JPS63260022A - Aligner - Google Patents

Abstract

PURPOSE:To minimize an error component and achieve alignment with a high accuracy by a method wherein the position of a 1st pattern estimated by the position of an optically detected mark is calculated and the 1st pattern and a 2nd pattern are relatively moved in accordance with the calculated results. CONSTITUTION:An alignment light LA is outputted from a light source 10 and an alignment mark is scanned and coordinates Xn and Yn observed by a mark position detector 14 are obtained. Then the observed values are inputted to a successive position estimating filter 22 and a mark coordinate estimated value Xsn, a mark interval estimated value DELTAXsn and predicted values X<m>n+1 and DELTAX<m>n+1 are calculated respectively. Among those estimated values, the mark coordinate estimated value Xsn is inputted to the shot position calculating part 26A of an exposure controller 26 and the predicted value X<m>n+1 is inputted to an alignment coordinate calculating part 28. In the shot position calculating part 26A, the position of an (n)th shot region is calculated in accordance with the input and inputted to a stage controller 32 which moves a stage S in accordance with the input and, by utilizing the successive position estimating filter, highly accurate registration exposure can be performed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application in Industry A] The present invention relates to an alignment apparatus suitable for, for example, a step-and-repeat reduction projection exposure apparatus.

[Prior Art] In a conventional exposure apparatus of this type, as schematically shown in FIG. The image is projected in a reduced size onto a wafer W set on a stage S via the wafer W set on the stage S.

Projection of such a pattern is performed for each shot area corresponding to a chip by moving the wafer W in the XY directions in the figure.

FIG. 3 shows an example of a wafer W, in which alignment marks MX and MY are formed in each shot area Emo corresponding to the above-mentioned XY directions.

The projection of the circuit pattern of the reticle R described above is, for example,
Shot area j5 E z, E +□,...
...Ell, E 21. E22. ...
・・・・・・E 27. E 28・・・・・・・・・
This is done by moving the stage S in the order of .

At this time, each shot area E,. In each case, the circuit pattern of the reticle R and the shot area EMN are aligned using the alignment marks MX and MY.

The coordinate positions of these alignment marks are determined photoelectrically by a predetermined detection means, and after positional deviations are corrected, overlapping exposure of circuit patterns is performed for each shot area.

[Problems to be Solved by the Invention] However, in the conventional techniques as described above, the alignment mark MX on the wafer W.

Due to errors in the detection means for detecting MY and distortions in the alignment marks MX and MY that exist depending on the state of the wafer W surface (all white noise-like components), the detected mark position coordinate values may have random errors. Errors may be added, resulting in the inconvenience that highly accurate alignment cannot be performed.

The present invention has been made in view of the above points, and provides an alignment device that can perform highly accurate alignment by minimizing random error components included in observed coordinate values of alignment marks obtained from a detection means. Its purpose is to provide.

[Means for Solving the Problems] In order to solve the above problems, the present invention optically detects either a plurality of first patterns to be aligned or marks each has, and observes its position. a mark position detection means for calculating a value; a predicted value of the position of the currently detected mark calculated based on the observed value and the estimated mark value obtained last time if a previous alignment has been performed;
a sequential position estimation filter that calculates an estimated value of the current detection mark position using a weighting coefficient determined corresponding to the array position of the first pattern; , calculation means for calculating the position of the first pattern estimated from the position of the detected mark; and relative movement of the first pattern and the second pattern based on the calculation result of the calculation means. The technical point is to have a driving means for performing the following.

[Operation] According to the present invention, first, the mark of the first pattern (pattern on the wafer) to be aligned this time is observed by the mark position detecting means film, and the observed value of the position is obtained.

Next, in the sequential position estimation filter in which the observed values are manually input, if the previous alignment has been performed, the predicted value of the position of the detected mark this time is calculated based on the mark estimated value obtained last time. Furthermore, a weighting coefficient is determined in advance in accordance with the arrangement position of the first pattern to be aligned.

In the sequential position estimation filter, an estimated value of the current detected mark position is calculated based on these predicted values and weighting coefficients and the input observed value of the detected mark.

Next, the estimated value is input to the calculating means, and the position of the first pattern to be aligned this time is calculated.

The calculation result of the calculation means is manually inputted to the drive means, and the first pattern is moved relative to the second pattern (pattern on the reticle).

[Embodiments] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. Note that the same reference numerals are used for the same components as in the prior art described above.

FIG. 1 shows the main parts of an embodiment of an alignment device according to the present invention. In this figure, alignment light LA, which is illumination light for alignment, is output from an alignment light source 10 and is directed to the alignment optical system 1.
It is designed to be incident on 2.

