JPS63232324A - Alignment - Google Patents

Abstract

PURPOSE:To improve accuracy of alignment, by measuring the amount of shift of an original plate to a substrate in prescribed regions on the substrate and performing theta directional alignment of the substrate on the basis of the results of measurement and next confirming theta correction in the measured regions or in the other regions. CONSTITUTION:The amount of position shift of an original plate RT to a substrate WF in prescribed regions on the substrate WF is measured by observing light which transmits a projection optical system LN and is reflected on or diffracted from marks on the substrate WF, and theta directional alignment of the substrate WF is performed on the basis of the results of measurement. Thereafter, measurement is performed again in the measured regions or in the other regions so as to confirm the alignment of the substrate WF, that is, theta correction. Hence, the alignment in the theta direction can be performed with good accuracy, and so observation is not required in global alignment and die.by.alignment of the following processes.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an alignment method, for example, for aligning an original plate such as a mask or reticle and a substrate such as a semiconductor square with high accuracy in an exposure apparatus that is a semiconductor manufacturing device. alignment)
Regarding how to.

[Prior Art] In recent years, as semiconductor integrated circuits such as ICs and LSIs have become smaller and more highly integrated, semiconductor exposure apparatuses have become increasingly sophisticated. Particularly at present, it is required that the original plate and the substrate to be aligned be superimposed on the order of submicrons.

A device called a stepper is known as an exposure device used in such semiconductor manufacturing. This stepper is
While the substrate, such as a semiconductor wafer, is moved step by step under a projection lens, the pattern image formed on the original plate, such as a reticle, is reduced by the projection lens, and multiple locations on one wafer are sequentially exposed to light. .

Then, as one alignment method for reticle and wafer in a stepper, TTL (Through) is used to align the reticle and wafer through a projection lens.
gh The Lens) method. Furthermore, TTL
Two methods of alignment are: ■Die-pie-die alignment method, in which alignment is performed for each shot, and ■Alignment is performed based on the results at an appropriate number of measurement points, and then the wafer is stepped according to the shot arrangement. There is a global alignment that performs exposure and the like.

[Problems to be solved by the invention] However, in general, the TTL die-pie-die method performs alignment for each shot of the wafer, so although the overlay accuracy is high, the processing time for one wafer is long. As a result, the overall throughput decreases. On the other hand, TTL global alignment reduces the processing time for one wafer, but has the problem of poor overlay accuracy for each shot.

SUMMARY OF THE INVENTION In view of the problems with the conventional method described above, it is an object of the present invention to provide an alignment method with extremely high alignment accuracy and high productivity (high speed).

[Means and operations for solving the problems] In order to achieve the above object, the alignment method according to the present invention provides TTL
The method measures the amount of positional deviation between the original plate and the substrate in a predetermined area on the substrate, aligns the substrate in the θ direction based on the measurement results, and then aligns the substrate in the θ direction in the measured area or other areas. The measurement will be performed again to confirm the above alignment, that is, the θ correction.

As a result, alignment is performed with high accuracy in the θ direction, and there is no need to look in the θ direction during subsequent global alignment or die-to-die alignment.

Note that measuring the amount of positional deviation using the TTL method is as follows:
This is measurement by observing the reflected or diffracted light of marks on the substrate that passes through the projection optical system.

[Example] Hereinafter, the present invention will be explained in detail using the drawings.

FIG. 1 is a schematic diagram of a semiconductor exposure apparatus to which an alignment method according to an embodiment of the present invention is applied.

In the same figure, ST is an XY stage that moves the wafer in the X and Y directions, and XM and YM are
, a drive motor that drives in the Y direction, ws a θ stage that rotates the wafer in the θ direction, WF the wafer, and OA an off-axis microscope that detects pre-alignment marks and performs rough alignment. In addition, LN is a printed projection lens, RS is a reticle stage that moves the reticle in 8 directions in X, Y, and θ directions, RT is a reticle, and PT is a reticle R.
The circuit pattern drawn at T and Ll is a lighting device for printing. LT is a laser tube, PM is a polygon mirror, M
is a motor that rotates the polygon mirror, Pl is a prism that divides the optical path of the laser beam, and P2. P3 is a mirror,
BSI and BS2 are beam splitters, Ll and L2 are objective lenses, MR and ML are mirrors, DR and DL are charging detectors, CB is a control box equipped with a control circuit consisting of a CPU (central processing unit) and memory, etc. This is a console that allows you to input various parameters.

