JPH0286117A - alignment device - Google Patents

Abstract

(57) [Abstract] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to an alignment technique, in particular, to measuring the positions of marks formed at multiple locations on an alignment target,
Regarding alignment technology that performs alignment work based on this measurement data, for example, in the manufacturing process of semiconductor devices, alignment (hereinafter referred to as alignment) is performed in a reduction projection exposure apparatus that superimposes and exposes circuit patterns on a wafer. ) Concerning effective technology that can be used in technology.

[Conventional technology]

As an alignment technique for a reduction projection exposure apparatus, for example, there is a technique described in Japanese Patent Laid-Open No. 61-44429.

In other words, in this alignment technology, the positions of marks formed at multiple locations on the wafer are measured, and the actual measurement data group is compared with the design data of the mark group, and the error between the actual measurement data and the design data is checked. Data with a value greater than a predetermined standard value will be deleted from the alignment data.
This deletion improves alignment accuracy.

[Problem to be solved by the invention]

However, in this alignment technology, in which actual measurement data and design data are compared to determine abnormalities in the actual measurement data, errors caused by wafer deformation included in the actual measurement data themselves are not considered, and the wafer Since only the measured data that has an error greater than the amount of deformation is deleted from the alignment data, there is a limit to the improvement in alignment accuracy.

An object of the present invention is to provide an alignment device that can also correct errors caused by deformation of the alignment target itself.

The above and other objects and novel features of the present invention will become apparent from the description of this specification and the accompanying drawings.

[Means to solve the problem]

An overview of typical inventions disclosed in this application is as follows.

That is, in an alignment apparatus configured to measure the positions of marks formed at multiple locations on an alignment target and perform alignment work based on this measurement data, measurement of the group of marks is performed. a deformation amount calculation unit that calculates the amount of deformation of the alignment target from the data; an error calculation unit that calculates an error for each mark based on the determined amount of deformation and measurement data for the mark group; The apparatus includes an exclusion section that excludes data having an error exceeding a standard value, and a correction calculation section that corrects the amount of deformation of the alignment target by using the remaining data group after exclusion. The present invention is characterized in that the alignment work is executed based on data for correction calculations.

[Effect]

According to the above-described means, the amount of deformation of the object to be aligned is determined, and this amount of deformation is subtracted from the actual measurement data, thereby correcting the deformation of the object to be aligned itself. Then, after this deformation correction, the error correction is performed again, so that the alignment accuracy is improved.

[Embodiment] Fig. 1 is a block diagram showing an alignment device used in a reduction projection exposure apparatus which is an embodiment of the present invention, Fig. 2 is a perspective view showing the reduction projection exposure apparatus, and Fig. 3 is a wafer A schematic plan view of a wafer showing the shot coordinate system in alignment, FIG. 4 is a schematic diagram showing shot offset, which is one of the causes of wafer error, and FIG. 5 is a schematic diagram showing shot expansion/contraction error. Figure 6 shows the same X shot.
Figure 7 is a schematic diagram showing the rotation of the axis, Figure 7 is a diagram showing the rotation of the shot on the Y axis, Figure 8 is a diagram showing the rotation error of the shot, and Figure 9 is a diagram showing the magnification error of the shot. 10 is a plan view of a wafer, FIG. 11 is a flow chart showing a wafer alignment sequence according to an embodiment of the present invention, and FIG. 12 is a flow chart showing a modification thereof.
FIG. 13, FIG. 14, FIG. 15, and FIG. 16 are diagrams for explaining the operation.

In this embodiment, an example in which the present invention is applied to the alignment of a reduction projection exposure apparatus, that is, the alignment of a stepper in the manufacturing process of a semiconductor device will be described.

Before explaining the positioning apparatus of the present invention, a reduction projection exposure apparatus will be briefly explained.

FIG. 2 is a perspective view showing an outline of the reduction projection exposure apparatus.

In the same figure, the original image of the semiconductor element pattern is on the reticle 6.
This image is projected onto the wafer 1 via the reduction projection lens 7. The wafer 1 is automatically transferred from the cassette 8 onto the loading table 9, and is placed in the pre-aligner 1o.
After rough positioning is performed, the transfer arm 11 vacuum-chucks the chuck 13 on the XY stage 12.

