JPS61131444A - Alignment device - Google Patents

Abstract

PURPOSE:To realize an extremely precise alignment by a method wherein a quotient is obtained, together with a surplus, by dividing the quantity of movement by the resolution factor of a pulse motor, the quotient is the portion that is to be covered by a pulse motor, and the surplus is the portion to be covered by another trimming means. CONSTITUTION:With a pulse motor ZW provided with a migration resolution of 2mum, a microprocessor 40Z supplies a quantity of migration DELTAd1 consisting of some units of 2mum to a register 41Z for migration up the axle Z of a wafer. As the result, the wafer surface position is now within approximately 2mum from the focal plane position. Here, the distance to the wafer surface is measured again. When air sensor nozzles AG1-AG4 have measured distances d9-d12, the microprocessor 40Z this time gives to a register 43Z an instruction DELTAd2=d0-(d9+d10+d11+d12)/4, regarding the direction and quantity of the migration of a piezoelectric element PZ. The register 43Z stores the instruction and at the some time, outputs the same to a digital analog converter 44Z and a piezo-electric element driving voltage generating circuit 46Z.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field] The present invention relates to an apparatus for printing or exposing circuit patterns used in the manufacture of high-density integrated circuit chips for semiconductor memories, arithmetic devices, etc., and particularly to an alignment apparatus.

[Prior Art J] Conventionally, this type of device has had drawbacks such as overlay accuracy, accuracy, productivity, flexibility with other devices, and large compound cone formation.

[Objective] The present invention solves the above-mentioned difficulties and achieves extremely high overlay accuracy,
The object is to provide an apparatus and method with high productivity (high speed), high flexibility, and simple configuration.

[Example] An example of the present invention will be described below with reference to the drawings.
- to do. FIG. 1 shows a circuit pattern surface CP formed within the surface of a circuit pattern mask, so-called reticle RT, on a wafer W.
FIG. 2 is a schematic configuration diagram of an exposure device for exposing onto F. 1
0 is the light source system for exposure, and the light source such as an ultra-high pressure mercury lamp -P
An ellipse I first is placed near the light source LP to effectively condense the luminous flux emitted from the light source LP, and then sequentially along the optical path to transmit most of the infrared light and reflect the ultraviolet light. cold mirror M2, integrator lens system Ll for making the distribution characteristics of the luminous flux uniform, shutter ST lens system L2, reflector M3, lens system L3, light shielding device 8m, lens system L4, reflection RM4, lens system 15. Reflector MS,
A lens system L6 and a reticle RT are sequentially arranged along the optical path, and reflections IM3, M4, and MS are
Each lens system L3 is for bending the optical axis at right angles to make the illumination system more compact, and the lens system L3 is for condensing the light from the light source LP to uniformly illuminate the light shielding device BL. . As is the so-called TTL (Through Th
elens) optical system for alignment, R8 is a drive stage for the reticle RT in the X, Y, and θ directions, and reduction lens system PO is an optical system for reduction exposure, 115 to 1/10
It has a reduction rate of OA is an off-axis optical system for alignment of wafer WF, and WS is x, y of wafer WF.
, z, θ directions, LZ is a laser interferometer, mirror M6 of reduction lens system PO and wafer stage W
The movement of wafer stage WS is controlled by mirror Ml of S.

FIG. 2 is a perspective view of the light shielding device BL. This light shielding device BL includes lens systems L4 and L5 shown in FIG.

With L6, the light-shielding surface is arranged in a conjugate relationship with the circuit pattern surface CP of the reticle RT, and the reticle RT is replaced with a reticle whose transparent part, such as glass, has a different thickness, and the refractive power changes. In this case, it is possible to move in the optical axis direction in order to maintain the above-mentioned conjugate relationship.

In addition, a light shielding device rotating mechanism (not shown) rotates in the θ direction in conjunction with the rotation of the reticle RT, as shown in FIG. It is possible to move.

The light shielding device BL includes four pulse motors P1 to PM4 on the substrate ST, four feed screw parts FG1 to FG4 which are fixed to the rotating shaft of each motor PM and are rotatable, each motor PM and the feed screw part FG. Four feed nut parts NAI to NA4 that are movable in one direction by the rotation of , and four light shielding plates 811 to 8L4 that are fixed on the feed nut part NA and have sharp side edges (edges) d1 to d4. are respectively arranged, and the side edges d1 to d4 of 41 constitute a rectangular opening.

In the above configuration, the light beam emitted from the light source LP in FIG. 1 is transmitted to the optical element M1 when the shutter ST is opened. M
2. L 1. M3. L 2. The light is reflected and refracted in the order of M4, and the light shielding device 8L is uniformly illuminated. The light beam irradiated to the outside of the opening of the light shielding device 8L is transmitted through the four light shielding plates B.
The light flux that is blocked by L1 to BL4 and passes through the opening illuminates the circuit pattern surface CP of the reticle RT, as shown by the solid line in the figure. Here, the light-shielding surface of the light-shielding device BL and the pattern IIcP of the reticle RT are arranged in an optically conjugate relationship, so that the edge of the opening of the light-shielding bag fBL has a clear outline on the circuit pattern surface CP. is projected with
The area outside the circuit pattern surface CP of the reticle RT can be completely shielded from light.