In the same figure, the laser beam emitted from the laser tube LT is connected to a polygon mirror PM rotated by a motor M.
The 0-divided laser beam is split into a left viewing system and a right viewing system by prism P1, and then passes through beam splitters BSI and BS2, objective lenses Ll and L2, and mirror MR.
, ML, and scan the alignment mark on the reticle and the alignment mark on the wafer, respectively. The reflected light from each mark returns along the original optical path, passes through beam splitters BSI and BS2, and enters photoelectric detectors DR and DL. The photoelectric detectors DR and DL photoelectrically convert this incident light and output it as an electrical signal. This output is input to the control box CB and binarized,
Based on this binary signal, the relative position of each mark (Figure 2 (d)
Therefore, ~1. ) is calculated. From the above, the amount of deviation between the reticle RT and the wafer WF can be determined.

Figure 2(a) shows the alignment mark W attached to the wafer.
, FIG. 2(b) shows an alignment mark M attached to the reticle, and FIG. 2(C) shows a state in which alignment is completed. FIG. 2(d) is a diagram showing how the alignment mark M on the reticle and the alignment mark W on the wafer are scanned with a laser beam. Figure 3 shows
FIG. 4 is a diagram showing the arrangement of alignment marks in each exposure area S (shot area) on the square. Alignment marks W1 and W2 are arranged for each shot area.

In this embodiment, as shown in FIG. 2(d) (specifically, as explained in FIG. 1 above), the alignment marks W and M are scanned with a laser beam, and the relative distance between each mark is 1
1~J'! Find I, find .

ΔX and ΔY are determined for each of the alignment marks (FIG. 3) attached to both sides of the shot area.

For example, by measuring the left mark of shot 1, Δx1 - ΔY1L is obtained, and by measuring the right mark, ΔY1L is obtained.
If X IR and ΔYIRL are obtained, the amount of deviation of the shot in the X, Y, and θ directions can be determined, for example, as follows.

(Amount of deviation in the X direction) - Ako (Amount of deviation in the Y direction) - A22 and 22△] 22 - (Amount of deviation in the θ direction) =
xtan-'△ko! However, ・L indicates the distance between the left and right marks. The amount of deviation in the θ direction is a so-called tip rotation.

In this embodiment, several predetermined shots are predetermined automatically or specified by the user, and the chip rotisshicon in the shot area is determined and averaged.
θ correction is performed in the shot area. In addition, the center position of a predetermined shot is determined using the equation for the Y-direction shift amount, and the difference between the center positions of the left and right shots is determined, and the wafer rotation is determined from this.

FIG. 4 is a diagram showing a shot layout on a wafer.

Next, the operation of the apparatus of this embodiment will be explained in detail with reference to the flowchart of FIG.

First, an overview of the processing will be explained.

Step 1.2> Wafer set and TV pre-alignment.

Steps 3 to 9> The amount of deviation is measured in a predetermined shot area on the wafer, the amount of deviation in the θ direction is calculated from the measurement result, and the rotational component is corrected by driving in θ. Furthermore, the measurement is performed again to confirm that the θ correction is appropriate. Through steps 3 to 9, the position of the wafer in the θ direction is guaranteed.

Steps 10 to 19> The amount of deviation is measured in a predetermined shot area on the wafer, and based on the results, a correction grid, that is, information including coordinates indicating the position of each shot area, is created. If a correction grid is created and stored, alignment with a predetermined accuracy is guaranteed by alignment using the correction grid points.

Note that when wafers coated with resist are aligned and baked, the baked results may be radially, spirally, or randomly shifted. This kind of misalignment does not occur on wafers that have not been coated with resist, so it is likely that the cause is due to resist coating. When creating a correction grid, the above offset amount is added to calculate the correction grid. Therefore, during alignment, the stage is driven with the offset added. This makes it possible to prevent regular misalignment among the above-mentioned unavoidable misalignments.

Next, the operation of the apparatus of this embodiment will be explained in detail.

First, in step 1, the wafer WF is placed on the wafer stage WS. Next, in step 2, television pre-alignment is performed using an off-axis microscope OA. This allows rough alignment of the wafer WF.