On the other hand, the reticle 6 is aligned by a reticle alignment optical system 20 so that its center coincides with the center of the reduction projection lens 7. In the case of the reduction projection exposure apparatus according to this embodiment, for positioning the reticle 6 and the wafer 1, a through-the-lens position detection X system 21 and a position detection Y system 22 are provided. This detection system irradiates the alignment mark formed on the wafer 1 with light of a wavelength to which the resist is not sensitive, and detects the regular reflection image from this mark.
The slit is scanned and detected by a photomultiplier tube, and the window pattern of the reticle 6 is also detected. Further, the wafer 1 is configured to have its position measured by a laser interferometric length measuring meter 30. Then, from these measurement data, the deviation of the wafer pattern from the designed grid position is determined.

The laser beam 31 emitted from the laser interferometer 30 is
It is separated by a spectrometer 32. One laser beam 31 is irradiated onto an X-axis peripheral mirror 33 attached to the XY stage 12. This irradiation light is reflected by the X-axis circumferential mirror 33 and returns to the laser interferometric length measuring meter 30.
Coordinates are detected. The other laser beam 31 is irradiated onto the Y-axis layer mirror 36 attached to the xy stage 12 via mirrors 34 and 35, respectively. The laser beam 31 irradiated and reflected by the Y-axis layer mirror 36 passes through the mirror 34, 35 and the spectrometer 32 and reaches the laser interference length measuring needle 30, so that the Y coordinate of the XY stage 12 is detected. It has become. The XY stage 12 is configured to be controlled to move with high precision in the X-axis direction by an X-axis circumferential motor 37, and to be controlled to move in the Y-axis direction with high precision by a Y-axis layer motor 38.

On the other hand, the wafer 1 on the chuck 13 whose work has been completed is transferred onto the unloading table 40 by the multi-feed arm 11. This unloading table 4
The wafers 1 transferred onto the cassette 41 are sequentially accommodated in a recovery cassette 41 by, for example, an air bearing mechanism configured on the unloading table 40.

In this embodiment, the alignment device 50 is a deformation amount calculation unit that calculates the deformation amount of the wafer 1 based on the measurement data regarding the mark group by the position detection X system 21 and the position detection Y system 22 and the design data from the design data unit 56. 51, and an error calculation unit 52 that calculates an error for each mark based on the measured data from the detection system and the amount of deformation from the deformation amount calculation unit 51.
, an exclusion section 53 that excludes data having an error greater than a predetermined standard value from the data from the error calculation section 52, and a correction calculation section 54 that corrects the amount of deformation of the wafer by using the data group after removal. and a controller 55 that controls the motors 37 and 38 based on the amount of deformation of the calculation unit 54.

Next, the alignment method used in this exposure apparatus will be explained.

The wafer 1 on which the pattern has already been formed is placed in the pre-aligner 1.
0 using the pattern of two points on the wafer 1,
Coarse alignment in the Y force direction and rotational direction is performed. At this time, in the case of this apparatus, since there is no rotation mechanism on the XY stage 12, the XY stage 12 is positioned on the pre-aligner IO so that the rotational error is reduced.

The wafer 1 transferred and attracted onto the chuck 13 is further subjected to global alignment using the two sitants 2 inside the wafer 1. In shot 2, all X marks A and Y marks A are arranged. For example, the third
As shown in the figure, shot A in wafer l is first placed on XY stage 12 under reduction projection lens 7.
Positioned by Then, the position detection Y system 22 detects an error value with respect to the design value in the Y direction, and the position detection X system 21 detects a similar error amount in the X direction. Next, in the same way, the XY position is detected for shot 2 B in the wafer 1, and the error with respect to the design value (design shot) is determined.

Therefore, from the above results, the rotation of the wafer 1, the offset (off) in the XY force direction, and the expansion and contraction of the wafer 1 in the XY force direction are calculated, and the exposure grid is determined.

In the case of exposure using conventional global alignment, the XY stage 12 positions the wafer 1 by step-and-repeat based on this exposure grating, and then the reticle image is transferred to the wafer 1. Now that even higher precision is required, after global alignment, X mark A and Y mark A are further detected for each shot, and the xY stage 12 is Error correction is performed by precise control or, as shown in FIG. 2, by a reticle fine movement system 42 that supports the reticle 6, and then exposure is performed.