The alignment light LA that has passed through the alignment optical system 12 passes through the reticle R and enters the projection optical system T.
Furthermore, the light passes through the projection optical system T and illuminates the alignment marks MX, MY (see FIG. 3) in a predetermined shot area EIIlo of the wafer W as a spot light (sheet-like spot).

The alignment marks MX, MY are formed, for example, as unevenness on the surface of the wafer W, and the incident alignment light is scattered by the machining/saw portions of the marks, and this scattered light is sent to the projection optical system T as the detection light LD. It is designed to be incident on .

The detection light LD travels in the opposite direction of the optical path of the alignment light LA, enters the alignment optical system 12, and further enters the mark position detector 14. This mark position detector 14 detects the predetermined coordinates of the alignment marks MX and MY on the wafer W from the incident detection light LD, based on the coordinate system on the stage side, and detects the coordinate position of the stage S. A detection signal from a laser interferometer 16 is input.

Next, the output side of the mark position detector 14 described above is connected to the sequential position estimation filter 22 of the control calculation section 20. This filter 22 constitutes a Kalman filter as shown in FIG. 4, for example. This sequential position estimation filter 22 calculates the coordinate position of the alignment mark of the shot area currently specified by the shot area specifying unit 24 that specifies the shot area using the estimated position value of the previously detected alignment mark. This function has a function of estimating the coordinate position of the alignment mark using the predicted value of the coordinate position.

Next, the output side of the sequential position estimation filter 22 is connected to an exposure control section 26 and an alignment coordinate calculation section 28, respectively. Among these, the exposure control section 26 is provided with a shot position calculation section 26A, which allows shot position data to be calculated.

Further, the alignment coordinate calculation section 28 is configured to obtain alignment coordinate data of the corresponding shot position.

Next, the exposure control section 26 is connected to an exposure light source 30, and after the alignment is completed, the exposure light source 30 is driven to expose a predetermined shot area of the circuit pattern of the reticle R.

The above-mentioned exposure control section 26 and alignment coordinate calculation section 28
The output sides of both are connected to a stage controller 32, and to this stage controller 32, the above-described laser interferometer 16 and a drive motor 34 for driving the stage S in the X and Y directions are respectively connected.

The stage controller 32 drives the drive motor 34 based on data manually inputted from the laser interferometer 16 and the control calculation unit 20, and performs eight rotations of the stage S to align the reticle R and the wafer W. be.

It is assumed that the projection position of the circuit pattern of the reticle R on the stage S side is determined in advance based on the device configuration.

Next, referring to FIG. 4, the sequential position estimation filter 2
2 will be explained in detail.

Note that although the illustrated sequential position estimation filter 22 is for the X direction, the device configuration for the Y direction is also the same.
It is configured by software or hardware as appropriate.

In FIG. 4, the output side of the mark position detector 14 of FIG. 1 is connected to an adder 40 as a positive side human power. The output side of this adder 40 is connected to a weighting factor multiplier 42, and the output side of this weighting factor multiplier 42 is connected to an adder 44 as a plus side input.

The output side of this adder 44 serves as the output of the sequential position estimation filter 22 on one side, and is connected to the predicted value calculation unit 48 via a delay circuit 46 on the other side.

Next, the output side of the predicted value calculation section 48 is connected to the plus input side of the adder 44 on one side, and to the predicted coordinate extraction section 50 on the other side. Further, the output side of the predicted coordinate extraction section 50 is connected to the minus input side of the adder 40.

Next, the sequential position estimation filter 2 configured as described above is
The effect of 2 will be explained.

In the following explanation, the observed coordinate values (actual measured values) of the alignment mark of the n-th shot area E an to be exposed are assumed to be X, , Yn, and the estimated values are assumed to be X @n*
Let Y an and the predicted values be xn", Yn", and n −
The estimated value of the alignment mark spacing between the first and nth shot areas is ΔX Mn+ ΔY. Let the predicted values at the same interval be Δxn" and ΔYn".

Furthermore, the operations for the alignment marks MX and MY are similar, and the operations for the X and Y directions of each mark are also similar, so the X direction of the mark MX will be described below.

First, the alignment mark MX of the n-th shot area Elfin to be aligned is scanned by the alignment light LA, and the observed value xn of the coordinate position of the mark MX is obtained by the mark position detector 14.

Next, the observed value xn obtained as described above is input to the adder 40 of the sequential position estimation filter 22. On the other hand, the adder 40 receives the predicted value X of the alignment mark MX from the predicted coordinate extraction section 50. " is input. This predicted value is obtained based on the results of alignment up to the previous time (n-1).

The adder 40 calculates the predicted value x from the observation (m
n'' is subtracted, and this subtracted value is manually input to the weighting coefficient multiplier 42.

This weighting factor multiplier 42 uses weighting factors α8. β8 are each selected,
These are multiplied by the human power value.

