately align each shot area on the wafer on which a circuit pattern has already been formed with a pattern image of a reticle to be exposed, that is, the alignment (alignment) between the wafer and the reticle. A conventional method of aligning a wafer in a stepper or the like is roughly as follows (for example, see Japanese Unexamined Patent Publication No.
-44429).

That is, a chip pattern including alignment marks (alignment marks) is provided on a number of shot areas regularly arranged based on predetermined arrangement coordinates on a wafer to be processed. Are formed. However, when another pattern is superimposed on the formed pattern, even if the wafer is stepped based on the set arrangement coordinates, satisfactory overlay accuracy is not always obtained due to the following factors.

(1) Remaining rotational error of wafer Θ (2) Orthogonality error W of stage coordinate system (or shot arrangement) (3) Linear expansion / contraction Rx, Ry of wafer (4) Offset of wafer (center position) (parallel) Move) O _X ,
O _Y

Since these four error amounts can be represented by six parameters, a two-row × two-column transformation matrix A composed of elements represented by the four parameters and an offset (parallel movement ) Consider a transformation matrix O of 2 rows × 1 column having O _X and O _Y as elements. Then, the designed array coordinate values (D _Xn , D _Yn ) (n = 0, 1, 1) of each shot area on the wafer
2, ‥‥) and the actual array coordinate values (F _Xn , F _Yn ) to be aligned by the step-and-repeat method are expressed as follows using these transformation matrices A, O: And

[0007]

(Equation 1)

At this time, array coordinate values (F) obtained by actually measuring a plurality of shot areas selected from the wafer.
M _Xn , FM _Yn ) and the calculated array coordinate values (F _Xn , F _Xn ,
The transformation matrices A and O are determined using the least squares method so that the average deviation from F _Yn ) is minimized. Conventionally, on the basis of the determined transformation matrices A and O and the designed array coordinate values (D _xn , D _yn ), the above formula (1) is used to calculate the position to be actually aligned. Array coordinate value (F
_Xn , F _Yn ) were calculated, and each shot area of the wafer was positioned based on the calculated coordinate values.

[0009]

However, even if the wafers are aligned in accordance with the calculated array coordinate values (F _Xn , F _Yn ) calculated as described above, due to the following factors, There has been a disadvantage that satisfactory overlay accuracy is not always obtained. (1) residual rotation error θ of the circuit pattern (chip pattern) on each shot area of the wafer, (2) orthogonality error w of the coordinate system (chip pattern) on the wafer, (3) two orthogonal directions of the chip pattern These are the displacement and rotation of the reticle from a predetermined position and the projection of the projection optical system when the chip pattern is first printed (first layer) on each shot area on the wafer. This is caused by a magnification error or distortion (distortion) of the projection optical system. Further, these factors change depending on the strain applied to the wafer by a processing process or the like.

[0010] The present invention has been made in view of the points mow斯, when performing alignment between each shot area and the reticle on the photosensitive substrate based on array coordinates predicted using a statistical method, is transferred to each shot area An object of the present invention is to suppress the influence of the expansion and contraction and rotation of the chip pattern itself, and to superimpose the projection image of the pattern of the reticle and the chip pattern of each shot area on the photosensitive substrate with higher precision. .

[0011]

Alignment method according to the invention According to an aspect of the order to transfer the pattern on the mask (2) on a substrate (8), it is two-dimensionally arranged on the substrate,
A method of aligning each of a plurality of shot areas (27-n) provided with respective marks with respect to a predetermined transfer position in a stationary coordinate system (X, Y) that defines a movement position of a substrate. in at least three sample shot area of the plurality of shot areas on a substrate (S
A), marks (29-n, 30-n,
Of 34-n, 35-n) , a first step of measuring the position information in the stationary coordinate system (step 103), the first transformation parameters relating to sequences on the substrate of multiple shot areas (A, O) and a conversion parameter relating to an error in the relative imaging relationship between the pattern and the pattern in the shot area.
And at least mask the shot area
Multiple shot areas on the substrate when the pattern is projected
Linear expansion and contraction in each of two-dimensional array directions
The second conversion parameter (B) individually including the error (rx, ry) is used as the position of the plurality of marks measured in the first step.
The information is converted into one coordinate transformation including the first and second transformation parameters.
Simultaneous calculation by statistical calculation using the equation (Equation 10)
And a second step (steps 104 and 107) of calculating array position information (G _xn , G _yn ) of each of the plurality of shot areas on the stationary coordinate system, and array position information calculated in the second step And a third step (step 108) of aligning each of the plurality of shot areas on the substrate with respect to the transfer position.

[0012]

In this case, when the mark arrangement is relatively simple and the chip pattern shape error is represented by two parameters (for example, the rotation error θ of the chip rotation and the isotropic chip scaling M), Two or more each of the reference positions (28-n) in each shot area (27-n) are designed to have a fixed relative positional relationship with each other.
Positioning marks indicating dimensional coordinates are arranged, and in the first step, at least four of the positioning marks (29-n, 30-n) indicating the two-dimensional coordinates are selected in advance. It is desirable to measure the coordinate position of the alignment mark on the stationary coordinate system (X, Y). Next, the exposure method according to the present invention uses the alignment method of the present invention.
The plurality of shot areas aligned using a
Over the area, the pattern of the mask is sequentially projected through the projection optical system.
Next projection, the pattern is sequentially transferred onto the substrate
It includes steps. Also, the device according to the present invention
The fabrication method uses the exposure method of the present invention to form a device pattern.
Is transferred onto the substrate.

[0014]

According to the present invention, a plurality of shot areas on a substrate
First conversion parameter (A, O) fool for array at
Relative to the pattern in the shot area
Second conversion parameter (B) relating to an error in a proper imaging relationship
Use the results of these calculations.
To all the reference positions of multiple shot areas on the substrate
On the other hand, it is important to reduce the alignment error on average.
At the same time, chip patterns on those shot areas
Overlay error with the mask pattern image
The difference can be reduced on average. Further according to the invention, each shot area (27-n) on the substrate (8)
The conversion parameters (parameters in the conversion matrices A and O) relating to the arrangement of the reference positions (28-n) provided in, and the magnification error and rotation error of the chip pattern (projection pattern) with respect to the reference position (28-n) Conversion parameters (parameters in the conversion matrix B) are calculated. Then, after correcting at least one error such as a magnification error and a rotation error of the chip pattern based on a conversion parameter related to rotation and the like, each shot area (27-27) is corrected based on a conversion parameter (parameters in the conversion matrices A and O). n), the reference position (28-n) of each shot area (27-n) on the substrate (8) is correctly aligned, and the chip pattern and the projected image of the mask are aligned. Overlap exactly.

In this case, since the positions of at least seven alignment marks are measured when converted into marks indicating one-dimensional coordinates, the conventional six parameters (Θ, W, Rx, Ry) are used. , O _X , O _Y ), at least one parameter relating to the chip pattern (for example, a rotation error θ of the chip rotation, an orthogonality error w of the chip pattern, a chip scaling rx in the x direction, or a chip scaling ry in the y direction. Are determined. Further, for example, when alignment marks indicating two-dimensional coordinates are used, the number of alignment marks to be measured is at least four, but one of the alignment marks is used. For the mark, it is only necessary to measure the coordinates in one direction among the two-dimensional coordinates.

The linear superposition error between the chip pattern in each shot area (27-n) and the projected image of the mask is, for example, four parameters (rotation error of the chip rotation) excluding the offset error. θ, the orthogonality error w of the chip pattern, the parameter rx in the x direction of chip scaling, and the parameter ry in the y direction of chip scaling. Of these four parameters, the orthogonality error w of the chip pattern may be ignored (w = 0), and the chip scaling may be isotropic (rx = ry = M) in some cases. . In this case, since the number of parameters to be obtained for the chip pattern is two, taking into account the offset,
It is sufficient that there is a positioning mark indicating one two-dimensional coordinate in addition to one two-dimensional positioning mark indicating two-dimensional coordinates on the substrate (8) in each shot area. In this case, in order to determine the values of the conventional six parameters and the two parameters related to the above-described chip pattern, two values are required.
In the case of a mark indicating dimensional coordinates, it is sufficient to measure the coordinates of at least four alignment marks on the substrate (8). However, it is needless to say that a positioning mark indicating one two-dimensional coordinate can be replaced with a positioning mark indicating two one-dimensional coordinates.