The next step 3 to step 9 is a loop for performing θ correction in positional deviation measurement. First, in step 3, the amount of deviation is measured by TTL for one shot (the shot shown in black) in FIG. This measurement is performed as explained in Figs. 2 and 3 above, and the position of the measurement shot is automatically selected according to a predetermined rule as described later, but it can be changed by command from the console CON. You can also do that.

Next, in step 4, it is determined whether the above measurement is good or not. If it is bad, the XY stage is stepped to an adjacent shot in step 4a, and the process returns to step 3 to measure the amount of deviation in that shot. In the case of poor measurement, adjacent shots are automatically selected according to predetermined rules. Further, here, the number of steps to adjacent shots is limited to two shots.

If the measurement is good in step 4, proceed to step 5 and step 1.
It is determined whether or not measurement has been completed for all shots, and if it has not been completed yet, the process advances to step 5a to proceed to the next shot to be measured, and then returns to step 3 to perform measurement. Note that the number of β is selected in advance from the console. However, it is assumed that the shot is selected from m shots, which will be described later, and λ≦m.

If it is determined in step 5 that measurement has been completed for all two shots, the process proceeds to step 6 to determine whether there is an abnormal shot. Here, when the maximum deviation of pitch error and chip rotation is larger than a predetermined value, it is considered as an abnormal value. In the case of an abnormal value, the abnormal value is rejected, step is performed to an adjacent shot in step 6a, and measurement is performed again from step 3. As with step 4a, up to two adjacent shots are automatically selected. If there is no abnormality in step 6, it is determined in step 7 whether the number of effective shots is greater than or equal to the minimum number of shots. The minimum number of shots is set, for example, as follows, corresponding to the number of measurement shots 1.

However, if the number of effective shots is smaller than this minimum value, the process proceeds to step 7a. In step 7a, it is determined whether the mode is automatic measurement, and if so, the process returns to step 3 and re-measurement is performed. That is, if there are an abnormally large number of abnormal value rejects, re-measurement is automatically performed. If it is not the re-measurement mode, the process proceeds to step 7b and an error occurs. In the event of an error, the device is temporarily stopped, and then the operator is informed through the console whether to continue, reload, or try again, and processes are performed according to the instructions.

Next, in step 8, chip rotation, wafer rotation, shift amounts in the X and Y directions, etc. are measured, and abnormal values are rejected. This abnormal value reject is performed when the residual deviation amount r/σ is larger than a predetermined value.

If the number of abnormal value rejects is greater than a predetermined value, the process branches to step 7a and the aforementioned process is performed.

Next, in step 9, it is determined whether the amounts of wafer rotation and chip rotation are within a predetermined tolerance. If not within tolerance, step 9a
and the reticle θ stage (hereinafter referred to as reticle θ/S)
After correcting and driving the 〇 stage (hereinafter referred to as wafer θ/S) and the XY stage (for shifting), the process returns to step 3 for re-measurement of the °C shot. Note that since remeasurement is limited to two times here, the determination in step 9 will be performed three times. In addition, since the loop processing of steps 3 to 9 is repeated in the re-measurement, the tolerance value is slightly increased in the third re-measurement, and the conditions for discrimination are relaxed.

This tolerance value can be changed by user input.

If it is within tolerance in step 9, step 1
The process proceeds to the θXY correction loop of positional deviation measurement from step 0 to step 18. In this θXY correction loop, θ correction, magnification correction, and shift correction of the wafer are performed by driving the XY stage. We also obtain various measurement values for creating a correction grid.

First, in step lO, TT is applied to the m shots in FIG.
Measure the amount of deviation at L. This measurement is performed in the same manner as the above-mentioned 1 shot. Further, the position of the measurement shot is automatically selected like the one shot, but it can also be changed by giving an instruction from the console CON.

Next, in step 11, it is determined whether the above measurement was good or not, and if it is bad, the XY stage is stepped to the adjacent shot in step lla, and the process returns to step 10 again to measure the amount of deviation in that shot. . In the case of poor measurement, adjacent shots are automatically selected according to predetermined rules. Further, here, the number of steps to adjacent shots is limited to two shots.