Note that the light shielding device BL is located on the circuit pattern surface C of the reticle RT.
It is rotatable in the θ direction (Fig. 2) for rotational alignment with P, and when the reticle RT is changed to a reticle with a different thickness, that is, a different refractive power, the above conjugate relationship It is movable in the optical axis direction (up and down arrows in FIG. 1) to maintain the Furthermore, the area and position of the opening of the light shielding device BL are determined by the pulse motors PMI to MP4 being rotated in a predetermined manner according to a signal output from the electronic processing section of this device.
Feed screw FG "~FG4" connected to the shaft of each motor
By moving the feed nuts NA1 to NA4 in one direction and thus moving the light blocking plates BLI to BL4, the area of the opening in the direction perpendicular to the optical axis can be arbitrarily changed. The movement and adjustment of the four light shielding plates BLI to BL, 4 can be performed at the same time, thus making it possible to expose an arbitrary area of the reticle RT.

FIG. 3 is a cross-sectional view of reticle stage R8 in FIG. 1, and FIG. 4 is a schematic top view thereof. In the figure, the board S
L is fixed to the reduction lens PO, a part of which protrudes upward, and reticle reference marks RKR, RKL are provided on its upper surface. This reticle reference mark RKR, RKL
The set marks R8R and R8L of the reticle are aligned to perform so-called reticle alignment.

RY is the Y-direction drive stage of reticle RT, RX, Rθ
is also a drive stage in the X direction and θ direction, θB (θ3
1. θ82. θ83) is the guide bearing for the rotation stage Rθ, RC is the reticle chuck and the tube TU
It has the function of attracting and fixing the eight reticles to the chuck RC using the suction force from R. RX, RY, and Rθ are pulse motors for driving each stage RX, RY, and Rθ, XL, YL, and θL are drive force transmission lever gears, and BXI, BX2, BYl, BY2, and Bθ are motors that resist the force of the motors. One-sided spring, XS.

YS is a guide block that cooperates with guide bearings XG and YG, respectively. The desired I is determined by the rotation of each motor.
Each stage is driven in the X, Y, and θ directions. For example, if you now rotate the pulse motor PX in the direction of the arrow, lever XL
rotates in the direction of the arrow, and its left end pushes the X stage RX,
X stage RX moves to the left against spring BX1.8X2. At this time, it is correctly moved only in the X direction by the guide block xB and guide bearing XG. Photo sensor PS and light shielding plate SM are detection systems for aligning the movement limit of reticle stage R8 and the center of reticle stage R8 with the optical axis of exposure lens system PO.

Further, when the pulse motor PO is driven and rotated, the substrate ST of the light shielding device BL in FIG. 2 is also rotated in the θ direction in conjunction with the rotational force transmission system DT shown by the dotted line in FIG.

FIG. 5 FIJ is a diagram schematically showing the TTL alignment optical system AS and off-axis optical system OA of FIG. 1. In the figure, 18 is a laser source, 2S is a condensing lens for focusing the laser system, 3S is a rotating polygon mirror, and 4S is an r-
The θ lens and 5S are beam splitters. The laser beam emitted from the laser source 1S is scanned as the rotating polygon 138 rotates, and enters the optical system below the beam splitter 5SIX, where 6S is a field lens, IS is a field dividing prism, and the prism IS is a scanning prism. Splits the laser beam into two optical paths. In this respect, the prism IS can be referred to as a field and space dividing prism. 8R, 8
L is a polarizing beam splitter, 9R and 9L are relay lenses, and 10R and IOL are beam splitters.The light reflected or passed through these elements enters objective lenses 11R and 111, is reflected by objective mirrors 12R and 12L, and is sent to the reticle. R
An image is formed on T and scanning is performed. Imaging lenses 13R, 13
The system from L to the photoelectric detectors 18R and 18L is a photoelectric detection system.

14R and 14L are color filters, and 15R and 15L are spatial frequency filters, which serve to block regularly reflected light and extract scattered light for photoelectric detection. 16R16L is a reflector, 17
R and 17L are condenser lenses. Light source 19R,
19L, condenser lenses 20R, 20L, and color filters 21R, 21L constitute an illumination optical system for observation,
Erecta 22S1 prism 23S, TV lens 24
S and image pickup tube CDo constitute an observation system.