In the case of such "in-situ exposure method" chip alignment, it is necessary to restrict the positions of the X mark and Y mark placed within the shot. In the case of this reduction projection exposure apparatus, which does not place restrictions on placement, the positions of X marks A, L, and Y marks A are detected for each shot within the range where the position of wafer 1 can be detected, and then exposure is performed based on the detection results. A grid is determined, and positioning exposure of the wafer 1 is performed by step-and-repeat.

However, in such an exposure method based on positioning, if the wafer 1 itself is deformed, the positioning accuracy will be lowered. Therefore, in this embodiment, by using the positioning method described in detail below, even if the wafer 1 is deformed, highly accurate alignment and exposure can be achieved.

As mentioned above, the present invention relates to a technique for determining an exposure grid after position measurement for each shot is completed, and its operation will be described below.

In this embodiment, the positions of a plurality of marks placed on a wafer as an alignment target based on design data are measured by chip alignment. Errors ΔX and ΔY expressed by the difference between the measured results and the design data are expressed by error factors shown in the following formula.

Δχ冑X. ,, +P, +θ, +Sθ+S 14+ S
.

+ε...
(1) ΔY=Yoyr + P y+θx + S e
+S, +S n+2
...(2) Here,
X0FF % YOFF is the offset of shot 2 with respect to the designed shot 43 in the XY axis directions, as shown in FIG. As shown in FIG.
This is the expansion/contraction error in the XY-axis directions. As shown in FIGS. 6 and 7, θ8 and θ are the rotations of shot 2 in the XY axis directions with respect to shot 43 in design.

As shown in FIG. 8, Sθ is designed for shot 4.
As shown in FIG. 9, the rotational error S of shot 2 relative to shot 43 is a magnification error of shot 2 relative to shot 43 in terms of design. Furthermore, Ss is a nonlinear distortion error within the seat, and ε is a nonlinear position error that occurs during step-and-repeat.

Among these error factors, components other than the aforementioned SI and ε can be calculated by statistical means based on the results previously measured in chip alignment as linear errors of the wafer. Using this method, each error factor except 30 and ε is expressed as a linear equation by averaging processing and regression calculation from the coordinate data determined by chip alignment within the wafer.

Hereinafter, the formula for calculating each first-order error will be explained.

In the case of this embodiment, only two X marks and Y marks within a shot are detected. As shown in FIG. 5, the position measurement error for any shot is ΔX i , J %
ΔY5, expressed as J.

Regarding the offset, it is calculated by the following formula as the average of all measured data.

Here, N is the number of measurement data, and M is a matrix.

At this time, the expansion and contraction of each axis P, , P, and the rotation θ of each axis
8, θ is calculated before the offset calculation.

That is, after calculating the expansion/contraction and axis rotation of the wafer, the offset is calculated.

The expansion and contraction of each axis P, , P, and the rotation of each axis θ8, θ,
is obtained by regression calculation using 1 averaged for each column and 7 averaged for each row in the following equation.

j=1 Here, S, , S is the exposure pitch.

1=1 N, , N is the number of measurement data for each matrix.

j=1 From the above results, the coordinates Xe8, Yo of the center of the shot to be exposed on the wafer are expressed by the following equation, with the wafer center as the origin.

X tx = X or t + P x −L x+
θy−L, y −03) Ytx””Yott +P
Y' L 3'+θx-Lx "QIJ here Lx
, Ly is the X from the wafer center of each shot center,
Indicates the distance of the Y component.

In the case of this exposure system, the wafer rotation error is not precisely corrected, so when global alignment is completed, the reticle rotation is corrected based on the result of calculating the wafer rotation amount, and the wafer rotation is as if it were O. Needless to say, the feed of the XY stage of the exposure apparatus must be corrected so as to achieve this. In addition, by using θX and θy determined from the shot measurement data and correcting the X-Y stage again using the average as the wafer rotation, and simultaneously correcting the changing value of XottyYott, high accuracy against wafer rotation errors is achieved. is measured.