If the measurement is good in step 11, the process proceeds to step 12, and it is determined whether or not the measurement has been completed for all m shots.If the measurement has not been completed yet, the process proceeds to step 12a, where the process moves to the next shot to be measured, and then steps again. lO
Go back and take measurements. Note that the number m is selected in advance from the console. For example, the value of m is m-thickness 2. 4. Select from 6.8.12.16 etc. Note that the two shots mentioned above are selected from among the m shots and are used in the θ correction loop. Therefore, in steps 3 to 9, θ by one shot is
When performing the measurement in step 10 for the first time after performing the correction loop, the measurement is performed by excluding l shots from m shots (only the mouth in FIG. 4). From the second loop onwards, all m shots (■ and mouth in FIG. 4) are subject to measurement.

Returning to the sequence description, in step 12 all m
If it is determined that the measurement for the shots (excluding the λ shot in the first loop) has been completed, step 13
to determine whether there are any abnormal shots. Here, the same check as in step 6 is performed, and if an abnormality is found, step is performed to an adjacent shot in step 13a, and measurement is performed again from step 10. Selection of adjacent shots is similar to steps 4a and 6a. If there is no abnormality in step 13, it is determined in step 14 whether the number of effective shots is greater than or equal to the minimum number of shots. As in the case of shot 1, a predetermined value is determined for the minimum number of shots in correspondence with the number m of measured shots. If the number of effective shots is smaller than this minimum value, an error is determined in step 14a, and the same processing as in step 7b is performed.

Next, if the number of effective shots is greater than or equal to the minimum number of shots in step 14, chip rotation is performed in step 15.
Calculates wafer rotation, shift amounts in x and y directions, and magnification, and rejects abnormal values. This abnormal value reject is to reject when the residual deviation amount r/σ is larger than a predetermined value. Note that step 14
If there are an abnormally large number of abnormal value rejects in step 15 and 15, automatic re-measurement may be performed in the same manner as in step a.

In this case, the process returns to step IO.

Next, in step lli, it is determined whether each component is within a specified range. i.e. chip rotation,
For wafer rotation, magnification, and alignment, it is determined whether each component is greater than a user-set value. If it is large, it is assumed that an error has occurred and the operation is stopped in step 16a. The processing at this time is similar to step 7b.

If each component is within the specified range in step 16, in step 17 the console selects whether or not correction driving is necessary. If correction driving is to be performed, it is determined in step 18 whether the rotation (wafer 0), shift (XY), and magnification are within tolerance. This tolerance value is entered from the console. Remeasurements may be made up to two times. Therefore, the determination in step 18 can be made up to three times. Step 1
If it is out of tolerance in step 18a,
Drive the Y stage for correction and return to step 10 again.
Take measurements.

Furthermore, if it is within the tolerance in step 18, a correction grid is created in step 19. Even if it is not necessary to confirm the correction drive in step 17, the process branches to step 19 and a correction grating is created immediately.

Thereafter, the exposure sequence from step 20 is executed. First, in step 20, a selection is made between global alignment and die-pie-die alignment. This selection is made by inputting from the console.

In the case of global alignment, the XY stage is moved in steps to expose the first shot area in step 21, focusing and exposure are performed in step 22, and it is determined in step 23 whether or not it is the final shot. If it is not the final shot, step 23a advances to the next shot, and the process returns to step 22 to perform focusing and exposure. This is repeated, and when the final shot is reached, the wafer is reloaded in step 31, and the process proceeds to the next step.

On the other hand, in the case of die-pie-die alignment, the XY stage is driven step by step to expose the first shot area in step 24, and in step 25, the offset and position of the shot is measured by TTL.

Next, in step 26, it is determined whether the amount of deviation is within the limit. The limit value is input in advance from the console CON. If this limit value is exceeded, the reticle is centered in step 30, and the process proceeds to step 28, where focusing and exposure are performed at the corrected grid points.If it is within the limit, it is determined in step 27 whether the amount of deviation is within tolerance.

If the tolerance is out of step 27, the reticle XY stage is driven in step 27a, and the process returns to step 25.

On the other hand, if it is within the tolerance in step 27, focus adjustment and exposure are performed in step 28. After exposure, it is determined in step 29 whether or not it is the final shot. If it is not the final shot, step to the next shot is performed in step 29a. , return to step 25. When the exposure of the final shot is completed, the wafer is loaded in step 31, and the process proceeds to the next wafer.

According to the global alignment described above, when moving stepwise to each shot area, the chip θ and wafer θ are corrected in advance by a θ correction loop, and a correction grid is created based on measurement results in several shot areas. Since positioning is performed according to the data, a certain level of accuracy is always guaranteed. Furthermore, since the stage is moved according to pre-stored correction grid data, the speed is high and the throughput of the apparatus is high.