In this example, in order to effectively use the front beam, the optical path of the scanning laser beam is divided into left and right sides by a field splitting prism IS placed on the conjugate plane of the reticle and wafer. The scanning line is orthogonal to the 'M line of the field dividing prism IS. In other words, the mirrors 10R, 1 are used to scan the laser in the vertical direction.
0L, 12R, and 12L are used.
- Also, as shown in the figure, since the left and right scanning optical systems are asymmetrically configured, the left and right objective lenses 11
R, 11L are alignment marks WR of reticle RT,
The mushrooms are arranged alternately in correspondence with the positions of WL and mushrooms. OAR and OAL are a pair of off-axis optical systems for alignment, and are used for alignment operations to be described later.

CR and CL are high-magnification photoelectric high-resolution image pickup tubes, CDR.

CDL is a converter for low magnification and consists of C, CD (charge coupled device), etc.

FIG. 6 is a partial perspective view of the wafer stage WS shown in FIG.
A direction moving stage Wx is mounted, and each X and Y stage W
X and WY are controlled by servo motors XM and YM, respectively.
Move along the guides GX and GYIC in the direction. XH-X
S and YH-YS are detection systems for initial reset of the X and Y stages, respectively. XO is a hole for a stage θ2 that rotates in the θ direction and moves up and down in the Z direction. A wafer chuck WC is mounted on the stage θ2 as shown in FIG. 7, and the wafer WF is suctioned and fixed thereon in the same manner as on the reticle side. The stage θ2 is fitted into the stage holder θH, and can be moved up and down in two directions and rotated in the θ direction by ball bearings BB and bl. Stage holder θH
is fixed to the X stage WX as shown. The pulse motors ZM and θM for driving in the Z and θ directions are connected to the stage holder θ.
Fixed at H.

A piezoelectric element PZ in which a large number of donut shapes are stacked is arranged at the center of the stage θZ. The piezoelectric element Pz, stage θZ, and wafer chuck WC are integrated with screws BW.

Is is an eddy current type position sensor fixed to the base ZD of the piezoelectric element PZ, and detects the upward movement Ha of the stage θ2. A lever ZL is further rotatably supported on the stage holder θH, and a nut N is fixed to the stage holder θH. Z direction drive mode 9
When 2M is driven, gear G1. G2 rotates and screws G3
When the lever ZL descends while rotating downward, the right end of the lever ZL is pushed by the ball b1 and rotates clockwise, and the left end of the lever ZL is connected to the piezoelectric element PZ and the sensor Is via the ball b2.
is pushed upward through the base ZD, so the integrated stage θZ and wafer chuck WC move upward, that is, Z
move in the direction. In this way, coarse movement for focusing is performed. Thereafter, the piezoelectric element Pz is driven from that position and extends in the Z-axis direction using the lever ZL as a fulcrum. Therefore, the wafer chuck WC and the θZ steady move upward by the extension valve of the piezoelectric element. The amount of movement is detected by measuring the gap Q using the sensor Is. This provides fine adjustment.

When the θ direction drive motor θM is driven, the gear trains G4, G
5, stage θZ and wafer chuck W via G6
C rotates smoothly by ball bearings 8B and b2.

OH, G3 is a detection system that determines a reference point in the rotation direction.

A G attached to the reduction projection lens PO 1. A G
3 is an air micro sensor nozzle, and AG2 (not shown)
．． The distance to the surface of the wafer WF is measured using, for example, four devices including AG4. If the distances from the end face of the reduction projection lens PO to the surface of the wafer WF measured by the nozzles AG1 to AG4 are respectively d1 d2, d3, and d4, then the average distance is (di + d2 + d3 + d4)/4. If the distance between the imaging plane position of a predetermined reduction projection lens PO and the end face of the reduction projection lens PO is do, then in order to move the wafer WF to the imaging plane position, ΔcJ=do −(dl +d2 '+d3 +d4
)/4, which is the amount Δd, by moving the wafer 2m structure.

As a result, the average plane of the wafer WF becomes the imaging plane position.

FIG. 8 is a block diagram for controlling the automatic focusing mechanism.
The microprocessor 402 performs various judgment processes,
Issue instructions appropriate to each case. A register 412 stores command information such as two rotational directions, two rotational amounts, and two rotational speeds from the microprocessor 402 to the pulse motor ZM. 42Z is a pulse motor control circuit, and register 4
Based on the movement amount command information of 1Z, continuous loop control of pulse motor ZM is performed. In the initial state, the wafer W
The surface position of F is, for example, two or more distances away from the imaging plane position. This is to prevent the wafer WF from colliding with the reduction projection lens PO even if the thickness of the wafer WF is greater than the specified thickness. In addition,
The range in which air sensor nozzles AG1 to AG4 can be accurately measured is when the distance from the end face of the nozzle to the wafer surface is within about 0.2 m. Therefore, the predetermined imaging plane position is 0.1 m+ from the end face of the nozzle. Assuming this, accurate measurements can only be made after the wafer surface moves upward and is within 0.1 mm below the imaging plane position.

49Z is a circuit that converts changes in the fluid flow rate of air sensor nozzles AGI to AG4 into voltage, and a reduction projection lens P.
The distance between O and the wafer surface di, d2.

d3, (141,: vs. 125'L/ta voltage output V1.V2, V3.