[0017]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of an alignment method according to the present invention. In this example, the present invention is applied to an alignment process of an exposure apparatus (stepper) that exposes a reticle pattern to each shot area on a wafer as a photosensitive substrate by a step-and-repeat method.

FIG. 2 shows the exposure apparatus of the present embodiment. In FIG. 2, the exposure light IL emitted from the illumination optical system 1 illuminates the reticle 2 with substantially uniform illuminance. The reticle 2 is held on a reticle stage 3, and the reticle stage 3 is supported so as to be able to move and minutely rotate in a two-dimensional plane on a base 4. A main control system 6 that controls the operation of the entire apparatus controls the operation of the reticle stage 3 via the drive device 5 on the base 4.

Under the exposure light IL, the pattern image of the reticle 2 is projected onto each shot area on the wafer 8 via the projection optical system 7. The wafer 8 is placed on a wafer stage 10 via a wafer holder 9. The wafer stage 10 holds the wafer 8 in a plane perpendicular to the optical axis of the projection optical system 7.
XY stage for two-dimensionally positioning the projection optical system 7
A Z stage for positioning the wafer 8 in a direction parallel to the optical axis (Z direction), a stage for slightly rotating the wafer 8, and the like.

The movable mirror 1 is provided on the upper surface of the wafer stage 10.
1 is fixed, and a laser interferometer 12 is arranged so as to face the movable mirror 11. Although shown in a simplified manner in FIG. 2, the movable mirror 11 is a plane mirror having a reflecting surface perpendicular to the X axis, with an orthogonal coordinate system in a plane perpendicular to the optical axis of the projection optical system 7 as the X axis and the Y axis. And a plane mirror having a reflecting surface perpendicular to the Y axis. In addition, laser interferometer 1
Reference numeral 2 denotes two X-axis laser interferometers that irradiate the movable mirror 11 with a laser beam along the X-axis and a Y-axis laser interferometer that irradiates the movable mirror 11 with a laser beam along the Y-axis. The X coordinate and the Y coordinate of the wafer stage 10 are measured by one laser interferometer for the X axis and one laser interferometer for the Y axis. Coordinate system (X, Y) composed of X coordinate and Y coordinate measured in this way
Is hereinafter referred to as a stage coordinate system or a stationary coordinate system.

The rotation angle of the wafer stage 10 is measured based on the difference between the values measured by the two X-axis laser interferometers. X coordinate, Y measured by laser interferometer 12
Information on the coordinates and the rotation angle is supplied to the coordinate measuring circuit 12a and the main control system 6, and the main control system 6 monitors the supplied coordinates and drives the wafer stage 10 via the driving device 13.
Control the positioning operation. Although not shown in FIG. 2, the same interferometer system as the wafer side is provided on the reticle side.

The projection optical system 7 of this embodiment is provided with an imaging characteristic control device 14. The imaging characteristic control device 14 adjusts, for example, the distance between predetermined lens groups of the lens groups constituting the projection optical system 7 or the pressure of gas in the lens chamber between the predetermined lens groups. Thus, the projection magnification and distortion of the projection optical system 7 are adjusted. The operation of the imaging characteristic control device 14 is also controlled by the main control system 6.

In this embodiment, an off-axis alignment system 15 is arranged on the side surface of the projection optical system 7, and in this alignment system 15, illumination light from a light source 16 is collimated by a collimator lens 17, a beam splitter 18, a mirror 19 and an objective. The light is irradiated to the vicinity of the alignment mark 29 on the wafer 8 via the lens 20. In this case, the objective lens 2
The baseline amount, which is the distance between the 0 optical axis 20a and the optical axis 7a of the projection optical system 7, is measured in advance. Then, the reflected light from the alignment mark 29 is
Mirror 19, beam splitter 18, and condenser lens 2
The light is radiated onto the index plate 22 via 1, and an image of the alignment mark 29 is formed on the index plate 22.

The light transmitted through the index plate 22 is directed to the beam splitter 24 via the first relay lens 23, and the light transmitted through the beam splitter 24 is converted by the second X-axis relay lens 25X into the X-axis of a two-dimensional CCD. Focused on the imaging surface of the imaging element 26X for the beam splitter 2
The light reflected by 4 is focused by the second relay lens 25Y for Y-axis on the imaging surface of the imaging element 26Y for Y-axis formed of a two-dimensional CCD. On the imaging surfaces of the imaging elements 26X and 26Y, an image of the alignment mark 19 and an image of the index mark on the index plate 22 are formed so as to overlap each other. The imaging signals of the imaging devices 26X and 26Y are both coordinate position measurement circuits 1
2a.

FIG. 3 shows a pattern on the index plate 22 shown in FIG. 2. In FIG. 3, an image 29P of a cross-shaped alignment mark 29 is formed at the center, and an orthogonal linear pattern image 29XP of this image 29P is formed. And XP perpendicular to 29YP
Direction and YP direction are respectively the wafer stage 1 of FIG.
0 is conjugate to the X and Y directions of the stage coordinate system. Then, two index marks 31A and 31B are formed so as to sandwich the alignment mark image 29P in the XP direction, and two index marks 32A and 32B are formed so as to sandwich the alignment mark image 29P in the YP direction.

In this case, in the XP direction, the index marks 31A,
Detection area 33 surrounding 31B and linear pattern image 29XP
The image in X is picked up by the X-axis image sensor 26X in FIG.
An image in the detection area 33Y surrounding the index marks 32A and 32B and the linear pattern image 29YP in the P direction is picked up by the Y-axis image sensor 26Y in FIG. Furthermore, the scanning direction when reading the photoelectric conversion signal from each pixel of the imaging devices 26X and 26Y is set to the XP direction and the YP direction, respectively, and by processing the imaging signals of the imaging devices 26X and 26Y,
Alignment mark image 29P and index marks 31A, 31
It is possible to determine the amount of positional deviation between B and 32A, 32B in the XP direction and the YP direction. Therefore, in FIG.
The coordinate measuring circuit 12a determines the stage coordinate system (X) of the alignment mark 29 from the positional relationship between the image of the alignment mark 29 on the wafer 8 and the index mark on the index plate 22 and the measurement result of the laser interferometer 12 at that time. , Y), and supplies the measured coordinate values to the main control system 6.

Next, a description will be given of the operation of the present embodiment in which each shot area on the wafer 8 is aligned with the pattern image of the reticle 2 to expose each shot area. FIG. 4A shows a wafer 8 of the present embodiment. In FIG. 4A, an orthogonal coordinate system (α,
along a plurality of shot areas 27-n (n = 0,
1, 2,...) Are arranged in a matrix, and a chip pattern is formed in each shot region 27-n by exposure and development in the previous process. In FIG. 4, five shot areas 27-1 to 27 out of a plurality of shot areas are illustrated.
Only 27-5 is shown as a representative.

A reference position is defined for each shot area 27-n. For example, assuming that the reference position is a reference point 28-n at the center of each shot area 27-n, the design coordinate values of the reference point 28-n in the coordinate system (α, β) on the wafer 8 are respectively ( C _Xn , C _Yn ). Each shot area 27-n has four alignment marks 29-
n, 30-n, 34-n, 35-n are additionally provided. In this case, each shot area 27-n is placed in parallel with the coordinate system (α, β) on the wafer shown in FIG.
As shown in (b), the coordinate system (x, y) on the shot area
Is set, the alignment marks 29-n, 30-
n, 34-n, 35- n coordinate system (x, y) on each coordinates of the design in the _{(S 1Xn, S 1Yn),} (S 2Xn, S
_2Yn ), ( _S3Xn , _S3Yn ) and ( _S4Xn , _S4Yn ).

Referring back to FIG. 4A, the wafer 8 is placed on the wafer stage 10 shown in FIG. 2, and a reticle is projected onto each of a plurality of shot areas where a chip pattern has already been formed by the step-and-repeat method. Exposure is performed by superimposing images sequentially. At this time, the correspondence between the stage coordinate system (X, Y) defining the movement position of the wafer stage 10 and the coordinate system (α, β) of the wafer is not always the same as the relationship in the previous process. Therefore, the stage coordinate system (X, C _Yn ) is calculated from the design coordinate values (C _Xn , C _Yn ) of the reference point 28-n of each shot area 27-n with respect to the coordinate system (α, β).
Y) Even if the coordinates on the upper side are obtained and the wafer is moved based on the coordinates, the shot areas 27-n may not be precisely aligned. Therefore, in the present embodiment, it is assumed that the positioning error is caused by the following four factors as in the conventional example.

