JPH11329943A - Semiconductor production system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a semiconductor production system in which highly accurate alignment can be conducted while enhancing throughput. SOLUTION: In a semiconductor production system where a pattern on an original mask plate is projection aligned to wafer 5 mounted on an XY movable stage, an alignment mark on the wafer 5 is measured by means of a storage type alignment sensor 12. The stage is subjected to simultaneous control of movement in the measuring direction and nonmeasuring direction of the alignment mark. When the alignment mark moves under the sensor 12 and the stage is stationary in the measuring direction, storage of the sensor 12 is started if the mark is present within a predetermined range even during movement of the stage in the nonmeasuring direction and the position of stored alignment mark is calculated. Furthermore, the order for measuring a plurality of alignment marks on the wafer 5 is selected to increase such movements as the moving time of the stage in the measuring direction is shorter than the moving time in the nonmeasuring direction.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a position measurement function of a semiconductor manufacturing apparatus, particularly to position measurement.

[0002]

2. Description of the Related Art A conventional stepper position measuring method will be described with reference to FIG. 1 showing a configuration of a semiconductor exposure apparatus, FIG. 7 showing mark arrangement and measurement order on a wafer, and FIG. 6 showing a stage driving pattern. I do. XY stage (7)
A position measurement mark is provided on the wafer (5) registered in the above. The measurement mark in the X direction is arranged on a horizontal scribe line or the like (Mx in FIG. 7). The measurement mark in the Y direction is arranged on a vertical scribe line or the like (see FIG. 7).
My). The XY stage moves in the X and Y directions so that the alignment mark attached to the measurement shot position (S1 to S4) on the wafer shown in FIG. 7 can be observed by the alignment sensor (12), and stops the stage. While stationary, the alignment mark is illuminated by the light source (11), and the reflected light is received by the alignment sensor (12) to perform photoelectric conversion. FIGS. 7Wy and Wx show the photoelectrically converted signals. An accumulation type linear sensor or the like is used as the alignment sensor. The photoelectrically converted signal is A / D-converted at 101 and stored as a plurality of pieces of signal information in a memory (102) in the processing device. The stored pixel information is subjected to signal processing in an operation block (103), and the center of the position measurement mark is obtained from the result.

When the XY stage is driven, for example, from the Mx mark of the S1 shot to the My mark of the S2 shot, the driving becomes as shown in FIG. The movement amount in the Y direction is different from the movement amount in the X direction, and the driving distance X in the non-measurement direction is longer and the driving distance Y in the measurement direction is shorter than the My mark of the S2 shot.

The driving speed pattern in the non-measuring direction is shown in FIG.
(A). As a target position a position for resting the mark position, when the difference between to that position and dp _um, the position control pattern to the target position is in Figure 6 (b). In the driving, when the distance to the target position approaches a predetermined distance, position servo control is applied. After the start of the position servo control, the state of driving to the target position is shown by enlarging the vertical axis with a bold line. Similarly, FIG. 6C shows a driving speed pattern in the measurement direction, and FIG. 6D shows a position control pattern. The stage drive in the non-measurement direction and the measurement direction is simultaneously started, and when it is determined that both stages have reached the target positions and have stopped, the alignment sensor has started accumulating the image of the wafer mark.

[0005]

By the way, in recent semiconductor manufacturing, an ability (throughput) for processing as many wafers as possible in a given time is required. In addition, high-precision alignment accuracy is also required.

In the above conventional example, the alignment mark is not detected unless both the stage positioning in the measurement direction and the stage positioning in the non-measurement direction are completed. Therefore, the positioning stationary time t _{m in} the measurement direction and the non-measurement direction are not detected. _Assuming that the positioning time t _um and the accumulation time t _c , when t _m > t _um , t _m + t _{c When} t _m <t _um , a time of t _um + t _c is required.