Furthermore, according to the die-pi-dia alignment described above, after performing θ correction and creating a correction grating,
Performing pi di-alignment. Therefore, since alignment has already been done in the θ direction, there is no need to look at it, and alignment only needs to be done in the X and Y directions, so alignment can be performed faster than normal die-pie-die alignment. In addition, since a correction grid has been created, the amount of movement during alignment is smaller than normal die-pie-dia alignment, and if an abnormal value is detected, a certain level of accuracy can be achieved by aligning according to the correction grid. Guaranteed.

In this embodiment, not only the wafer 〇 but also the chip θ are measured and averaged for alignment. This is particularly effective when the alignment mark is not placed off-center, that is, at a position where the straight line connecting the center of the chip and the mark is parallel to the X or Y direction. In other words, if the chip is off-center or measured with a single mark, the chip center position calculated based on the measurement results may shift, and this error will affect the correction value for wafer 〇, resulting in the correction. This is because the accuracy of the grid deteriorates.

In addition, in this embodiment, a predetermined shot area is measured using the TTL method instead of off-axis, so the error between the running direction of the XY stage and the off-axis microscope is the error of the correction grid. Nothing will happen.

Switching between global alignment and die-pie-die alignment can be performed automatically or can be specified manually. If this is done automatically, it can be determined based on the number of rejected shots due to measurement abnormalities, or the dispersion of measured values ((I4 difference). If there are many abnormal shots or deviations, use global alignment to average the position. However, if the number is small, it is better to use die-pie-die for more accurate alignment.

In this embodiment, after performing the θ correction, the process returns to the beginning and remeasures one shot. By performing such correction confirmation, more accurate θ correction can be performed, and subsequent measurements can be processed satisfactorily.

Furthermore, in order to improve accuracy, various methods can be considered for rejecting abnormal values of measured values. For example, it is better to compare it with a predetermined absolute value, to make a judgment based on the variance of the measured data value (for example, 3σ), or to make a judgment by a combination of these.Also, if there are an abnormally large number of abnormal value rejects, It is also good to re-measure automatically.

In this embodiment, the wafer θ is corrected by the wafer θ/S, and the chip θ is corrected by the reticle θ/S.

Therefore, highly accurate θ alignment is possible.

Next, a basic algorithm for selecting m shots, which are sample shots, will be explained.

First, the index Ind(n) of shot n is defined as the smallest number among the four when counting the number of shots from shot n up, down, left and right to the outermost circumference.For example, as shown in Figure 6(a) Then, Ind (n) =1.

The set of shots with index k is written as Ik.
Also! When the number of elements in , is a, write Ik=a. In Figure 6(b), ◎ is all elements of I1, ll-1
It is 8. Note that the index of the outermost shot is 0.

In addition, it is defined as follows.

(1) The given shot array is Io, It.

I2. . . . is composed of indices up to Ik.

(2) T is the total number of shots, T-I0+I, +...
...+Ik.

(3) S is global alignment or die-pie
This is the number of sample shots selected for die alignment.

(4) Pn is It when Pn' is a vector extending from the center of the shot array (not necessarily the shot center) to shot n. '1 Note that this is the case when P n"" (Xn, yn).

(5) el is a unit vector i=1.2 extending in each direction from the center of the shot array when 360° is divided into S equal parts.
......, S.

Next, general rules for selecting sample shots m to be measured will be explained.

■ First 1. Focusing on 11≧S, divide 360° into S equal parts, el, e2...
...Create an eg.

Next, for each eI (iml~S), the inner product elPj
(ixt~s, j=1.), and let j that gives the maximum value be the sample shot.

■ Calculation of the inner product is performed counterclockwise until a certain el
If there are multiple shots that give the same inner product, the most counterclockwise shot is taken as the sample shot.

■ If the same Pn is selected for multiple e's, only the sample shot for the first el calculated (i.e. 01 on the - table clock side) is adopted, and sample shots for other el's are suspended. , and continue calculating the inner product. This is to prevent double selection.

(2) If r+<s, all I1 are taken as sample shots, and the remaining S -1r shots are selected from I2. That is,
Replace S with I, -5, replace ■ with I2, and perform the processing from ■.