Wafer rotation: This is represented by a residual rotation error の in the wafer coordinate system (α, β) with respect to the stage coordinate system (X, Y). The orthogonality of the stage coordinate system (X, Y): This is caused by the fact that the feed of the wafer stage 10 in the X-axis direction and the Y-axis direction is not exactly orthogonal, and is represented by an orthogonality error W.

Linear expansion and contraction (wafer scaling) in the α and β directions in the coordinate system (α, β) of the wafer: This means that the entire wafer 8 expands and contracts due to a processing process or the like. This amount of expansion and contraction is represented by wafer scaling Rx and Ry in the α and β directions, respectively. Here, the wafer scaling Rx in the α direction is the ratio between the measured value of the distance between two points in the α direction on the wafer 8 and the design value, and the wafer scaling Ry in the β direction is the measured value between the two points in the β direction and the design value. Shall be expressed as the ratio of The stage coordinate system (X, β) of the coordinate system (α, β) on the wafer
Offset with respect to Y): This is caused by the entire shift of the wafer 8 with respect to the wafer stage 10 by a small amount, and is represented by offset amounts O _X and O _Y.

When the above error factors are added,
With respect to the shot area whose design coordinate value of the reference point is (C _Xn , C _Yn ), coordinates (C ′ _Xn , C ′ _Yn ) on the stage coordinate system (X, Y) to be positioned for actual exposure.
Is expressed as follows.

[0033]

(Equation 2)

Here, the orthogonality error W and the residual rotation error Θ
Is a very small amount and the first approximation is performed, (Equation 2) becomes as follows.

[0035]

(Equation 3)

Up to this point, it has been described that the reference position on each shot area 27-n (in this example, the reference point at the center of each shot area) is accurately aligned. However, just because the reference points of each shot area are correctly aligned, the entire chip pattern in each shot area and the projected image of the reticle do not always overlap exactly.

Next, the overlay error in each shot area will be described. As described above, FIG.
In any of the shot area 27-n on the coordinate system (x, y) coordinate values of the design on the _{(S 1Xn, S 1Yn) ~} (S
_4Xn , S _4Yn )
9-n, 30-n, 34-n and 35-n are formed. In this example, it is assumed that the overlay error in each shot area is caused by the following factors.

Rotation of chip pattern (chip rotation): This is, for example, when the projection image of the reticle 2 is exposed on the wafer 8, the reticle 2 is rotated with respect to the stage coordinate system (X, Y). Alternatively, this occurs when yawing is mixed in the movement of the wafer stage 10, and the coordinate system (x,
It is represented by a rotation error θ with respect to y). Chip orthogonality error: This is, for example, when a projection image of the reticle 2 is exposed on the wafer 8, this is caused by distortion of the pattern itself on the reticle 2 or distortion (distortion) of the projection optical system 7. This is an error of the orthogonality, and is represented by an angle error w.

Linear expansion and contraction of the chip (chip scaling): This is caused, for example, by an error in the projection magnification when the projection image of the reticle 2 is exposed on the wafer 8 or by the wafer 8 processing process. This is caused by expansion and contraction. Here, the chip scaling rx in the x direction, which is the ratio of the measured value of the distance between two points in the x direction of the coordinate system (x, y) of the shot area to the design value, and the distance between the two points in the y direction It is assumed that linear scaling in two directions is represented by chip scaling ry in the y direction, which is the ratio between the measured value and the design value.

For example, FIG. 5A shows a wafer 8A in which a rotation error and a magnification error have occurred in the chip pattern of each shot area 27-n formed in the previous step.
In (a), an example of a shot area in the case where there is no rotation error and magnification error is surrounded by a shot area 36-6 to a dotted line.
Expressed as 36-10. On the other hand, the shot areas 27-6 to 27-10 actually formed on the wafer 8A have different rotation angles and magnifications. These errors are shown in FIG.
As shown in FIG. 5B, the chip rotation error in which the shot area 27-n is inclined with respect to the original shot area 36-n, and the magnification of the shot area 27-n as shown in FIG. Can be separated into a chip scaling error different from the original magnification of the shot area 36-n.

However, in the example of FIG. 5, there is no orthogonality error w of the chip pattern and the chip scaling rx in the x direction.
And the chip scaling ry in the y-direction are equal. When the above error factors are added, the design coordinate values on the shot area 27-n become (S _NXn , S
_NYn ) (N = 1 to 4) alignment marks 29-n, 3
With respect to 0-n, 34-n, and 35-n, the coordinate values (S ′ _NXn , S ′ _NYn ) of the shot area to be actually aligned on the coordinate system (x, y) are expressed as follows. You.

[0042]

(Equation 4)

Here, when the first-order approximation is performed on the assumption that the orthogonality error w and the rotation error θ are minute amounts, (Equation 4) is expressed by the following equation.

[0044]

(Equation 5)

In FIG. 4B, the array coordinate value of the reference point 28-n of an arbitrary shot area 27-n on the stage coordinate system (X, Y) is (C _Xn , C _Yn ). Therefore, the stage coordinate system (X, X) of any alignment mark (29-n or 30-n) on that arbitrary shot area
The design coordinate values (D _NXn , D _NYn ) on Y) are expressed as follows. However, as described above, the alignment marks 29-n to 35-n are distinguished by the value of N (1 to 4).

[0046]

(Equation 6)

The above three errors (1) to (4) occur when a chip pattern is printed on a layer on which alignment marks of each shot area on the wafer 8 are printed. In practice, the alignment mark 29 is further affected by the above-mentioned error caused by the processing process of the wafer 8.
-N, 30-n, 34-n, and 35-n are coordinates (F _NXn , F _NXn , F
_NYn ) (N = 1 to 4), this coordinate value (F _NXn , F
_NYn ) is expressed as follows from ( _Equation 3) and ( _Equation 5).

[0048]

(Equation 7)

Next, in this embodiment, in order to facilitate the application of the least squares method, the wafer scaling Rx in the α direction and the wafer scaling Ry in the β direction in (Equation 7) are respectively set to new parameters Γx and It is represented as the following (Equation 8) using Γy. Similarly, the chip scaling rx in the x direction and the chip scaling ry in the y direction in (Equation 7) are set to new parameters γx and γ, respectively.
It is expressed as the following (Equation 8) using y.

[0050]

(Equation 8)

By rewriting (Expression 7) using these new four parameters Γx, Γy, γx, and γy indicating the changes in the linear expansion and contraction, (Expression 7) becomes approximately as follows: Become.

[0052]

(Equation 9)

In Equation (9), when the two-dimensional vector is regarded as a matrix of 2 rows × 1 column, Equation (9) can be rewritten into a coordinate conversion equation using the following conversion matrix.

[0054]

(Equation 10)

However, each transformation matrix of (Equation 10) is defined as follows.

[0056]

[Equation 11]

That is, in (Equation 10), the matrix F _Nn of 2 rows × 1 column is represented by the addition of the matrix AC _n , the matrix BS _Nn and the matrix O. Ten error parameters (Θ, W, Γx (= Rx−1), Γy, O _X , O _Y , θ, θ) included in the transformation matrices A, B, and O in the coordinate transformation formula of (Equation 10)
w, γx (= rx−1), γy) can be obtained by, for example, the least square method. In this example, the coordinates of each shot area of the wafer 8 on the stage coordinate system (X, Y) and the error of each chip are obtained based on the coordinate transformation formula of (Equation 10). Then, after correcting the chip rotation error and the chip magnification error based on that,
Each shot area of the wafer 8 is aligned with the reticle. Note that it is not always necessary to apply the least squares method to (Equation 10). For example, ten error parameters may be obtained at the stage of (Equation 7).