As a measure for improving the throughput, a method has been proposed in which a mark on a wafer is detected by an alignment sensor without stopping the stage. Two such examples are as follows: 1) Mirror scan method This is a method in which a mirror in the optical path is scanned so that the movement of the stage can be canceled with respect to a fixed alignment sensor. 2) Stage scan method Light is applied to the mark on the wafer, and the returning light is detected by a photo sensor. This method determines the position of the mark by referring to the peak position of the light and the position of the stage. Generally, scattered light at a mark edge is detected in order to capture the peak position of the mark. However, both methods have problems. In the mirror scan method, driving such as rotation of a mirror must be performed in synchronization with movement of the stage.
However, drift occurs in the deflection angle of the mirror, and the position of the alignment mark shifts, resulting in drift of the measurement position. Synchronizing the drive of the stage and the mirror also involves a problem that complicated control is required such as a change in stage speed. In addition, since the optical path length changes by rotating and scanning the mirror with respect to the sensor, a defocus, a change in magnification, and a change in signal intensity occur on the sensor surface, and a complicated optical system for imaging a correct alignment mark. A system is required.

In the stage scanning system, the peak position of the scattered light must correspond to the measurement position of the interferometer accurately. However, the measurement resolution of the interferometer is limited. Furthermore, if the sampling interval of the interferometer is constant, it becomes impossible to match the peak position with the sampling position. Therefore, the resolution of the interferometer, sampling errors, and detection signal errors are included.

In the stage scan method, the position of the signal of the scattered light due to the shape distortion of the mark edge, the low step structure, the ultra-low step structure subjected to the CMP process, the unevenness of the resist coating applied to the upper surface of the mark, and the like. And the intensity are also easy to change, and may not be detected correctly.

In order to improve the measurement accuracy, the number of measurement marks is increased to a plurality. However, in the stage scan, the position of each mark is measured and the stage is scanned and detected by a photo sensor. for that reason,
As described above, an error of the interferometer occurs at each mark position, and an error of the measurement mark remains.

[0011]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems of the prior art, the present invention provides a semiconductor manufacturing apparatus for projecting and exposing a pattern on an original mask to a wafer mounted on an XY movable stage. In, the positioning mark on the wafer is measured by an accumulation type alignment sensor, the stage simultaneously controls the movement in the measurement direction and the movement in the non-measurement direction with respect to the positioning mark, and the positioning mark is located below the alignment sensor. Move, if the stage in the measurement direction is stationary, if there is a mark in a predetermined range even while the stage in the non-measurement direction is moving, the accumulation of the alignment sensor is started, and the accumulated The position of the positioning mark is calculated. Here, the measurement direction and the stage position tolerance in the non-measurement direction are set, and the tolerance in the non-measurement direction is
It can be larger than the tolerance in the measurement direction.

Further, while the alignment sensor is accumulating the mark image, the position of the stage in the mark measurement direction and / or the stage minute rotation angle is continued to be averaged over the accumulation time to determine the average position of the stage during accumulation. ,
The average position can be reflected on the calculated positioning mark position.

[0013] Further, the alignment mark is constituted by at least one mark, and is provided in the X direction or the Y direction.
It is possible to measure at least one of the directions.

Further, according to the present invention, there is provided a semiconductor manufacturing apparatus for projecting and exposing a pattern on a mask original to a wafer mounted on an XY movable stage.
The order of measuring the plurality of positioning marks on the wafer is selected so that the movement of the stage in the measurement direction becomes shorter than the movement time in the non-measurement direction.

The procedure for aligning the semiconductor device of the present invention will be described below. In order to measure the alignment mark with high precision without stopping the stage by the global alignment method, a storage type sensor is used as the alignment sensor. The stage is moved while measuring the X and Y measurement marks on the wafer. If the movement distance in the non-measurement direction is long, the stage position in the non-measurement direction changes the image of the alignment mark in the measurement direction after the movement in the measurement direction is completed. After entering the allowable range for accumulation, accumulation of sensors is started.

By simultaneously picking up images of a plurality of marks by the accumulation type linear sensor, there is no stage position error between the measured mark positions. By providing the accumulation time, the mark images taken through the optical path are averaged, and are reduced to a level at which, for example, optical fluctuation does not affect the measurement.

While the alignment sensor is accumulating light, the position of the stage in the measurement direction and the minute rotation angle of the stage are stored and averaged a plurality of times and averaged and reflected on the mark measurement value, so that the mark of the fluctuation of the stage position during mark imaging is recorded. Eliminate the effect on measurement. The accumulation of the alignment sensor and the storage of the position of the stage are synchronously controlled.

Since the scanning of the stage and the optical system is not performed in the measurement direction of the mark, the above-mentioned problem of the mirror scan and the stage scan is not affected.