In addition, in the case of through-the-lens, alignment, and in-situ exposure systems, position measurement and exposure are repeated for each shot, so it is necessary to finish the measurement exposure for shots within the wafer that can perform position measurement, and then use that data. Because the above-mentioned calculation determines the exposure grid for shots whose positions could not be measured, it is possible to achieve higher accuracy than the exposure grid determined by conventional global alignment. Improves accuracy.

In the above embodiments, the position measurement data of all shots whose positions can be measured have been basically treated.
Furthermore, in order to shorten the time for position measurement and improve throughput, it is possible to reduce the number of position measurement shots within the wafer using the following equation.

Y-3ln(θ)X -OS where θ is 0.30" 60° 90°
! ...06) Here, r is 2/3R (R: wafer diameter).

In Fig. 5, 13 shots (see Fig. 3), which are the sum of the shot of the intersection coordinates where formula (to) holds and the wafer origin (center) (marked with an x), are selected and the positions are measured, and the positions are measured. Determine the exposure grid according to the following.

In the case of an even matrix, the center of the a4.051 formula is 1/2
It is necessary to shift the shot from the wafer center.

In this way, considering the symmetrical position of the measurement shot wsy with respect to the wafer center is because it is necessary to reflect the features of the entire surface inside the wafer in determining the exposure grid, compared to selecting the measurement shot arbitrarily. be. In addition, the reason why the number of points is 13 points is that this calculation method% formula % In this example, the error term SM of formulas (1) and (2),
Sθ was omitted. However, in a device that has a magnification control mechanism for a reduction lens, it is possible to correct S, and by providing position detection marks at three or more points within a shot, the surface method (7) to Using P,,P,
, θ8, θa, and S.

Finally, the average value of P and Py, and Sθ is θ8 and θ
By correcting the average value of , it is possible to further improve the accuracy.

By the way, with wafers, the wafer itself is deformed due to the process, the shape of the position detection mark becomes poor, and marks with low S/N of the detection system are generated, resulting in large errors in position measurement. may occur. In this embodiment, if the result of position detection for each shot is larger than a certain value with respect to the grid determined by global alignment, that position data is excluded.

The alignment sequence according to this embodiment will be explained below using the flow diagram shown in FIG.

First, as described above, global alignment of the wafer 1 is performed by measuring two or three marks on the wafer 1, and rough positioning is performed.

Next, as described above, alignment position measurements are performed at multiple points within the wafer 1, and final statistical processing alignment data is sampled.

For data that exceeds a certain standard value a for the actual measured values ΔX and ΔY, which are the position measurement results, there is a possibility that the alignment is erroneously detected or that only the alignment mark is misaligned. It will not be used and will be excluded from the actual measurement data.

Here, in the case of this example, the standard value a is a value stipulated for the data ΔX and ΔY that includes the orthogonality error, which is the difference between the wafer expansion/contraction error and the rotation error of the X and Y axes. , the processing process, and the management method of the exposure apparatus are set in advance.

Now, if the frequency distribution diagram of the error ΔX before excluding data exceeding the standard value a is as shown in Figure 13, then the frequency distribution diagram after excluding data is, for example, as shown in Figure 14. As shown, the scope is narrow. However, since there is almost no difference in the peak positions of the frequency distribution, there is no problem in the averaging process.

After this exclusion, the actual measured values ΔX and ΔY are used to calculate the error term, the X-axis expansion/contraction amount Px, and the Y-drawing expansion/contraction amount py.
, x-axis rotation amount θX1, Y-axis rotation amount θy, X-direction offset X-FF, Y-direction offset Y. .

, are required respectively.

In the conventional alignment flow, the exposure spot position is then determined from each error term, and the wafer is exposed. In this conventional alignment method, the specific value of the standard value a needs to be set to about 1.5 μm to 2 μm, and abnormal data smaller than that cannot be excluded.

In this example, after each error term is calculated by statistical processing calculation, each error term component is subtracted from the actual measured values ΔX and ΔY, and the remaining error data ΔX'ΔY' exceeds a certain standard value. The data is excluded from the actual measurement value data because the original actual measurement values ΔX and ΔY are not used in subsequent data processing.

This standard value a is a value specified for data after subtracting each error term, so it is a smaller value than the standard value a. Also, like the value a, this value is set in advance based on the processing process, type of exposure apparatus, management method, etc. When using a high-precision stepper for all layers, the specific value of this standard value is set to about 0.15 μm.