In this example, as will be described below, the coordinate position of the selected alignment mark is further measured in a shot area (sample shot) previously selected from the shot areas on the wafer. To (Equation 1
0), ten parameters are obtained by the least squares method, and the array coordinates of each shot area are calculated based on these parameters. The calculation for obtaining a predetermined (10 or less) error parameter from the measurement result of the alignment mark (sample shot) selected in advance in this manner is performed by an enhanced global alignment (hereinafter, referred to as an enhanced global alignment).
A method of performing positioning based on the result is called an EGA type alignment.

Next, referring to the flowchart of FIG.
An example of the alignment operation and the exposure operation of this example based on the coordinate conversion formula of (Equation 10) will be described. First, in step 101 of FIG. 1, the wafer 8 to be exposed this time is loaded onto the wafer holder 9 of FIG. A chip pattern has already been formed in each shot area of the wafer 8 in the previous step. Further, FIG.
As shown in FIG. 3, four cross-shaped alignment marks 29-n are provided in each shot area 27-n on the wafer 8.
n, 30-n, 34-n and 35-n are formed. Also, alignment of reticle 2 has been completed,
The amounts of displacement of the reticle 2 in the X, Y, and rotational directions with respect to the rectangular coordinates defined by an interferometer (not shown) are substantially zero.

Next, in step 102 of FIG. 1, the origin of the wafer 8 is set (pre-alignment). Thereafter, in step 103, five or more alignment marks (29-n, 30-n, 34-n or 3-n) are formed on the wafer 8 using the off-axis alignment system 15 shown in FIG.
5-n) on the stage coordinate system (X, Y).
M _NXn , FM _NYn ). Since one alignment mark has two components in the X and Y directions, the values of ten or more parameters can be determined by actually measuring the coordinate values of five or more alignment marks.

The alignment marks actually measured in this example are 3
It is necessary to select one or more shot areas 27-n, but it is not necessary to select four alignment marks 29-n to 35-n from one shot area 27-n. One alignment mark (29-n, 30-n, 34-n or 35-n) may be selected from each of 27-n.

In this case, the design coordinate values (C) of the reference points 28-n of the plurality of shot areas 27-n selected on the wafer 8 on the coordinate system (α, β) on the wafer 8 are set. _Xn ,
C _Yn ) and the design coordinate values (relative coordinate values) (S _NXn , S _NYn ) of the measured alignment mark in the coordinate system (x, y) on each shot area 27-n are known in advance. I have. Therefore, in step 104, on the right side of (Equation 10 ), the design array coordinate values (C _Xn , C _Yn ) of the reference point of the shot area to which the measured alignment mark belongs, and the design related to the reference point of the alignment mark By substituting the above relative coordinate values (S _NXn , S _NYn ), calculated coordinate values (F _NXn , F _NYn ) whose alignment marks should be on the stage coordinate system (X, Y) are _obtained .

Then, ten error parameters (Θ, W, Γx, Γy,
_{O X, O Y, θ,} w, γx, γy) Request. In particular,
The difference (E _NXn , E _NYn ) between the actually measured coordinate value (FM _NXn , FM _NYn ) and its calculated coordinate value (F _NXn , F _NYn ) is considered as an alignment error. Therefore, E _NXn = FM _NXn -F
_NXn , E _NYn = FM _NYn -F _NYn . Then, five or more sets of alignment errors (E _NXn , E _NYn ), that is, the sum of the squares of the ten or more alignment errors, are sequentially partially differentiated with these ten error parameters, and the values are each 0.
By setting an equation as follows and solving these ten simultaneous equations, ten error parameters can be obtained. This is the EGA calculation of the present example.

Thereafter, in step 105, (Equation 1)
In order to correct the rotation error θ of the chip rotation in the conversion matrix B of (0), the reticle 2 is appropriately rotated through the reticle stage 3 in FIG. The rotation of the chip pattern with respect to the system (X, Y) is corrected. This is (Equation 10 )
Means that the reticle 2 or the wafer 8 is rotated in accordance with the rotation error θ constituting the element of the transformation matrix B shown by.

However, when the wafer 8 is rotated, the offset error (O _X , O _Y ) of the wafer 8 may be changed. It is necessary to calculate the error parameter again by performing the EGA calculation of Therefore, for example, the wafer 8 is set at the angle θ.
When the wafer 8 is rotated only by at least three times, at least 3
The coordinate values of the alignment marks of the shot areas in the stage coordinate system (X, Y) are measured again. Then, from the result, six error parameters (Ｗ, W, Rx, Ry, O
_X , O _Y ) are determined, and each shot area is aligned based on the array coordinates calculated from the result to perform exposure.

Next, the orthogonality error w of the chip cannot be corrected in a strict sense, but by rotating the reticle 2 appropriately, the error can be reduced. Therefore,
It is also possible to optimize the rotation amount of the reticle 2 or the wafer 8 so that the sum of the absolute values of the rotation error 誤差, the rotation error θ, and the orthogonality error w is minimized.

Next, in step 106, (Equation 1)
The projection magnification of the projection optical system 7 is adjusted via the imaging characteristic control device 14 of FIG. 2 so as to correct the chip scaling error in the conversion matrix B of 0). This means that the projection magnification of the projection optical system 7 is adjusted in accordance with the chip scalings rx and ry constituting the elements of the transformation matrix B shown in (Equation 9).

Thereafter, in step 107 of FIG.
Using the conversion matrices A and O including the elements consisting of the error parameters obtained in step 104, the design array coordinate value (C _Xn) of the reference point 28-n of each shot area 27-n on the wafer 8 is calculated by the following equation. , C _Yn ), a calculated array coordinate value (G _Xn , G _Yn ) of the reference point 28-n on the stage coordinate system (X, Y) is obtained. However, as described above, when the wafer 8 is rotated to correct the rotation error in step 105, each of the reference points 28-n is determined based on the coordinates of the alignment mark measured again.
Of the array coordinates (G _Xn , G _Yn ) on the stage coordinate system (X, Y).

[0069]

(Equation 12)

Then, in step 108, based on the array coordinates (G _Xn , G _Yn ) obtained by the calculation and the previously obtained baseline amount, the reference point 28-n of each shot area 27-n on the wafer 8 is determined. n is sequentially aligned with a predetermined position in the exposure field of the projection optical system 7 in FIG.
The pattern image of the reticle 2 is projected and exposed on the shot area 27-n. After the exposure of all shot areas on the wafer 8 is completed, processing such as development of the wafer 8 is performed.

In this case, as shown in (Equation 10) in this example, not only the transformation matrices A and O but also the transformation matrix B composed of chip rotation, chip orthogonality error, and chip scaling parameters are considered. Therefore, the effects of the expansion and contraction and rotation of the chip pattern itself transferred to each shot area can be minimized, and the chip pattern of each shot area on the wafer and the projected image of the reticle pattern can be superimposed with higher accuracy. it can.

In the above embodiment, as shown in FIG. 4B, a cross-shaped alignment mark 29-n that can be simultaneously positioned in the X and Y directions on the stage coordinate system.
~ 35-n on the x-axis on the coordinate system of the shot area,
And two on the y-axis. However, for example, 1
If four alignment marks are not arranged on a straight line, such an arrangement is not necessarily required. For example, alignment marks may be arranged at four corners of each shot area. Further, each alignment mark may be formed on a street line area between each shot area. Even if a one-dimensional alignment mark for the X direction and a one-dimensional alignment mark for the Y direction are separately provided, the conversion matrix A , B and O can be obtained.

However, when a one-dimensional alignment mark (diffraction grating mark or the like) that can detect only the position in the X or Y direction is used instead of the cross mark that can specify the two-dimensional position as in this example. In order to determine the values of ten parameters, it is necessary to actually measure the coordinate values of ten or more one-dimensional alignment marks. In the above embodiment, in order to obtain the rotation error θ of the chip rotation, the orthogonality error w of the chip, and the chip scaling rx, ry, 4 shots are set in each shot area 27-n of the wafer 8.
The two-dimensional alignment marks 29-n to 35-n are provided. However, even if the offset (x-direction and y-direction) of the reference point of each shot area 27-n is taken into consideration, there are six parameters to be obtained, so each shot area 27-n has three two-dimensional Only alignment marks (for example, 29-n, 30-n, and 34-n) may be provided. Thus, when using a two-dimensional alignment mark, two alignment marks are always selected. However, in the case of a one-dimensional alignment mark, it is necessary to form six alignment marks in each shot area 27-n.