To be more specific, the weighting coefficients α8β8 are, for example, as shown in FIG. 5(A).
, CB), the shot area E
It is changed corresponding to the shot position n of an, and the later the exposure is performed among the shot areas E0 adjacent in the same direction, the value becomes smaller than "1" and the degree of weight is reduced. Set.

For example, in FIG. 3, in the shot area Ell, the weighting coefficient α8. β8 is “1” and becomes smaller in the last shot area EL6 lined up in the Y direction.

However, in shot area E21, which is the next step shifted by one column in the X direction from that column, the weighting coefficients α and β8 return to the original values as in shot area Ell, and decrease in shot area E2IS. do. The same applies to the X direction.

In this way, the weighting coefficient α8. The reason why β8 is changed is to take into consideration the amount of movement of the stage S when sequentially exposing the shot areas E1lln by step-and-repeat. This coefficient α8. The change in β8 and the operation control of the Kalman filter that operates during stepping in the X direction and the Kalman filter that operates during stepping in the Y direction are performed based on information from the shot area specifying unit 24.

The human input values are multiplied by the weighting coefficients α8 and β8 selected as described above, and the resulting equations (1) and (2) are input to the adder 44, respectively.

α, (X, -Xo“)・・・・・・・・・・・・(1)
β, (X, -X〆) (2) On the other hand, the adder 44 receives the predicted value X of the alignment mark MX described above. ", and the predicted value Δxn of the alignment mark interval of the (n-1)th and nth shot areas" are respectively input.

In the adder 44, each of these inputs is added, and n
The optimal coordinate estimate x of the alignment mark of the th shot area EIIIn! n and the optimal mark spacing estimate Δx
, n are calculated as in equations (3) and (4), respectively. These estimated values are sequentially output from the position estimation filter 22.

x, n = x o” + αX (Xn - Xo゛)...
...(3) ΔX, n=Δx7"+βX (X n-X
n') ・(4) On the other hand, the above estimated value is
6, and is used to predict the coordinate values of the alignment mark of the n+1-th shot area E□, , in the next step.

First, using the estimated values of equations (3) and (4), the predicted value calculation unit 48 performs calculations of equations (5) and (6) to predict the alignment mark coordinates of the n+1-th shot area Effln+1. Value X. . 1'' and mark interval predicted values ΔX n, , `` are respectively determined.

X Oh+”=X sn+ΔX fo+・・・・・・
...(5) ΔX n+ 1" = ΔXyo.
(6) These predicted values are input to the adder 44, and the predicted coordinate values X n
il'' is extracted by the predicted coordinate extraction unit 50 and input to the adder 40.

In addition, in FIG. 4, the contents of calculations in each part are shown by matrices. Further, as described above, the same applies to the Y direction.

As described above, according to the sequential position estimation filter 22, n
Using the estimated values of the coordinate position and mark interval of the alignment mark of the -1st shot area E+an-1, the predicted value of the mark coordinate position and mark interval of the nth shot area E mn is calculated, and Using these predicted values, estimated values of the mark coordinate position and mark interval of the n-th shot area E +nr+ are calculated.

The sequential position estimation filter 22 configured in this manner applies a Kalman filter to an array model as shown in FIG. This has the effect of minimizing noise components.

Next, the overall operation of the above embodiment will be explained.

For example, assume that the exposure of the (n-1)th shot area E a+n-1 has been completed and the shot area specifying unit 24 has specified the nth shot area E0.

In this case, since the predicted values of the alignment mark coordinate position 9 interval values of the n-th shot area have been sequentially determined by the position estimation filter 22, these are input to the alignment coordinate calculation unit 28, where the n-th Alignment coordinates of the shot area are calculated. Here, a stage position is calculated such that the predicted mark position is directly below the alignment light LA.

The calculated value is then used by the stage controller 32.
is input, and the stage S is moved so that the alignment mark of the n-th shot area E11In to be aligned is illuminated by the alignment light LA (spot light) as shown in FIG.

The alignment light is directed to the alignment light irradiation position by drive control of the driver 34 by the stage controller 32. That is, the stage controller 32
Predicted XY coordinate position P of the alignment mark in the shot area Ean input from the alignment coordinate calculation unit 28
The drive motor 34 is driven in accordance with xn and Pyn while monitoring the coordinate position of the stage S based on the output of the laser interferometer 16.

Next, the alignment light LA is output from the light source 10, and the alignment mark is scanned, and the alignment mark is aligned with the position detector 1.
Observed values Xn and Yn based on 4 are obtained, respectively. These observed values Xn, Y, are sequentially input to the position estimation filter 22, where the above-mentioned calculations are performed to obtain the mark coordinate estimated value X sn, the mark interval estimated value ΔX sn, and the predicted value X.
"n+1+ΔX". , 1 are calculated respectively.