Thereafter, each final error term is calculated by statistical processing using the final remaining actual measured values ΔX and ΔY. The exposure shot position is calculated using this error term, and wafer positioning exposure is performed.

Now, if the straight line regarding the ideal X coordinate with no errors for multiple mark groups is the straight line shown in FIG. 15, the error actually measured at each coordinate position is ΔXl ~ Δ
X7, and assuming that the averaging line (regression line) of the actual measured data is the straight line Lo, the averaging line after excluding data exceeding the large standard value a will be La.

As shown in FIG. 16, the errors due to the measured data after correction based on this averaged line La are ΔX,'~
ΔXt', and the averaged line of the actual measured data is the straight line L.
Assuming that o', the averaging line after excluding data exceeding the small standard value S becomes Lb.

Straight line Lal1!1. As is clear from the comparison with Lb,
With respect to the original straight line LO, the straight line Lb is closer to the ideal straight line than the straight line La.

Now, if we consider the case where data exceeding the small standard value S is excluded from the beginning, Δ
Since even the data that should be used, such as Xs and ΔX4, are excluded, proper correction will not be performed, resulting in a decrease in accuracy.

It comes alive.

By the way, in the case of this example, the standard value 0.15 μm is 1710 times smaller than the standard value a of 1.5 μm, so the force is smaller than the standard value a (with a larger error than the standard value). If there is a lot of data, the precision of each error term is poor, so it is possible that the remaining data without errors may exceed the standard value.

Therefore, the alignment flow shown in Fig. 12 is used, and the standard value is first set to a large value close to the standard value a, and from the remaining data, the actual measured values ΔX and ΔY with large errors are excluded. Then, statistical processing calculations and error term calculations are performed.

Next, the standard value is reset, and this process is repeated until the specific value of the standard value reaches the final small value of 0.15 μm, which prevents normal data from being excluded. .

In addition, in the alignment flow of this embodiment, offset error, expansion/contraction error, orthogonality error, and rotation error are corrected as error terms. rotation of,
It can also be applied to increase or decrease error terms such as magnification errors.

By the way, in wafer global alignment,
When performing 3-point or 5-point alignment, roughly estimating each error term, and performing multi-point alignment position measurement using the corrected coordinate system, and setting the standard value a for the data at this time, Since the value can be set to a smaller value than the conventional standard value, the same effect can be obtained.

According to the embodiment described above, the following effects can be obtained.

(1) The deformation of the alignment target can be corrected by determining the amount of deformation of the alignment target and subtracting this amount of deformation from the actual measurement data, so after correcting this deformation, the error is corrected again. By doing so, alignment accuracy can be greatly improved.

(2) In the conventional alignment method, the standard value a is 1
．． 5μm, and if one erroneous detection of about 1μm occurs, the overlay accuracy within the wafer is 3σzO07μm.
It gets worse to some degree. In contrast, according to this embodiment, the standard value of 0.15 μm makes it possible to exclude false detections of about 0.3 μm, and the in-wafer overlay accuracy is 3.
σ<0.2 μm can always be maintained.

(3) When exposing a pattern on the main surface of the wafer using a step-and-repeat method, the positioning of each shot must be
Offset error in the Y-axis direction (X OF F, YOFF
) 'Stretching error (P,,P,), rotation error of the entire wafer in XY axes (θ8, θy), shot rotation error (SQ
), the magnification error is determined for shots at multiple locations, and then based on this measurement data, the alignment of each shot is determined by averaging processing and regression calculation, and
Before calculating the offset, the shot expansion and contraction (P, ,P,
) and the shaft rotation (θ8, θy) are calculated first, thereby eliminating these errors, and as a result, highly accurate positioning can be achieved.

(4) The effect of highly precise alignment of each shot on the main surface of the wafer can be obtained.