In the above embodiment, the reticle (or wafer) is formed by using four parameters relating to the chip pattern (rotation error θ of chip rotation, orthogonality error w, chip scaling rx, chip scaling ry).
And correction of the magnification of the projection optical system. However, it is not always necessary to correct the rotation of the reticle (or wafer) and the magnification of the projection optical system, and it is only necessary to align each shot area in accordance with the array coordinate values obtained in the above embodiment. At this time, magnification correction using, for example, scaling parameters Rx and Ry (or Δx and Δy) for the wafer may or may not be performed.

Further, in the above embodiment, four parameters relating to the chip pattern (rotation error θ of chip rotation, orthogonality error w, chip scaling rx (= 1 +
γx) and chip scaling ry (= 1 + γy)) are all obtained, but (Equation 7) (or (Equation 9)) may be used focusing on only one of these four parameters. Specifically, when focusing only on the rotation error θ,
The orthogonality error w is assumed to be 0, and the chip scalings rx and ry are each assumed to be 1 and (Equation 7) is used. In this regard, chip scaling rx and ry
When attention is focused on, it is sufficient to select rx = ry, that is, assume that the linear expansion and contraction is isotropic, and use (Equation 7) to select according to the type (characteristic) of C.

Further, four parameters relating to the chip pattern (rotation error θ of chip rotation, orthogonality error w, chip scaling rx, chip scaling r
y), the orthogonality error w is regarded as 0, and the chip scaling is regarded as isotropic (that is, rx = ry =
M) and (Equation 7) are simplified as follows.

[0077]

(Equation 13)

If the isotropic chip scaling M is (1 + δM) and (Equation 8) is used and the second-order minute amount is ignored, (Equation 13) can be approximated as follows.

[0079]

[Equation 14]

In this (Equation 14), since there are two parameters relating to the chip pattern, each shot area 27-n of the wafer 8 has two parameters shown in FIG. Dimensional alignment mark 29-
It is only necessary to provide n and 30-n. Since the error parameters are eight in total, the least squares method is
In order to determine the values of these error parameters by applying the method described above, the number of two-dimensional alignment marks to be measured should be at least four.

Next, an example of an alignment mark that can be actually used will be described with reference to FIG. First, as an alignment mark (two-dimensional mark) indicating two-dimensional coordinates, in addition to the cross-shaped alignment mark 29 used in the above-described embodiment (this is also shown in FIG. , T-shaped, or C-shaped marks. Further, when an alignment system of a so-called two-beam interference system or an alignment system of a laser step alignment system is used, the two-dimensional grating pattern 41 as shown in FIG.
It becomes a dimension mark. When an alignment system of a laser step alignment system or an alignment system of an imaging system as in the above embodiment is used, as shown in FIG. The alignment mark 43 in which the line and space patterns 42Y in the Y direction are arranged in parallel is also a two-dimensional mark.

The selection of one such two-dimensional mark is based on the fact that 10 (or less) parameters in (Equation 10) can be obtained by using the X-coordinate and Two data for the Y coordinate are obtained.
Therefore, selecting one cross-shaped alignment mark (for example, 29-1) in the above embodiment is equivalent to selecting an alignment mark (one-dimensional mark) indicating two one-dimensional coordinates. However, even when a two-dimensional mark is selected, only one coordinate data of the X coordinate or the Y coordinate may be used.

As one of the one-dimensional marks indicating the coordinates in the X direction, as shown in FIG. 7A, a line and space pattern (or diffraction grating) arranged at a predetermined pitch in the X direction is used. Mark) 42X, and as a mark indicating the coordinate in the Y direction, as shown in FIG.
There is a line and space pattern (or diffraction grating mark) 42Y arranged at a predetermined pitch in the Y direction. Further, the two-dimensional grid pattern 41 in FIG. 6B can be regarded as a one-dimensional mark in the X direction or a one-dimensional mark in the Y direction.

Next, another example of a method of selecting a sample shot and an alignment mark to be measured will be described in detail. In this case, it is premised that the exposure apparatus performs the EGA calculation using the measurement result of the alignment mark formed in the exposure step of the previous layer. Therefore, the design position and the number of alignment marks attached to each shot area on the wafer are the same in all shot areas.

First, in a conventional alignment method of the EGA system which does not consider an error in a chip pattern,
There are at least 6 error parameters (Θ, W, Rx, Ry, O _X , O _Y ), so that at least 6
It is necessary to measure the coordinates of each alignment mark.
That is, the X coordinate and the Y coordinate at each of at least three measurement points that are not on the same straight line on the wafer may be measured. At this time, the X coordinate and the Y coordinate of one measurement point do not need to measure the one-dimensional mark 42 in the X direction and the one-dimensional mark in the Y direction in the same shot, respectively, and are attached to one of two different shot areas. The one-dimensional mark in the X direction and the one-dimensional mark in the Y direction attached to the other shot area may be individually measured, and the measurement results may be used.

FIG. 8 shows a method for selecting such an alignment mark. In FIG. 8, each of the shot areas 44A, 44B,... On the wafer 8 has a one-dimensional mark (hereinafter, referred to as "X" coordinate). X mark) 42
XA, 42XB,... And one-dimensional marks (hereinafter referred to as “Y marks”) 42YA, 42YB,
… Is attached. Therefore, when measuring the coordinates of, for example, six alignment marks, first, as shown in FIG. 8A, three shot areas 44A, 44B and
Y marks 42YA, 42YB and 42 attached to 4C
The Y coordinate of YC is measured, and three parameters (Θ, Ry, O _Y ) are obtained from the measurement result. Thereafter, as shown in FIG. 8B, the three shot areas 44A, 44B and 4
X marks 42XA, 42XB and 42 attached to 4D
The Y coordinate of the XD may be measured, and three parameters (W, Rx, O _X ) may be calculated from the result. In this case, the shot area 44D used in FIG. 8B is different from the shot area 44C used in FIG. 8A, but there is no problem.

It should be noted that the X mark and Y
By measuring the mark, the X coordinate and the Y coordinate at a predetermined measurement point may be calculated in the calculation, and the EGA operation may be performed using the coordinates. Next, from among the four error parameters (rotation error θ of chip rotation, orthogonality error w, chip scaling rx, chip scaling ry) introduced in the above-described embodiment, in addition to the conventional six error parameters, A method of selecting an alignment mark when obtaining the value of the selected predetermined error parameter will be described for the following cases with reference to FIG.

(1) When only chip scaling ry (or γy) in the y direction is calculated: In this case, since seven error parameters need to be obtained, at least three X marks and four Y marks must be obtained. Coordinates need to be measured. At this time, as shown in FIG.
X mark 42X and one Y mark 42Y,
One Y mark 46 having a different Y coordinate from the Y mark 42Y
Y must be attached. The selection condition of the Y mark used for the EGA calculation is to select one Y mark (for example, 46Y) that is not on the straight line 45 that passes through the selected one Y mark 42Y and is parallel to the X axis. is there. However, the two Y marks that satisfy the condition need not exist on the same shot area.

Also, the mark arrangement as shown in FIG. 9A is not usually employed, and as shown in FIG.
The two-dimensional marks 47 and 48 may be provided. In this case, the four Y marks are sorted and selected as the Y coordinate of the two-dimensional mark 47 in the first shot area and the Y coordinate of the two-dimensional mark 48 in the same or another shot area.

(2) When calculating only the rotation error θ of the chip rotation: In this case, in addition to at least three X marks and three Y marks, the position of at least one X mark or Y mark is measured. There is a need. Last X
As shown in FIG. 9C, the mark selection condition is to select an X mark 51X that is not on a straight line 50 that passes through the already selected X mark 42X and is parallel to the X axis. Also, X
When selecting a Y mark instead of a mark, it is necessary to select a Y mark 51Y that is not on the straight line 49 that passes through the already selected Y mark 42Y and is parallel to the Y axis. Even in this case, two X marks (or Y marks) satisfying the above conditions need not be present in the same shot area 44.

(3) When calculating the chip scaling ry in the y direction and the rotation error θ of the chip rotation: In this case, in addition to at least three X marks and three Y marks, two Y marks (of which (One can be X mark)
You need to choose. Further, in addition to satisfying the above conditions (1) and (2), it is necessary to select three marks that are not on the same straight line. (4) When calculating only the chip scaling rx (or γx) in the x direction: In this case, it is necessary to measure the coordinates of at least four X marks and three Y marks. above
As in the case of (1), the selection condition of the four X marks is Y
Choosing two points that are not parallel to the axis.