Among these estimated values, the mark coordinate estimated value X an is
The predicted value X"n□ is input to the shot position calculation section 26A of the exposure control section 26, and is manually input to the alignment coordinate calculation section 28.

The shot position calculation unit 26A calculates n based on this input.
The position of the second shot area is calculated and inputted to the stage controller 32.

The stage controller 32 performs the adjustment of the stage S based on the input, and the alignment is completed.

After this operation, the exposure light source 30 is driven by the exposure control section 26, and the n-th shot area is exposed.

The above operations are repeated for each shot area to perform step-and-repeat exposure.

As explained above, according to this embodiment, random errors that exist in the mark state on the detector or wafer or in the coordinate detection system can be minimized and the best estimated value of the mark coordinates can be found sequentially. Since the position estimation filter is used, highly accurate overlapping exposure can be performed.

It should be noted that the present invention is not limited to the above-mentioned embodiments, and can be applied to inspection devices other than exposure devices, for example.

In addition, when random error components fluctuate due to wafer conditions, etc., the influence of error fluctuations can be suppressed to a smaller extent than before by adjusting the position estimation filter to be the best for that error. is possible.

Furthermore, if you want to improve alignment throughput compared to the die-by-die method that detects (actually measures) the mark position one shot at a time, alignment is performed at one-shot intervals, and the wafer is positioned in the unaligned shot areas based on predicted values. Exposure may also be performed.

Furthermore, in the above embodiment, a case has been described in which alignment and exposure are performed for each shot area E mn arranged on the wafer W by directly detecting the alignment mark of that shot area. Alternatively, alignment and exposure of the corresponding shot area may be performed indirectly by detecting the alignment mark of the area.

As such an indirect alignment method, for example, as disclosed in Japanese Patent Application Laid-Open No. 61-44429,
A method of actually measuring mark positions associated with some shot areas at representative positions among a plurality of shot areas on a wafer in advance, and certifying the arrangement of the shot areas on the wafer using a statistical method; That is, a certain type of wafer global alignment is known. It is also possible to use a combination of such global alignment method and the alignment method according to the present invention. In this case, by starting the shot (exposure) on the wafer based on the alignment result obtained by the global alignment method, it is possible to suppress errors at the start of the shot.

Alternatively, the noise component (error) weighing on the mark coordinate values obtained by the above global alignment method
The weighting coefficient (α) of the Kalman filter is determined according to the size of the variance.

β) may be fixed as a constant value. In this case, the sequential position estimation filter is not configured as a Kalman filter, but takes the form of a simple weighted average equation, which has the advantage of simplifying internal processing.

Further, although the alignment system shown in the embodiment of the present invention is designed to detect only the marks on the wafer, there is an alignment system that simultaneously detects marks attached to the reticle and marks attached to the shot area on the wafer. A similar effect can be obtained by using a system.

[Effects of the Invention] As explained above, according to the present invention, it is possible to suppress the random error component included in the coordinate values of the alignment mark obtained from the detection means to the minimum range and perform highly accurate alignment. There is an effect.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram showing an embodiment of the present invention, FIG. 2 is a schematic explanatory diagram showing an example of a reduction projection type exposure apparatus, and FIG. 3 is a pattern array having alignment marks on the wafer shown in FIG. FIG. 4 is a circuit block diagram showing an example of the circuit configuration of a sequential position estimation filter, and FIG. 5 is a diagram showing an example of weighting coefficients of the filter. [Description of symbols of main parts] 14... Mark position detector, 20... Control calculation unit,
22... Sequential position estimation filter, 26... Exposure control section, 26A... Shot position calculation section, 28... Alignment coordinate calculation section, 32... Stage controller, 42... Weighting factor multiplication section , 48... Predicted value calculation section, 50... Predicted coordinate extraction section, Eyu. ...Shot area, MX, MY...Alignment marks Y, R...
Reticle, S...stage, T...projection lens,
W...Wafer.

Claims (1)

[Scope of Claims] A first pattern is formed using any of the marks each of a plurality of first patterns arranged at approximately equal intervals in the same direction on a substrate set on a stage movable within a predetermined plane. In an alignment device that performs relative alignment between a first pattern and a second pattern, a mark position detection means for optically detecting one of the marks of the first pattern and obtaining an observed value of the position; , an observed value of the detected mark position and, if a previous alignment has been performed, a predicted value of the currently detected mark position calculated based on the previously obtained mark estimate;
a sequential position estimation filter that calculates an estimated value of the current detection mark position using a weighting coefficient determined corresponding to the array position of the first pattern; , calculation means for calculating the position of the first pattern estimated from the position of the detected mark; and relative movement of the first pattern and the second pattern based on the calculation result of the calculation means. An alignment device characterized by comprising: a driving means for performing.