Generates V4. 50Z is an analog-to-digital converter (A
DC), and the voltage V1 generated in the voltage conversion circuit 49Z
, V2, V3, and V4 are converted into digital signals and sent to the microprocessor 402. Here, since the initial position of the wafer WF is more than 2 m away from the imaging plane position, the microprocessor 4oz sends a movement command to the pulse motor ZM to the register 41Z until the wafer Z axis rises and enters the measurement range of the air sensor nozzle. Continue to give. Pulse motor Z
When the wafer Z-axis rises due to the rotation of M and the wafer comes within 0.1 of the image forming surface position, the air sensor nozzle AG1
~AG4, the voltage conversion circuit 49Z, and the analog-to-digital conversion circuit 5oz, the microprocessor 402 detects that it has entered the measurement range, sends a stop command to the register 41Z to the pulse motor ZM, and stops the lifting of the wafer WF. Next, the microprocessor 40Z controls the air sensor nozzles AG1 to AG4. The surface position of the wafer WF is measured via the voltage conversion circuit 49Z and the analog-to-digital conversion circuit 5oz, and the movement amount of the uni-huffing mechanism Δdl = do - (dl + d2 + d3 + d4
)/4. The movement resolution by the pulse motor ZM is 2 μm, and the microprocessor 40z applies a movement amΔd1 in units of 2 μm to the register 41Z to raise the wafer Z-axis. As a result, the surface position of the wafer is located within an accuracy of about 2 μm with respect to the focal plane position. At this point, the distance to the wafer surface is also measured. Air sensor nozzle A
If the distances measured by GI to AG4 are respectively d9 to d12, the microprocessor 4oz will now read the register 4.
Δd2=do −(d9 +d10+dll+−
612) Issue a command for the moving direction and moving amount of the piezoelectric element PZ to be /4. The register 43Z stores this command and sends the command to the digital-to-analog converter 44, respectively.
Z and the piezoelectric element drive voltage generation circuit 46Z.

The digital-to-analog converter 44Z outputs the digital value of the register 43Z as an analog voltage to the differential amplifier 45Z as a command voltage. 46Z is a piezoelectric element drive voltage generation circuit, which generates a voltage up and down centered on a voltage that is approximately one-half of the maximum voltage VH applied to the piezoelectric element PZ, and generates a voltage across the differential amplifier 45.
Occurs in response to the output of 7'. When the wafer WF moves up and down by driving the piezoelectric element PZ, the amount of movement can be detected and measured by the eddy current type position sensor Is. The output of the eddy current type position sensor Is is converted into a voltage proportional to the amount of displacement by a displacement voltage conversion circuit 47Z, and the output is converted to a voltage proportional to the displacement amount by a differential amplifier 45. ,
Z and is output to the analog-to-digital converter 48Z.

The differential amplifier 45Z successively compares the amount of movement of the wafer 2 mechanism by the piezoelectric element PZ detected by the eddy current position sensor Is and the amount of movement instructed by the register 43Z,
Drive until the difference falls within the error range. As a result, the surface of the wafer WF can be accurately positioned with respect to the predetermined imaging plane position.

The analog-to-digital converter 48Z is an eddy current type position sensor I
The amount of movement of the piezoelectric element Pz detected by S is converted into a digital amount and transmitted to the microbe and processor 40z. The processor 402 detects this and knows that the imaging plane position has been reached, and moves on to the next control.

Alternatively, simultaneous control as shown below is also possible.

That is, the operation classification of the pulse motor ZM and the piezoelectric element PZ is based on the following equation. .

Measured value/pulse motor resolution - Quotient (pulse, motor drive) Remainder (pressure, electric element drive) In other words, when the remainder is calculated by dividing the average value measured by air sensors AGI to AG4 by the resolution of pulse motor ZM. , the remainder is driven by the piezoelectric element PZ. This allows coarse driving and fine driving to be performed suitably. Further, the quotient and the remainder are simultaneously stored in the registers 417 and 432, and the pulse motor ZM and piezoelectric element PZ are driven simultaneously, so that the focus position can be reached at high speed.

The area indicated by A in the figure is the first shot area that is exposed first. In this way, the step-and-repeat type projection printer sequentially prints the wafer WF placed on the Uni Hachisec WC by moving it in the X and Y axis directions. By the way, when printing area A, since the reduction projection lens PO and air sensor nozzles AG1 to AG4 are located as shown in the figure with respect to the wafer WF, the air sensor nozzles AGI and Ac2
- A and G3 can detect and measure the surface position of wafer WF, but air sensor nozzle AG4 cannot detect and measure the surface position of wafer WF. That is, as the wafer WF approaches the reduction projection lens PO, the microprocessor 402 can detect that the air sensor nozzles AG1, Ac2, and Ac1 are sufficiently within the measurement range, but has no response input. Therefore, the microprocessor 402 determines that air sensor nozzle AG4 cannot be measured, and air sensor nozzle AGI・AG2
・Measurement value d1 of AG3 ・Take out the values of dl・d3 and average them to find the average distance to the wafer WF (dl + d2
+d3)/3. Focus positioning of area A is performed based on this calculated value.