(5) When calculating only the orthogonality error w: In this case, one X mark, for example, is selected in addition to at least three X marks and three Y marks. Four X
The condition for selecting a mark is to select two points that are not parallel to the X axis, as in the case of (2) above. (6) Chip scaling rx and orthogonality error w in x direction
Is calculated: In this case, in addition to at least three X marks and three Y marks, two X marks are selected.
The selection condition of the four X marks is to select three points that are not on the same straight line, as in the above cases (3) and (4).

In the above description, the rotation error θ of the chip rotation may be calculated by using any coordinates of the Y mark and the X mark, and the orthogonality error w may be calculated by using any of the X mark and the Y mark. It may be calculated using coordinates. However, since the orthogonality error w is expressed in (Formula 7) (or (Formula 9)) of the above-described embodiment using the X mark, the above-described embodiment uses the X mark. Further, the above conditions for selecting an alignment mark are not limited to selection within the same shot area.

Further, for example, when it is assumed that the orthogonality error w of the chip is 0, the number of error parameters to be determined by (Equation 10) becomes nine. Therefore, the number of parameters determined in step 104 of FIG. 1 is nine, and in step 103 of FIG.
It is only necessary to measure the coordinate values of nine or more one-dimensional alignment marks that define the position in the direction.

Further, the orthogonality error w of the chip is regarded as 0, and the linear expansion and contraction of the chip is isotropic (rx
= Ry), two alignment marks (eg, 29-n and 30-n) are included in each shot area 27-n.
May be provided. In this case, since there are eight parameters to be determined, in step 103 of FIG. 1, the coordinate values of four or more two-dimensional alignment marks (or the coordinate values of eight or more one-dimensional alignment marks) are changed. Just measure it.

Conversely, more than four alignment marks may be provided in each shot area 27-n. In this case, the measurement error when the coordinate values of the alignment mark are actually measured is averaged, and when the chip pattern is first printed on the wafer (first layer), the alignment mark is distorted by the projection optical system. There is an advantage that the influence of the deviation from the design position is averaged.

In particular, unlike the case where an off-axis type alignment system is used as in the above-described embodiment, a TTL (through-the-lens) for directly observing or detecting an alignment mark via a projection optical system to perform alignment. In the case of using an alignment system of the type), by increasing the number of alignment marks, it is possible to reduce a measurement error of coordinate values due to distortion of the projection optical system. In addition, by increasing the number of alignment marks in each shot area in consideration of the arrangement, not only rotation and linear expansion and contraction of the chip pattern but also nonlinear distortion in the chip pattern can be obtained.

For example, by providing three alignment marks in each shot area along the X axis, nonlinear distortion in the X direction can be obtained. By increasing the number of alignment marks in this manner, so-called distortion components of the chip pattern (translation, rotation, magnification, orthogonality, trapezoidal distortion, barrel distortion, pincushion distortion, and the like of the chip pattern) are comprehensively detected. It is also possible. That is,
By increasing the number of alignment marks, the number of error parameters can be increased, and it is possible to more accurately grasp the state of error occurrence.

The alignment method in which the shot arrangement error in the wafer is linear as in the above-described embodiment belongs to the EGA method. Further, in the EGA method of the above-described embodiment, the chip rotation and the chip magnification error are determined assuming that the error of the chip rotation and the chip magnification (including distortion) for each shot area is constant within the same wafer. For this reason,
If the local arrangement error on the wafer or the fluctuation (non-linearity) of the distortion component is large, it is difficult to improve the overlay accuracy in the above embodiment. Therefore, there is a demand for an alignment method capable of satisfactorily correcting the chip rotation and the chip magnification error even for a wafer having non-linear distortion and performing highly accurate alignment. In the following, an embodiment will be described in which the above-described EGA method is improved to enable more accurate positioning. This alignment method is obtained by applying the alignment method proposed in Japanese Patent Application No. 4-297121 to the above embodiment.

FIG. 1 shows another embodiment of the present invention.
0 will be described. In this embodiment, the projection exposure apparatus shown in FIG. 2 is used. In this embodiment, the EG used in the first embodiment is used.
A first weighted enhanced global alignment method (hereinafter, referred to as a “W1-EGA method”) in which the alignment in the A method is further improved is performed. This W1-EGA type alignment is effective for “regular non-linear distortion”, and even if the substrate has “regular non-linear distortion”, the alignment error in a local region on the substrate is small. Almost the same. " In the W1-EGA alignment, weighting is performed according to the distance from the sample shot as described later.

FIG. 10 shows the wafer 8 to be exposed in this example. In FIG. 10, when determining the calculated coordinate position of the i-th shot area ESi on the wafer 8, the shot areas ESi and m Distances LK1 to LK between the sample shots (m = 9 in FIG. 10) SA1 to SA9
Depending on the 9, the weight W _in is given to each of the measured coordinate positions of the alignment marks in the nine sample shots (alignment data). Specifically, two alignment marks MA of sample shot SA1
1, the measured coordinate positions of the MB1 is the weight W _i1 corresponding to the distance LK1 is given. Note that, more strictly, it is desirable to assign weights according to the distance from the reference point of the shot area ESi to each alignment mark in each sample shot. In each sample shot, it is not always necessary to measure the coordinates of two alignment marks.

In this W1-EGA system, the EGA system
Instead of a mere sum of squares residual error component in
A residual error component Ei represented by (Equation 15) is defined. this
In (Equation 15), the coordinate value (FM_NXn , FM_NYn )
Is the Nth alignment in the nth sample shot
Coordinate value, coordinate value (F_NXn ,
F _NYn ) Are the calculated coordinate values.

[0103]

(Equation 15)

Then, ten error parameters (Θ, W, Γx, Γy, O _X , O _Y , θ, θ) satisfying (Equation 10) so that the residual error component Ei defined as described above is minimized.
w, γx, γy). Here, the sample shots SA1 to SA used for each shot area ESi are used.
9 is the same, but naturally the distance to each sample shot SAn differs for each shot area ESi. Therefore,
Weight W _in giving to the coordinate position (alignment data) of sample shot SAn varies for each shot area ESi. The error parameter for each shot area ESi (Θ, W, Rx, Ry, O X, O Y, θ, w, rx, ry)
Is determined, first, the wafer rotation error Θ in the conversion matrix A in (Equation 10) and the chip rotation error θ in the conversion matrix B are corrected, and the chip magnification error ( Chip scaling γx, γ
Correct y).

Thereafter, the error parameters (Θ, W, Γx,
Γy, O _X, O _Y) using the transformation matrix A and O containing more composed elements (number 12) to the shot area E on the wafer 8
By substituting the designed array coordinate values of the Si, the stage coordinate system (X,
Y) Obtain the calculated array coordinate value in the above. As described above, when the wafer is rotated, an array coordinate value is obtained by a normal EGA calculation based on the coordinates of the alignment mark measured anew.

As described above, in the W1-EGA method, the wafer 8
For each shot area ESi above, each sample shot S
The weight W _in changes to the An of the coordinate data. The weight W _in an example, i-th shot area ESi and n
It is expressed as follows as a function of the distance LKn from the sample shot SAn. However, the parameter S is a parameter for changing the degree of weighting.

[0107]

(Equation 16)

As is apparent from this (Equation 16), the i-th
Sample in which the distance LKn to the shot area ESi is short
The weight given to the alignment data as the shot SAn
Only W _inIs becoming larger. Also, (Equation 1)
In 6), if the value of the parameter S is sufficiently large,
The result of the calculation operation is obtained by the EGA method of the above embodiment.
It is almost equal to the result. On the other hand, the system to be exposed on the wafer
All the shot areas ESi are used as sample shots SAn,
When the value of the parameter S is sufficiently close to zero, each shot area
Performs alignment by measuring the position of the wafer mark for each area
Almost equal to the result obtained by the so-called die-by-die method
Become. That is, in the W1-EGA method, the parameter S
By setting it to the appropriate value, the EGA method and die-by
-An intermediate effect from the die method can be obtained. example
For example, for a wafer with a large nonlinear component, the parameter
By setting the value of S small, the die-by-die method
The same effect (alignment accuracy) as that of
And eliminates alignment errors due to non-linear components.
Can be In addition, the measurement of the alignment sensor
If the actuality is poor, increase the value of the parameter S.
As a result, it is possible to obtain almost the same effect as the EGA method.
And reduce the alignment error by the averaging effect
Can be.