Immediately after the sensor accumulation is completed, the XY stage is driven to the next measurement mark position. The position of the alignment mark and the amount of fine movement of the stage during accumulation are calculated during the movement of the stage. In global alignment, the mark position is determined accurately using the position of the alignment mark on the stage + the amount of deviation of the alignment mark actually obtained by the sensor + the difference from the target position of the stage being accumulated, and the shift of the wafer is performed. , Rotation and magnification, and calculate the center position of the exposure shot.

The difference between the target position of the stage during accumulation and the target position will be described with reference to FIG. The difference from the target position is 2
The synthesis of the components. (1) The difference between the stage position and the stage target position where the alignment sensor stores the mark signal during accumulation. (2) A stage minute rotation component generated from the stage minute rotation angle θ component stored while the alignment sensor stores the mark signal. If the rotation center of the minute rotation is on the mark, L in FIG. 2 is 0, and the difference from the target position generated by the rotation component is 0. If the distance L between the rotation center and the mark position is known in advance, the minute rotation component can be calculated.

The stage movement pattern will be described with reference to FIG. 3. A drive pattern to a target position in the measurement direction is appropriate for the positioning time shown in FIG. 3 (d). On the other hand, the speed pattern in the non-measurement direction is shown in FIG. When reaching the allowable range in which accumulation is possible in the non-measurement direction, an accumulation enable signal to the sensor is generated. The length of the normal mark in the non-measurement direction is slightly shorter than the width of the scribe line.
m. Since the above-mentioned allowable range is possible even when the mark becomes half in the non-measurement direction, it can be generated 30 μm before. The time required to enter the allowable range in the non-measurement direction is defined as t _{um '} .

Then, as compared with the conventional method, t _m > t
If _{um '} , the time can be shortened by t _um -t _m . If t _m <t _um , the time can be reduced by t _um −t _{um ′} . t _m> t _um if not a shorter time.

By devising the measurement order of the alignment marks, the time can be further reduced. It is the above t _m
That is, the order of measurement is selected so as to reduce the case where> t _um is obtained.

As shown in FIG. 7, S1My → S1Mx → S2
My → S2Mx → S3My → S3Mx → S4My → S4
As shown in FIG. 4, the measurement order of Mx is S4Mx → S1Mx → S1.
My → S2My → S2Mx → S3Mx → S3My → S4
By setting My, the stage drive that _satisfies t _m > t _um is reduced.

[0025]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 A first embodiment of the present invention will be described with reference to FIG. The alignment mark is on the wafer (5) mounted on the XY stage (7). Light emitted from the light source (11) is guided to a half mirror (10) by a fiber or a dedicated optical system, passes through the mirror (13) and the reduction projection lens (4), and illuminates the position measurement marks (Mx and My in FIG. 7). I do. The image of the alignment mark is reflected by the wafer (5), passes through the same path again, passes through the half mirror (10), and hits the alignment sensor (12).

The alignment sensor (12) uses a storage type sensor or the like. A CCD linear sensor, a CCD array sensor, or the like is suitable as the accumulation type sensor, but any sensor having a plurality of pixels capable of controlling the start and end of accumulation can be used. The image of the alignment mark received by the sensor is photoelectrically converted. At this time, the time for accumulating light is controlled so as to be a predetermined time depending on the sensor driving device CLK. Also,
The control time is transmitted from the stepper control device HP to the alignment processing device AP and set to CLK. The timing for accumulating light is specified by the stage processing device SP. The signal photoelectrically converted by the alignment sensor is 10
The digital signal is A / D converted at 1 and stored in a memory (MEM) in the processing device as a plurality of digital signal information. From the stored digital signal, the mark center is calculated by the alignment processing device (AP).

A plurality of alignment marks are arranged on the wafer (5). This is shown in FIG. S1 in FIG.
S4 represent alignment mark measurement shots in the case of global alignment for measuring four alignment marks on the wafer. The measurement direction of each mark is indicated by an arrow. The number of shots is not limited to four. The XY stage measures mark positions sequentially from S4Mx. For example, to move from S1MX to S1My, it is necessary to accelerate and decelerate and stop at the mark position (target position) in the shortest time. The stage is moved by the stepper controller H
P instructs the target position (coordinates) to the stage processing device SP of the stage control device 200, drives the motor (MOT), applies the servo while referring to the interferometer (PM), and moves to the target position accurately.

When the stage reaches the target position and stops at the mark position on the wafer and a mark image can be picked up, an accumulation enable signal is sent to the sensor driver CLK to start accumulation of the mark. You.