When exposure of area A is completed and area 8 is to be exposed next, the wafer surface position of exposure area B is detected in advance by air sensor nozzle AG1 and the distance is measured before exposure of area B. The microprocessor 402 provides the value to the register 43Z as a command light. Next, the wafer WF is moved in the X direction so that the exposure area B is exposed to the reduction lens PO.
The piezoelectric element Pz continues to be driven based on the previously measured value until the piezoelectric element Pz is positioned below the piezoelectric element Pz, the amount of movement is confirmed by the eddy current type position sensor Is, and the driving is ended when the movement of the predetermined amount is completed. In this way, while the wafer WF is moving from the exposure area A to the exposure area B, focusing in the next exposure area B can be completed. Therefore, it is possible to realize an exposure apparatus that utilizes the operation time of stage movement and has little wasted time.

FIG. 9 is a block diagram of the X, Y stage control circuit.

WX (WY> is the X, Y stage, D $, t DC motor, and the X, Y stage and the DC motor are coupled with a ball screw. The DC motor is driven by the motor driver MO. A tacho generator (speed signal generator) T is added, and the output of the tacho generator T is used as a driver M for speed control.
Feedback is given to D. The positions of the X and Y stages are determined by the current position counter P based on the output signal of the length measurement 21LZ.
Measured at CP. A laser interferometer is used as the length measuring device. The output of this length measuring device is a relative position output, and
, an origin sensor XH,S (YH,S) and an origin detection circuit SO are provided to detect the origin of the current position of the Y stage, and the outputs of these detect the origin of the X and Y stages, and the gate circuit AP Length measurement is started, and the current positions of the X and Y stages are measured by the current position counter POP. The microprocessor CP that controls the entire X and Y stage has a current position latch circuit PL.
The current position I of the X and Y stages can be known through P.

CLP is a target position latch for the X and Y stages, and the target position of the stage is set by the microprocessor CP. DIF is a difference device, and current position counter PC
It outputs the difference between P and the target position latch CLP. That is, it indicates the amount of movement from the current position to the target position, and is used as a feedback signal for the position servo and at the same time as means for detecting the timing for setting the drive pattern of the X and Y stages. As a position servo feedback signal, the output of the difference device is used as a D/A converter D.
It is input to a bit converter BG to match the bit number of AP, the bit-converted output is input to a D/A comparator 2-1, and its analog output is input to a position servo amplifier GA as a position feedback signal. There is also a comparator COMP to generate a timing signal to set the drive pattern of the X and Y stages.The output of the bit converter BC and the movement II regular latch RPL are compared.5 When they match, the drive pattern is set from the comparator COMP. timing signal is output. The timing signal is input to the address generator RAG of the random access memory RM in which drive pattern information is stored, and
The necessary RAM address is generated and driven at that timing:
Pattern information is output from random access memory RM. The timing signal is also input to the interrupt generator INT to generate an interrupt signal, the timing of which can be sensed by the microprocessor CP''.

In addition to the movement amount data mentioned above, the drive pattern information includes DC
There is command value information for Ha of the motor, and the command value information includes DC
There is current command value initial information that actually drives the motor, target command value information that becomes the target value, and various information that controls the process to reach the target command value. PCv is a current command value counter, which generates a command value for driving the DC motor. C.L.V.
is a latch circuit for target command value information. FG
is a function generator that sets the frequency division ratio of the frequency divider DIV.By dividing the pulse frequency of the oscillator DIV to a desired frequency, the value of the current command value counter PC■ is set to the target command value latch. Controls the process to reach the CLV value. COMV is a comparator whose role is to compare the value of the current command value counter PCv and the value of the target command value latch CLV and enable the gate AV until they match.At the same time, it informs the microprocessor of the matching timing of the interrupt generator INT. It is notified by an interrupt signal. The value of the current command 1st shift counter PCv is input to the D/A converter DAV, and its analog output is input to the driver MD through the ON side of the changeover switch SW during speed servo control and becomes the speed command value.

Further, during position servo control, the position command value is input to the adder circuit AD, and the output of the converter DAP, that is, the addition signal with the position feedback signal is input to the driver MO through the position servo amplifier GA and the OFF side of the changeover switch SW.

DMG is a drive mode generator, and this output signal turns the drive mode changeover switch SW 0N10FF. For example, the ON side enters speed servo control mode, the output of the D/A converter DAV is input to the driver MD, and the operation T
In the R candy section, the X and Y stages are driven by speed servo control. On the O'FF side, it becomes position servo control mode,
The output of the position servo amplifier GA is input to the driver MO, and

The Y stage is driven by position servo control during the operation end section.