[0109] In addition, also be able to set the weighting function of equation (16), prepared separately for X and Y coordinates of the alignment mark, the weight W _in independently by the X and Y coordinates Good. In this case, the degree (large or small) of the nonlinear distortion of the wafer, regularity or step pitch,
That is, even if the center-to-center distance between two adjacent shot areas (a value substantially corresponding to the shot size, depending on the width of the street line on the wafer) is different between the X direction and the Y direction, the value of the parameter S is not changed. By setting them independently, shot arrangement errors on the wafer can be corrected with high accuracy. At this time, the value of the parameter S may be different between the X coordinate and the Y coordinate as described above,
Further, regardless of whether the values of the parameter S of the X coordinate and the Y coordinate are the same or different, the value of the parameter S depends on the magnitude of the “regular non-linear distortion”, the regularity, the step pitch, or the measurement reproduction of the alignment sensor. What is necessary is just to change suitably according to sex etc.

As described above, the effect can be changed from the EGA system to the die-by-die system by appropriately changing the value of the parameter S. Therefore, alignment is performed for various layers and further for each component (X direction and Y direction) in accordance with, for example, characteristics of nonlinear components (for example, magnitude, regularity, etc.), step pitch, and the reproducibility of measurement of the alignment sensor. The alignment can be flexibly changed, and alignment can be performed under optimum conditions for each layer and each component.

Next, an alignment method of the second weighted enhanced global alignment method (hereinafter, referred to as “W2-EGA method”) will be described with reference to FIG. Here, for the sake of simplicity, it is assumed that non-linear distortion, particularly point-symmetric, occurs regularly in the wafer W, and the center of the point symmetry coincides with the center of the wafer W (wafer center).

FIG. 11 shows the wafer 8 to be exposed in this example. In FIG. 11, the deformation center point of the wafer 8 (the point symmetric center of the nonlinear distortion), that is, the wafer center Wc,
Let LEi be the distance (radius) between the i-th shot area ESi on the wafer 8 and the wafer center Wc and m (m = 9 in FIG. 11) sample shots SA1 to SA.
The distance (radius) between each of the light emitting elements 9 is LW1 to LW9. In this W2-EGA system, W1-EG
Like the A-system, in accordance with the distance LEi and distance LW1～LW9, giving a weight W _in 'to each of the measured coordinate positions of the alignment marks in the nine sample shots SA1～SA9 (alignment data). This W2-EG
In the A method, after detecting two alignment marks (MAi, MBi) for each sample shot, the residual error component Ei 'is defined by the following (Equation 17), as in (Equation 15). 17) so that the minimum error parameters (number 10) (Θ, W, Γx , Γy, O X, O Y, θ, w,
γx, γy) are determined.

[0113]

[Equation 17]

[0114] Similar to the W1-EGA method in this W2-EGA method, since the weight W _in providing the alignment data 'is to be changed for each shot area ESi, shot area E
A statistical operation is performed for each Si to obtain error parameters (Θ, W, Γ
_{x, Γy, O X, O} Y, θ, to determine w, γx, the γy),
The chip rotation, the chip orthogonality, the chip magnification error, and the calculated array coordinate value are determined.

Then, each shot area ES on the wafer W
for each i, 'to change the weights W _in the equation (17)' weight W _in for each sample shots, i th shot area ESi and the wafer center W on the wafer 8
The distance (radius) to c is expressed as a function of LEi as follows. However, the parameter S is a parameter for changing the degree of weighting.

[0116]

(Equation 18)

As is apparent from (Expression 18), the distance LWn from the sample shot SAn to the wafer center Wc is close to the distance LEi between the wafer center Wc and the i-th shot area ESi on the wafer W. degree, so that the weight W _in 'to be given to the alignment data becomes large. In other words, the largest weight W _in 'is given to the sample shots alignment data located on a circle of radius LEi around the wafer enter Wc, the weight W for alignment data with distance radially from the circle _in ′
Is getting smaller.

The value of the parameter S in (Equation 18) is determined by the alignment accuracy and the characteristics of nonlinear distortion (for example, large and small, regularity, etc.) required similarly to the W1-EGA method.
What is necessary is just to determine suitably according to step pitch, the quality of measurement reproducibility of an alignment sensor, etc. That is, when the nonlinear component is relatively large, the value of the parameter S is set to be smaller, so that the distance LWn from the wafer center Wc is set.
Can significantly reduce the effect of sample shots. On the other hand, when the nonlinear component is relatively small, by setting the value of the parameter S to be larger, it is possible to prevent a decrease in alignment accuracy in an alignment sensor (or layer) having poor measurement reproducibility.

Further, in the W2-EGA method, a plurality of shot areas which are substantially equidistant from the point symmetry center on the wafer W, that is, a plurality of shot areas located on the same circle centered on the point symmetry center. in each, the weight W _in 'is the same given to the course of the sample shot alignment data. For this reason, when a plurality of shot areas are located on the same circle centered on the point symmetry center, the weighting and the statistical calculation are performed on only one of the shot areas to obtain the error parameters (Θ, W). , Γx,
y, O _X , O _Y , θ, w, γx, γy), the chip rotation, chip orthogonality, chip magnification error, and coordinates of the remaining shot areas are directly used using the error parameters calculated previously. The position can be determined. This has the advantage that the amount of calculation is reduced.

By the way, it is desirable that the arrangement of sample shots suitable for the W2-EGA method be specified so as to be symmetric with respect to the point symmetric center of the nonlinear distortion, that is, the wafer center Wc. What is necessary is just to specify a character shape, a cross shape, etc. Besides that,
The arrangement may be the same as in the W1-EGA system. When the point of symmetry of the nonlinear distortion is other than the wafer center Wc, an X-shaped or cross-shaped arrangement based on the point of symmetry may be used. When determining the value of the error parameter, the weighting function shown in (Equation 18) may be set independently in each of the X direction and the Y direction.
It goes without saying that a weighting function may be set independently for each alignment mark.

The W1-EGA system and the W2-EGA
In the method, when correcting a rotation error or a magnification error based on four error parameters (θ, w, rx, ry) relating to a chip pattern, each of the shot areas is corrected using the parameters obtained for each shot area. Correction may be performed. Alternatively, one set of parameters obtained for each shot area may be averaged to obtain one set of parameters, and correction may be performed only once for the entire wafer based on these parameters. Further, the wafer may be divided into a plurality of blocks, and correction may be performed for each block. Further, in the W2-EGA method, in a shot area located on a concentric circle with respect to the center of point symmetry, it is not necessary to obtain a parameter for each shot area, and a parameter obtained for any one shot area is used. May be.

The present invention is applicable not only to a step-and-repeat type exposure apparatus (for example, a reduction projection type stepper or a 1: 1 projection type stepper), but also to a so-called step-and-scan exposure type exposure apparatus or proxy. It can be widely applied to a mitty type stepper (such as an X-ray exposure apparatus). In addition to the exposure device, this is a device (defect inspection device, prober, etc.) that inspects semiconductor wafers and reticles with multiple chip patterns. The present invention can be similarly applied to an apparatus for performing position alignment.

When the above-described alignment method is applied to a scanning type exposure apparatus such as an exposure apparatus of the step-and-scan exposure type, a predetermined offset ( The scanning exposure is performed after the wafer is positioned at a position to which a value uniquely determined according to the pattern size, the reticle, the running section of the wafer, and the like are added.

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

[0125]

According to the present invention, the alignment error is reduced on average with respect to all of the reference positions of the plurality of shot areas on the substrate to be processed, and at the same time, the chip pattern on all of the shot areas is reduced. As a result, the overlay error with the pattern image of the mask is reduced on average. Therefore, the influence of the expansion / contraction or rotation of the chip pattern itself transferred to each shot area can be suppressed to be small, and the chip pattern of each shot area on the substrate and the projection image of the mask pattern can be superimposed with higher accuracy. There are advantages. As a result, the number of non-defective chips obtained from one substrate to be processed increases, and the productivity of chips such as semiconductor elements can be improved.