A pattern in which the stage moves from S1MX to the S1My mark position (target position) will be described with reference to FIG. This pattern is a case where the movement amount in the non-measurement direction is longer than the movement amount in the measurement direction. FIG. 3A shows a moving speed pattern of the stage in the non-measurement direction. The vertical axis represents speed (v _um ) and the horizontal axis represents time (t). At this time, the residual to the target position is shown in FIG. Vertical axis residuals (dp _um)
It is. In this case, the non-measurement direction is the X direction. On the other hand, a moving speed pattern in the measurement direction of the stage is shown in FIG. The vertical axis is velocity (v _m) horizontal axis indicates time (t). At this time, the residual to the target position is shown in FIG. The vertical axis represents the residual (dp _m). The measurement direction is Y.

Regarding the measurement direction, an accurate position cannot be detected unless the stage is stationary. The tolerance (allowable range) that enables the alignment mark on the wafer to be measured with high precision is set in the stage processing apparatus SP as tolm. The tolerance is an allowable range of a difference between the position of the stage and the target position. When the difference from the target position of the stage is within ± tolm, the measurement direction is assumed to be stationary. The amount of tolm is usually less than the pixel resolution of the sensor.

In the non-measuring direction, the stage does not need to be stationary. If the sensor is within the observable range, the sensor can capture a good mark image. Tolum 'is set in advance in the SP as the tolerance in the non-measurement direction with the above range. The size of the volume “tol ′” is set to about の of the width of the mark in the non-measurement direction. Normally, the scribe line is marked, so it is about 30 _um .

In global alignment, the position of the measurement mark in the measurement direction is important, but the position in the non-measurement direction is not important because the length of the mark in the non-measurement direction is short.

The accumulation of the alignment sensor can be performed if the stage position in the measurement direction falls within tolm and the stage position in the non-measurement direction falls into tolm '. (E) of FIG.
Shows the accumulation timing. The stage control in the non-measurement direction is controlled so as to have any one of the following patterns when it reaches to volume '.

Three types of drive patterns in the non-measurement direction are selected. 1) Do not perform stage deceleration control. Mark existence range t
Even if it reaches olum ', the speed of the stage does not change. When the direction of the next measurement mark in the non-measurement direction does not change. The fastest measurement is possible. 2) When the mark existing range turum 'is reached, deceleration is stopped and driving is performed at a constant speed. When the alignment sensor accumulation time is long and the distance moved during the accumulation time is shorter than the mark width. 3) After reaching the mark existence range torum ',
Continue deceleration and stop.

When the direction of the stage needs to be changed depending on the position of the mark to be measured next. However, it is not necessary to precisely servo the position for stopping. The selection of the stage drive in 1), 2), 3) is determined optimally by the stepper control device HP and instructed to the stage processing device SP because the relationship between the mark positions is calculated in advance. When accumulating marks by the method of 1), if the stage speed is too high, the distance moved during the accumulation time may exceed the observable range.
In that case, the speed is decelerated in front, and the method 2) is adopted. When it reaches tol ', the SP sends a storable signal to CLK.

The SP samples the stage position in the measurement direction at the SP immediately after sending the accumulation enable signal to the CLK, and monitors and stores the fine movement and drift of the stage during the accumulation period. Furthermore, the drift of the minute rotation angle of the stage also accumulates.

When CLK reaches a predetermined accumulation time, the completion of accumulation is notified to the stage processing device SP by a signal, and the monitoring and storage of the stage position in the measurement direction are terminated. In the SP, the average position of the stage and the average stage angle in the measurement direction during the accumulation by the alignment sensor are calculated. The alignment processing device AP processes the mark image stored in the memory and calculates the position of the mark. It instructs the stepper processor HP that the accumulation has been completed. By doing so, it is possible to move to the next measurement mark position.

The above processing is repeated for all measurement marks, and global alignment is performed. In global alignment, the following information is provided for each of a plurality of mark positions. The average position of the measured mark position (target position) measured mark position deviation amount measured mark
The shift, rotation and magnification of the wafer are calculated. When the movement in the non-measurement direction is shorter than the movement in the measurement direction, conventional stage control is performed.