FIG. 10 is a diagram showing changes in the speed of the stage when the horizontal axis is time and the vertical axis is speed. The section SS from time to to t4 in FIG. 10 is speed control, and from time t4 to t
The section PS up to 6 is a position control section.The speed control section is an acceleration section that accelerates at a constant acceleration.8, a constant speed V
It consists of a constant speed zone 1BC in which the vehicle moves at wax, a deceleration zone CD in which the vehicle decelerates at a constant deceleration, and a constant speed zone DE in which the vehicle moves at a constant speed Vain. Acceleration/deceleration straight line A8.8C, CD
The DOWN speed switching point is determined by the difference between the current position point A and the target position point P, that is, the moving distance.

This is determined, for example, by the microprocessor CP referring to a data table for the acceleration/deceleration, maximum speed, and DOWN speed switching point corresponding to the moving distance. The obtained data are stored in the random access memory RM shown in FIG. 9 by the microprocessor CP. FIG. 11 is a diagram showing the contents of the random access memory.

The contents of the random access memory RM are three blocks P.
The blocks are divided into HASE1, PHASE2, and PHASE3, and the contents of each block are composed of four pieces of data.

The data of PHASEL is the data for controlling from the start point A to the DOWN speed switching point C.
The data of SE2 is the data that controls from the DOWN speed switching point C to the position servo switching point E, and the data of PHA
The data SE3 is data for controlling from the position servo switching point E to the stop point P.

Next, use FIG. 9, FIG. 10, and FIG. 11 to make X.

A method of controlling the Y stage will be explained. First, the microprocessor CP writes data necessary for driving into the random access memory RM. Next, set the target position using the target position latch CL.
P and sends a start signal ST to the RAM address generator. As a result, the address of PHASEI is generated from the RAM address generator RAG, φ speed data is stored in the current command value counter PCv from the random access memory RM, MAX speed data is stored in the target command value latch CL and V, and acceleration gradient data is stored in the function generator FG. , and the movement resetting latch RPL are set with the amount of movement up to the DOWN speed switching point C, and the drive mode generator DMG sets the switch SW to the ON side. The X and Y stages move toward the target position by the output of the converter DAV.
0 Start the acceleration operation as shown in Figure A-B. That is, until the value of the current command value counter PCV becomes equal to the value of the target command value CLV, the counter P counts the output of the frequency divider DI■.
Counter PC whose input to the frequency divider DIV is cut off after the acceleration operation AB shown in Fig. 10 is performed by the variable output of C■ and a match is reached.
A constant speed operation BC is performed by a constant output of v.

Next, at the DOWN speed switching point C, a match signal is output from the comparator COMP and input to the RAM address generator RAG. As a result, the address of PI-IAsE2 is generated from the RAM address generator RAG,
Current command value counter P from random access memory RM
MAX speed data in Cv, target command value latch CLV
The MIN speed data is set in the function generator FG, the deceleration gradient data is set in the function generator FG, and the amount of movement up to the position servo switching point E is set in the movement amount setting latch PRL, and the X and Y stages begin deceleration operation. , that is, the current command value counter PC
Until the value of V becomes equal to the value of the target command value CLV, the deceleration operation CD is performed in the same manner as described above, and the matched aggressive speed operation DE
Do the following.

Next, at the position servo switching point, comparators Co to I
A match signal is output from P and the RAM address generator RΔG
is input. This causes the RAM address generator to
The address of HASE3 is generated, and the random access memory RM stores data corresponding to the movement m to the position servo switching point E, for example, 25μ before the target value, to the current command value counter PCV, and the target position data to the target command value latch CLV. , the position servo gradient data is set in the function generator FG, and the target stopping point P is set in the movement amount setting latch, and at the same time, the position l! is sent to the drive mode generator DMG. Itll
The mode is set, the switch SW is set to the OFF side, and the X and Y stages are driven for position control. Next, at the control end point F, comparators COMV and GOM
A match signal is output from P, and the interrupt generator IN
The signal is input to T and an interrupt signal is generated. The microprocessor CP detecting this assumes that the basic X, Y stage control has been completed, and determines the permissible value (hereinafter referred to as tolerance) of the stage stop position accuracy. The microprocessor CP inputs the current position data from the current position counter PCP via the current position latch PLP, determines whether the difference from the target position is within the tolerance, and determines whether the stop position accuracy and fluctuation are within the tolerance. At this point, the control is completed and the movement of the X and Y stages ends.

Figure 12 shows off-axis optical system O for TV alignment.
An embodiment of A is shown, and R11°111 in the figure is a light source for illumination, for example, a halogen lamp is used. R12,
L12 is a condenser lens, R13A.