In some shot areas on the substrate to be processed, the positions of the alignment marks belonging to the shot areas are actually measured, that is, the position measurement using alignment marks of the same shape is repeated a plurality of times. Therefore, there is also an advantage that a mechanical or electrical random error of the detection system is reduced. If two or more two-dimensional alignment marks are set in each shot area and the coordinate positions of at least four or more pre-selected alignment marks are measured, In addition to the array coordinates of the shot areas, two linear errors (such as chip rotation and isotropic chip scaling) of the chip pattern in each shot area can be obtained.

[Brief description of the drawings]

FIG. 1 is a flowchart showing an alignment operation and an exposure operation to which an embodiment of a positioning method according to the present invention is applied.

FIG. 2 is a configuration diagram illustrating an example of a projection exposure apparatus in which an alignment operation and an exposure operation of the embodiment are performed.

FIG. 3 is an enlarged view showing an image of an alignment mark on the index plate of FIG. 2;

FIG. 4A is a plan view showing an example of an arrangement of shot areas on a wafer according to the embodiment, and FIG. 4B is an enlarged plan view showing the shot areas in FIG. 4A.

FIG. 5A is a plan view showing an example of a wafer including a chip pattern rotation error and a chip magnification error, and FIG.
FIG. 7 is an explanatory diagram of a chip rotation error, and FIG. 7C is an explanatory diagram of a chip magnification error.

FIG. 6 is a diagram illustrating an example of an alignment mark indicating two-dimensional coordinates.

FIG. 7 is a diagram illustrating an example of an alignment mark indicating one-dimensional coordinates.

FIG. 8 is an explanatory diagram showing an example of selecting an alignment mark when performing alignment by a normal EGA method.

FIG. 9 is an explanatory diagram showing an example of selecting an alignment mark when performing alignment by the method of the embodiment.

FIG. 10 is a plan view showing a shot arrangement on a wafer on which W1-EGA alignment is performed in another embodiment of the present invention.

FIG. 11 is a plan view showing a shot array on a wafer on which W2-EGA alignment is performed in another embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 1 illumination optical system 2 reticle 6 main control system 7 projection optical system 8 wafer 10 wafer stage 12 laser interferometer 14 imaging characteristic control device 15 off-axis alignment system 27-1 to 27-5, 27-n shot area 28 -1 to 28-5, 28-n Reference point 29-1 to 29-3, 29-n Alignment mark 30-1 to 30-3, 30-n Alignment mark 34-1 to 34-3, 34-n Alignment Mark 35-1 to 35-3, 35-n Alignment mark

Claims (18)

(57) [Claims] In transferring a pattern on a mask onto a substrate, each of a plurality of shot regions two-dimensionally arranged on the substrate and provided with respective marks is moved to a moving position of the substrate. A method of aligning with respect to a predetermined transfer position in a stationary coordinate system defining the static coordinate system of a mark respectively attached to at least three sample shot areas of the plurality of shot areas on the substrate. a first step of measuring the position information on a first conversion parameter relating to the sequences on the substrate of the plurality of shot areas, to the error of the relative imaging relationship between the pattern in the shot area conversion
Parameters and at least the shot area
The base when the mask pattern is projected onto
Each of two or more shot areas on the plate in the two-dimensional array direction
Second conversion parameters and a plurality of Ma measured in the first step includes a linear expansion errors separately in the direction of
Network information including the first and second conversion parameters.
Statistical calculation using one coordinate transformation formula
A second method of calculating the array position information of the plurality of shot areas on the stationary coordinate system at the same time .
And a third step of aligning each of the plurality of shot areas on the substrate with respect to the transfer position based on the array position information calculated in the second step. How to align. 2. Each of the plurality of shot areas includes:
A plurality of first marks for measuring position information in a first direction on the stationary coordinate system and a plurality of second marks for measuring position information in a second direction orthogonal to the first direction are attached. The positioning method according to claim 1, wherein the positioning method is provided. 3. In the first step, a first mark provided along with a predetermined sample shot for measuring position information in a first direction on the stationary coordinate system, and a first mark orthogonal to the first direction. Measuring a second mark attached to a specific sample shot in order to measure position information in a second direction to perform the measurement, and among the plurality of first or second marks measured in the first step, 3. The alignment according to claim 1, wherein an arrangement relation of at least one mark to the sample shot area is different from an arrangement relation of another measured first or second mark to the sample shot area. Method. 4. The alignment method according to claim 1, wherein, in the first step, one mark attached to each sample shot area is measured for each of the sample shot areas. 5. The method according to claim 2, wherein in the first step, a plurality of pieces of position information in the first direction or a plurality of pieces of position information in the second direction are measured for at least one sample shot area. The alignment method described. 6. In the third step, before performing the positioning, correcting an error of a relative imaging relationship with the pattern in the shot area based on the second conversion parameter. The positioning method according to any one of claims 1 to 5, characterized in that: 7. The error of the imaging relationship includes a relative rotation error with respect to a pattern of the mask in the shot area, and in the third step, the mask and the substrate are relatively rotated. 7. The positioning method according to claim 6, wherein the rotation error is corrected by the correction. 8. When the rotation error is corrected by rotating the substrate in the third step, the first step and the second step are executed again, and the first conversion parameter and the first Newly calculating array position information, and in the subsequent third step, aligning each of the plurality of shot regions with respect to the transfer position based on the newly calculated array position information. The method according to claim 7, wherein 9. The image forming relation error includes an orthogonality error of the shot region, and in the third step, the mask is rotated with respect to the substrate to correct the orthogonality error. 7. The alignment method according to claim 6, wherein: 10. The error of the imaging relation is caused when the pattern of the mask is projected on the shot area.
Wherein the linear expansion error, by adjusting the projection magnification of the projection optical system, said linear
What claims 6-8, characterized in that to correct the expansion and contraction error
The alignment method according to any one of the preceding claims. Wherein said second conversion parameters according to claim 1, wherein the rotation of the shot area, which further comprises a parameter relating to at least one of the orthogonality
11. The alignment method according to claim 10. 12. The first conversion parameter includes a rotation of the substrate, an orthogonality of the stationary coordinate system, a linear expansion / contraction of an array coordinate system of the plurality of shot areas on the substrate, and a stationary state of the array coordinate system. The method according to claim 1, further comprising a parameter relating to at least one of offsets with respect to a coordinate system. 13. The method according to claim 12, wherein in the second step, rotation of the substrate is performed.
The orthogonality of the stationary coordinate system, and a conversion matrix A for linear expansion and contraction in the array coordinate system, and coordinate information C _n on the array coordinate system of the reference point in the shot area on the array coordinate system,
Using the transformation matrix O relating to the offset, the transformation matrix B relating to the second transformation parameter, and the design relative coordinate information S _Nn on the array coordinate system relating to the reference point of the mark, The coordinate information F _{Nn of the} position that should actually be on the stationary coordinate system is calculated by the following relational expression, and F _Nn = A · C _n + B · S _Nn + O Measurement of the measurement mark measured in the first step Value and coordinate information F _{Nn of the} measured mark calculated by the relational expression _.
The method according to claim 12, wherein the first and second conversion parameters are calculated based on the first and second parameters. 14. In the second step, the position information of the plurality of marks measured in the first step is weighted in accordance with the arrangement of the sample shot area, and the weighted plurality of position information is calculated. 14. The positioning method according to any one of claims 1 to 13, wherein the first and second conversion parameters are obtained by statistically calculating. 15. The alignment method according to claim 1, wherein the mark is a two-dimensional mark that can specify a two-dimensional position of the mark on the stationary coordinate system. . 16. Two or more two-dimensional marks are arranged in a fixed relative positional relationship with respect to a reference position in each of the shot areas in design, and in the first step, among the two-dimensional marks, 16. The positioning method according to claim 15, wherein position information of at least four marks selected in advance is measured. 17. A method of projecting a pattern of the mask via a projection optical system onto the plurality of shot areas aligned by using the alignment method according to claim 1. And exposing the pattern onto the substrate sequentially. 18. A device manufacturing method, comprising a step of transferring a device pattern onto the substrate by using the exposure method according to claim 17.