Next, the flow of processing will be described with reference to the flowchart of FIG. Since the processing device is divided into HP, AP, and SP, a flow of parallel processing is provided. It is assumed that the measurement mark S1Mx on the wafer moves from S1Mx to S1My. It is assumed that the movement amount X in the non-measurement direction is longer than the movement amount Y in the measurement direction.

The HP sends the Y measurement mark coordinates to the SP,
Instruct to move to the position to be measured (HPS00
1). When receiving the alignment sensor accumulation enable signal from the SP to the AP, the HP starts accumulating the mark image, and issues an instruction to execute a process of calculating the mark position after the accumulation is completed (HPS0).
02). After issuing the instruction, it waits until the accumulation is completed (HPS003).

The SP receives an instruction from the HP (SPS00
1) The movement of the X stage (non-measurement direction) and the movement of the Y stage (measurement direction) are started (SPS002). Control is continued until the position in the measurement direction falls within the tolerance (SPS003). Further, the control is continued until the stage position in the non-measurement direction enters the measurable range (non-measurement direction tolerance) (SPS004).

When the position in the measurement direction stops and the non-measurement direction reaches the measurable range, the stage speed in the non-measurement direction is maintained at a constant speed, and it is determined that the alignment sensor can be accumulated.
Signal to AP via LK (SPS00
5).

Upon receiving the accumulation permission signal from the SP via CLK (APS002), the AP accumulates the mark image for a predetermined time (APS003). At the same time, the SP monitors and stores the stage position in the measurement direction with the output of the interferometer while the AP accumulates the mark image (SPS006). The storage of the stage position in the measurement direction is continued until it is confirmed that the accumulation is completed (SPS007).

When the mark image accumulation of the alignment sensor is completed, the AP sends a signal to the SP and HP via CLK to calculate the center position of the mark image (APS004).
On the other hand, if the end of the accumulation can be confirmed (SPS007), the SP obtains the average position of the stage during accumulation and sends the position data to the HP (SPS008).

When the HP confirms that the accumulation has been completed (H
PS003) Store the mark position of the AP, the average position of the stage of the SP, and the target position of the measurement mark instructed to the SP. The above is the 1-mark measurement flow. When one mark can be measured, if there is still a mark to be measured, processing for moving to the next mark is performed (HPS00
5). The AP and the SP wait for the next instruction from the HP (APS
006, SPS009).

In this embodiment, the observation is made with a TTL scope, but any method such as an off-axis method or a TTL method may be used.

Second Embodiment The second embodiment has the same apparatus configuration and stage control flow as the first embodiment, and a description thereof will be omitted. When global alignment measurement is performed by the method described in the first embodiment, high-speed processing can be performed. Therefore, a description will be given of the order of measurement of the alignment marks for faster processing. As described in the first embodiment, when the movement distance in the measurement direction is shorter than the movement distance in the non-measurement direction, the timing at which the alignment sensor 12 starts accumulation is earlier.

In the conventional measurement order shown in FIG. 7, the non-measurement direction is longer than the measurement direction in two steps of S1Mx → S2My and S3Mx → S4My, and the measurement time is reduced by 2/7.

The reason why the non-measurement direction is longer than the measurement direction in the measurement order in FIG. 4 is as follows: S4Mx → S1Mx, SlMx → S1My, S1My →
S2My, S2Mx → S3Mx, S3Mx → S3My,
This is a six-step process of S3My → S4My. When the measurement order shown in FIG. 4 is adopted, the measurement time can be reduced by 6/7, and the effect three times as large as that of the related art can be obtained.

(Embodiment of Device Manufacturing Method) Next, an embodiment of a device manufacturing method using the above-described semiconductor manufacturing apparatus will be described. FIG. 8 shows a micro device (IC or LSI)
2 shows a flow of manufacturing a semiconductor chip such as a liquid crystal panel, a CCD, a thin-film magnetic head, a micromachine, and the like. In step 1 (circuit design), a device pattern is designed. Step 2 is a process for making a mask on the basis of the designed pattern. On the other hand, in step 3 (wafer manufacture), a wafer is manufactured using a material such as silicon or glass. Step 4 (wafer process) is called a pre-process, and an actual circuit is formed on the wafer by lithography using the prepared mask and wafer. The next step 5 (assembly) is called a post-process, and is a process of forming a semiconductor chip using the wafer produced in step 4, and includes processes such as an assembly process (dicing and bonding) and a packaging process (chip encapsulation). including. In step 6 (inspection), inspections such as an operation confirmation test and a durability test of the semiconductor device manufactured in step 5 are performed. Through these steps, a semiconductor device is completed and shipped (step 7).