R13B, L13A, and Li2S are the bright field diaphragm and dark field diaphragm that can be attached and removed interchangeably.In the figure, the bright field diaphragm R13A
, L13A is installed in the optical path, so the condenser lenses R12 and L12 connect the light sources R11 and 111 to the bright field aperture R.
(L) Image is formed on 13A. R14 and L14 are relay lenses for illumination, and R15a-b and L15a-b are cemented prisms. This cemented prism has the function of making the optical axis of the illumination system and the optical axis of the light receiving system coaxial, and has an inner reflective surface R15a.
, L15a and semi-transparent reflective surface R15b, l-15b
Equipped with Here, light source R, L11, condenser lens R
, 112, bright or dark field diaphragm R, L13A, B1 relay lens R, L14, cemented prism R, l-15a.

b. The objective lenses RL and LL constitute an illumination system, and the light flux emitted from the objective lenses RL and LL passes over the alignment marks CRL' (LR) 11, 12 or WPR (L) 1 of the wafer WF in FIG. Use epi-illumination.

R and L116 are relay lenses, R and L17 are mirrors that switch the optical path from high magnification to low magnification, and R and L18 are index glass plates having reference marks TPR and TPL for TV alignment.
The reference mark TPR(L) has the function of providing the origin of the coordinates, so to speak. Therefore, the alignment mark is detected as an X coordinate value and a Y coordinate value. R,
L19 is the imaging lens, R9L20 is the NA limited aperture,
The above cemented lens R, L15a, b1 relay lens R, L16,! 1R.

117, index glass plate R, L17, Ili image lens R,
L19 and the high-magnification image pickup tubes CR and CL form a light receiving system, and the optical path passing through the objective lenses Rt and LL is reflected by the inner reflective surfaces R and L15a of the cemented prism.
5b, and is reflected again by the inner reflective surfaces R,', L15a, and heads toward the relay lenses R, L16. Alignment mark image CRL (LR) 11 on wafer WF in FIG.
．． 12 is formed on the index glass plate R, 118 having the reference mark TPR(L), and then the reference mark l*TPR(
L) together with high magnification R@tube CR.

The image is formed on the imaging plane C. To explain in detail the operation of the high-magnification optical system with the above configuration, the luminous flux from the illumination light sources R and Lll is converged by the condenser lenses R and L12, and passes through the aperture of the bright field diaphragm R and L13A or the dark field diaphragm R9113B. It then passes through the illumination relay lens R9L14, passes through the semi-transparent surfaces R and L15b of the cemented prism, and is illuminated by the reflective surface R.
, L15a, and illuminates the wafer WF through the objective lenses RL and LL.

The light beam reflected on the surface of the wafer WF is passed through the objective lens R(L)
L receives an image forming action, enters the cemented prisms R2115a and R2115b, is reflected by the reflective surfaces R and L15a, is then reflected by the semi-transparent surfaces R and 1 isb, and the reflective surfaces R and L15ar, and is ejected from the relay lens R. , 116 to form an image on the index glass plate R, 118, and the imaging lens R, 11
9, images are formed on the image pickup tubes CR and CL. Next, switch to dark field mode to clearly detect the alignment mark image.
This is imaged to detect the position of the alignment mark image. Depending on the position of the alignment mark detected by electrical processing to be described later, wafer stage WS aligns the first shot (exposure) area of wafer WF with projection lens Po.
move so that it occupies the prescribed position in the projection field. R, L2
1 is a reflection mirror, R and L22 are erectors, R and L23 are apertures similar to R and L20-, and CDR and CDL are C for low magnification.
C[) performs the same action as above at a lower magnification. These optical systems are not necessarily a pair, but only one each is sufficient.

However, if they are paired, they can be detected simultaneously, so high speed and high accuracy can be expected.

Figure 14 is a block diagram of the entire device.
In addition to the HT shown in the figure, it includes a sub CPU and a drive circuit, that is, each unit control circuit as shown in FIGS. 8 and 9, for example. Also low magnification television (TV) camera CDO, CDR
, CDL connects the signal line L1 to the first TV receiver 117VI.
The high magnification TV cameras CR and CL are connected to the second TV.
It is connected to the receiver TV2 by a signal line L2.

The control box CB includes a main CPU and a control section MC including a high-speed arithmetic circuit, as well as a ROM.

Includes RAM. The ROM stores instructions as shown in the flowchart described later. KO3 uses the console to set various parameters and perform various other controls, and the printer PRT prints out various statuses of the device. FIG. 15 shows an example of a display monitor when performing off-axis and TTL alignment. A in the diagram is a high-power TV
2 is used to perform off-axis alignment, and the reference mark TPR (shi) on the index glass plate R(L) 18 of the off-axis optical system OA in FIG. 12 and the wafer WF in FIG. 13 are shown. High magnification alignment mark CRL
(LR) 11.12 will be displayed, allowing you to check the alignment of both marks. B shows how alignment is performed using the low magnification TV 1 by comparing the low magnification mark WPR(L)1 in FIG. 13 with an electronically set reference (cursor) line KSL. C is TTL alignment optical system As
Use the wafer WF female mark WKR on the low magnification TVI.
.