■ Perform the above calculation II-hI2-IO-413-...
...Proceed up to Ik.

However, shots where the distance from the center of the scimitar arrangement to the shot center is 20 mm or less are excluded.

(2) Let T' be the sum of the number of one-eye shots and the elements of an index set with four or less elements.

If T-T'<S, there are too many S, so the process is stopped as an error.

Next, selection of sample shot m in global alignment will be explained in more detail.

That is, sample shots are selected under the following conditions.

■ Exclude the outermost circumference to avoid abnormal offset check values. This is because the measurement accuracy is poor at the outermost circumference. Note that the offset check value is the amount of deviation of each shot in the X, Y, and θ directions calculated based on the measured values of the sample shots.

■ Translational (XY direction) component (S,,S,).

Rotation (θ direction) component (θ-1ey).

The precision (dispersion) of each expansion/contraction rate (β8.βy) should be approximately equal.

v (S, l') 4V (sy)■(θ
, )4V(θy) (βX) to V(βy) (2) The overall dispersion is made as small as possible under the following conditions. That is, w=v (s) +v (e) + v (β) is made to be the minimum.

From the conditions of ■ and ■, the i-th shot [[(Xt,
y5), then Σxl! Σ'j+ -ΣXt yI -0Σx1
2=Σy12 W′ wealth□ −□ −minimum Σx , 2 Σy , 2 . Therefore, it is best to select an arrangement that is symmetrical with respect to the origin and retains its shape even when rotated by 90 degrees from among shots that are as far away from the origin as possible.

From this, it can be seen that for the measurement shot, it is better to exclude the central area to the query and select a spot area that is point symmetrical with respect to the center.If the measurement shot is 0 point symmetrical, the center position can be calculated with high accuracy.

■ Consider stage accuracy and cross-correlation of offset check values, and create shot arrays that are as spatially dispersed as possible.

■ Even if there is a poorly measured shot or an abnormal shot,
Make sure there is no significant deterioration in accuracy.

By automatically selecting sample shots under the above conditions, averaged deviation needle position measurements can be obtained, and the accuracy in the θ correction in steps 3 to 9 in Figure 5 described above and those created in steps 10 to 19 can be obtained. The accuracy of the correction grid is guaranteed. In addition, all of the shots are automatically selected, so there is no reduction in throughput.Furthermore, even if the measurement in a certain shot is abnormal and you want to re-measure with an alternative shot, you can automatically select it in the same way as above, which is convenient. be.

[Effects of the Invention] As explained above, according to the present invention, the amount of positional deviation between the original plate and the substrate in a predetermined area on the substrate is measured by TTL, and the θ direction of the substrate is determined based on the measurement result. Since alignment is performed and then the θ correction described above is confirmed in the measured area or in an area other than the area, alignment can be performed with extremely high alignment accuracy and high productivity (high speed).

[Brief explanation of drawings]

, FIG. 1 is a schematic configuration diagram of a semiconductor exposure apparatus to which an alignment method according to an embodiment of the present invention is applied, and FIG. 2 shows an alignment mark and how the alignment mark is scanned with a laser beam. 3 is a diagram showing the arrangement of alignment marks in each exposure area on the wafer, FIG. 4 is a diagram showing the shot layout on the wafer, and FIG. 5 is a diagram showing the operation of the apparatus of the above embodiment. FIG. 6 is a schematic diagram for explaining the sample shot selection algorithm. ST: stage; Equipment, LT: Laser tube, PM: Polygon mirror, M: Motor, P1 Nibrism, BSl, BS2: Beam splitter, DR, DL: Photoelectric detector, CB: Control box, CON: Console. Patent Applicant Canon Co., Ltd. Agent Patent Attorney Tatsuo Ito Agent Patent Attorney Tetsu Ito Iroki 1 Figure (b) / VM 2nd FIJ I 3rd Figure Column (to empty) Figure 4 @ 6 Figure Procedure Amendment (Voluntary) June 23, 1986

Claims (1)

[Claims] 1. Prior to sequentially projecting the pattern drawn on the original onto a plurality of areas arranged on the substrate via a projection optical system, the above-mentioned The amount of positional deviation between the original plate and the substrate is measured in a predetermined area on the substrate, the substrate is aligned in the θ direction based on the measurement result, and then the alignment is performed again in the measured area or an area other than the area. An alignment method characterized by measuring and confirming the alignment.