FIG. 9 shows a detailed flow of the wafer process. Step 11 (oxidation) oxidizes the wafer's surface. Step 12 (CVD) forms an insulating film on the wafer surface. Step 13 (electrode formation) forms electrodes on the wafer by vapor deposition. In step 14 (ion implantation), ions are implanted into the wafer. Step 15
In (resist processing), a photosensitive agent is applied to the wafer. Step 16 (exposure) uses the above-described exposure apparatus to print and expose the circuit pattern of the mask onto the wafer. Step 17 (development) develops the exposed wafer. In step 18 (etching), portions other than the developed resist image are removed. In step 19 (resist stripping), unnecessary resist after etching is removed. By repeating these steps, multiple circuit patterns are formed on the wafer. By using the production method of this embodiment, it is possible to manufacture a highly integrated device, which was conventionally difficult to manufacture, at low cost.

[0052]

As described above, according to the present invention,
Since the processing time for measuring the alignment mark can be reduced, the throughput can be improved. The alignment mark is a multi-mark, and the position of each mark does not include an error of the stage and the optical system. In addition, the movement of the stage during accumulation is averaged, and the fluctuation can be reflected in the global alignment, so that not only the throughput but also the alignment accuracy is improved, and the yield of semiconductor manufacturing is improved.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a semiconductor exposure apparatus illustrating an embodiment of the present invention.

FIG. 2 is a diagram illustrating a difference from a target value of a stage during accumulation.

FIG. 3 is a diagram showing a driving pattern of a stage according to the present invention.

FIG. 4 is a diagram illustrating mark arrangements and measurement orders on a wafer for explaining the first and second embodiments.

FIG. 5 is a flowchart illustrating a flow of a process according to the first embodiment.

FIG. 6 shows a driving pattern of a stage for explaining a conventional example.

FIG. 7 illustrates a mark arrangement on a wafer and a measurement order for explaining a conventional example.

FIG. 8 is a diagram showing a flow of manufacturing a micro device.

FIG. 9 is a view showing a detailed flow of a wafer process.

[Explanation of symbols]

1: exposure light source, 2: shutter, 3: reticle, 4: reduction projection lens, 5: wafer, 7: XY stage, 10: half mirror, 11: illumination light source, 12: alignment sensor, 100: mark position calculation device, 200: stage controller, AP: alignment processor, SP: stage processor, HP: stepper controller, Mx, My: alignment mark.

Claims (6)

[Claims] 1. A semiconductor manufacturing apparatus for projecting and exposing a pattern on a mask original plate onto a wafer mounted on an XY movable stage, wherein a positioning mark on the wafer is measured by an accumulation type alignment sensor, and the stage is used for the positioning. The movement in the measurement direction and the movement in the non-measurement direction are simultaneously controlled with respect to the mark, and the positioning mark moves below the alignment sensor. If the stage in the measurement direction is stationary, the stage in the non-measurement direction moves. In particular, when a mark exists within a predetermined range, accumulation of the alignment sensor is started, and a position of the accumulated positioning mark is calculated. 2. The semiconductor manufacturing apparatus according to claim 1, wherein a stage position tolerance in the measurement direction and a non-measurement direction is set, and the tolerance in the non-measurement direction is larger than the tolerance in the measurement direction. 3. While the alignment sensor is accumulating a mark image, the position of the stage in the mark measurement direction and / or the stage minute rotation angle is continued to be referred to, averaged over the accumulation time, and the average stage position during accumulation is determined. 2. The semiconductor manufacturing apparatus according to claim 1, wherein the average position is reflected on the calculated positioning mark position. 4. The semiconductor manufacturing apparatus according to claim 1, wherein the alignment mark comprises at least one mark and can measure at least one of the X direction and the Y direction. 5. A semiconductor manufacturing apparatus for projecting and exposing a pattern on a mask original to a wafer mounted on an XY movable stage, wherein the order of measuring a plurality of positioning marks on the wafer is determined by moving the stage in the measurement direction. The semiconductor manufacturing apparatus is selected so that the number of movements whose time is shorter than the movement time in the non-measurement direction is increased. 6. A device manufacturing method, comprising manufacturing a device using the semiconductor manufacturing apparatus according to claim 1.