(5) As a result of measuring several shots with alignment marks on the main surface of the wafer and using these measurement data to position other shots with high precision, it is possible to measure the positions of other shots with high accuracy for incomplete shots without alignment marks. It is also possible to achieve the effect of highly accurate positioning. That is, conventionally, for incomplete shots that are located around the wafer and are partially missing and have no alignment mark, alignment has been performed using measurement data from global alignment. As a result, conventionally, for example, the shot positioning accuracy of the peripheral portion of the wafer was 3σ≦0.
Although the accuracy is good at 2 to 0.1 μm, since the accuracy of incomplete shots around the wafer is low, the accuracy deteriorates to 3σ≦0°3 to 0.2 μm over the entire surface of the wafer including the incomplete shots. However, according to this embodiment, the positioning accuracy of the entire wafer is as high as 3σ≦0.2 to 0.1 μm.

(6) Positioning during wafer exposure is highly accurate over the entire wafer, and incomplete shots around the wafer can also be used, resulting in synergistic effects such as improved chip yield and reduced production costs.

Although the invention made by the present inventor has been specifically explained above based on Examples, it goes without saying that the present invention is not limited to the Examples and can be modified in various ways without departing from the gist thereof. Nor.

In the above description, the invention made by the present inventor is mainly applied to exposure technology in semiconductor device structures, which is the background field of application, but the invention is not limited thereto.

For example, it can be applied to a step-and-repeat X-ray stencil, a projection-type stepper of equal magnification, and a measurement technique that measures the electrical characteristics of each chip using a step-and-repeat method.

The present invention can be applied to at least a technique of repeating step-and-repeat on an alignment target and working on each unit area of the alignment target.

〔Effect of the invention〕

A brief explanation of the effects obtained by typical inventions disclosed in this application is as follows.

The deformation of the alignment target can be corrected by determining the amount of deformation of the alignment target and subtracting this amount of deformation from the actual measurement data. Positioning accuracy can be greatly improved.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing an alignment device used in a reduction projection exposure apparatus which is an embodiment of the present invention, Fig. 2 is a perspective view showing the reduction projection exposure apparatus, and Fig. 3 is a shot shot in wafer alignment. Fig. 4 is a schematic diagram showing the shot offset, which is one of the causes of error in the sea urchin, Fig. 5 is a schematic diagram showing the shot expansion/contraction error, and Fig. 6 is a schematic diagram showing the shot expansion and contraction error. Figure 7 is a schematic diagram showing the X-axis rotation of the shot, Figure 7 is a schematic diagram showing the Y-axis rotation of the shot, Figure 8 is a schematic diagram showing the rotation error of the shot, and Figure 9 is the same magnification of the shot. FIG. 10 is a plan view of a wafer; FIG. 11 is a flow diagram showing a wafer alignment sequence according to an embodiment of the present invention; FIG. 12 is a flow diagram showing a modification thereof; 14, 15, and 16 are diagrams for explaining the operation. 1... Transfer arm, 12... XY stage, 13.
... Chuck, 20 ... Reticle alignment optical system, 21 ... Position detection X system, 22 ... Position detection Y system,
30... Laser interferometer length meter, 31... Laser light, 3
2...Spectroscope, 33...X-axis circumferential mirror 34.35
...Mirror 36...Y-axis circumference mirror 37...X-axis circumference motor, 38...Y-axis motor, 40...Unloading table, 4th work...Recovery cassette, 42
...Reticle fine movement system, 43...Shot by design, 5
0... Positioning device, 51... Deformation amount calculation unit, 5
2...Error calculation section, 53...Exclusion section, 54...Correction calculation portion, 55...Controller, 56...Design data section. IKI Figure 4 Figure 5 Figure 6 Figure 7 [2! 1 Figure 8 Figure 1 o Figure 9 Figure 12 Figure 13 Figure 15 Mar 7f IX Office! 14th figure 116 mark

Claims (1)

[Scope of Claims] 1. In an alignment device configured to measure the positions of marks formed at a plurality of locations on an alignment target and execute alignment work based on the measured data, a deformation amount calculation unit that calculates a deformation amount of the alignment target from measurement data about the mark group; and an error calculation unit that calculates an error for each mark based on the determined deformation amount and measurement data about the mark group. , an exclusion unit that excludes data having an error greater than a standard value from the error data group; and a correction calculation unit that uses the remaining data group after exclusion to perform correction calculations on the amount of deformation of the alignment target. What is claimed is: 1. A positioning device comprising: a positioning apparatus configured to perform positioning work based on data for this correction calculation. 2. The alignment device according to claim 1, wherein the standard values are set in stages.