An example is shown in which the male marks WSR and WS1 of WKL and reticle RT are displayed, and the state in which both marks are aligned at TTL can be confirmed. In addition, when using a wafer for which automatic alignment is not possible, special manual alignment marks can be printed on the scribe area of the wafer WF, and alignment with the special manual alignment marks on the reticle can be performed manually. .

In this case, cross marks such as those shown in A and B are preferable to recommended marks because they are easier to visually align.

FIG. 16A shows an example of the configuration of the first reticle, B shows the second reticle, and C shows how the first reticle RT1 is sequentially exposed on the wafer WF surface.
D shows how the contents of the second reticle RT2 are overlapped and exposed one after another.

In the figure, cpl, CR2 has 8 reticles 1. RT2
Circuit pattern (actual element) provided above, 5CR1,5
CL1, 5CR2, 5CL2 are the scribe areas provided on the left and right sides of the actual element, and the first reticle R'TI has 2 scribe areas.
Eye mark WKRI used for alignment with the second reticle RT2.

WKLl is provided. Further, if necessary, the aforementioned special manual alignment mark MAR1°MALI is provided in place of the mark WKR(L)1 or in parallel as shown. The lower scribe area SCU includes a high-magnification alignment mark CRR that is smaller than the low-magnification alignment mark WPR(L).

Two CRLs are prepared. It is preferable to provide a pair of two on the left and right as shown in FIG. 13, as this increases the probability that they will fall within the field of view of the objective lenses RL and LL of the off-axis optical system OA shown in FIG. WPR.

WPL is low magnification alignment mark 1. R2H, R8L
is a male mark for alignment of the reticle, and the position of the reticle is set in alignment with the reference marks RKR and RKL on the lens Po shown in FIG. RKR, RKL are WKRI
, has a female mark shape like WKLI, and T
Reticle auto-alignment is performed by aligning the recommended marks in the same way as in TL auto-alignment.

On the second reticle RT2, female marks WKR2, WKL2 for next process alignment and male marks WSRI, WSL1 for main process alignment are provided. R.C.N.I.
, RCN2 each indicate a reticle number and are provided in a coded manner, and by reading these with the T T'L alignment optical system AS shown in FIG. 5, the reticle number can be automatically identified.

This is done at the same time when creating the circuit pattern and each memory.
Similarly, WON in FIG. 13 indicates the coded wafer number, and TTL alignment optical system A
s and read by the TL alignment optics As or the off-axis optics OA. Of course, these operations can also be performed by determining the reticle and wafer numbers based on information specified in advance from the console KO3 in FIG. 14 or information sequentially specified in real time. In FIG. 16, when the first reticle RTI is inserted as shown in FIG. 5, the blade BL in FIG. Area 5 CRI
, 5CLI are exposed. In this state, as shown in Fig. 16C, 1, 2, 3.・
...and is exposed to light.

That is, in the first shot (exposure) area 1, the special manual marks MARI and MALL of the reticle RT1 are MAR.
ll, MALII, mamasu mark WKR1, W
KL1 is exposed (printed) as WKRll and WKLll, and the circuit pattern CP1 is exposed (printed) as CPll. Note that in reality, the printed pattern on the reticle RTI is projected through the projection lens PO, so an image that is horizontally and vertically inverted is printed on the wafer WF, but for ease of understanding, the illustration assumes that the same image is printed. . The subsequent images are sequentially printed in the same manner. At that time, for example, circuit patterns CP11 and C
The scribe area 5CR12L11 between P12 is shared by the circuit patterns cpii and 12 to save wafer space. Therefore, a pair of left and right marks, for example WKRll and WKLII, are shifted vertically from each other. With this configuration, for example, marks WKR12 and WKil do not overlap, which is preferable. In this way, exposure and steps are repeated sequentially in the order of shot numbers 1 to 45 shown in FIG. 13, and when a specific shot, for example, the 20.26th area in FIG. The opening of the blade 8L is set so that the area SCU is exposed. As a result, the high magnification alignment mark CRL of reticle RT1
.

As shown in FIG. 13, CRRs are printed in the 20th and 26th lower side scribe areas of shots as CLR11, CRLll and CLR12, CRL12, respectively.

Also, specific shot 41. and 45, Figure 17 C1
Blade B so that the low magnification alignment marks WPR and WPL of the reticle RTI are exposed as shown in D.
After setting the aperture of L, the shot area 41. in FIG. 13 is set.
Low magnification alignment marks W on the right and left sides of each of 45
Printed as PR1, WPLI. In this manner, the printing on the first wafer WF is completed. This first
The shot order shown as 1 to 45 in FIG. 3 is preferable because the amount of movement of the wafer stage WS is the shortest. The wafer that has been baked is replaced with the next wafer WF, and the same process is performed. When 10 wafers have been completed, the first reticle RTI is ejected and the second reticle RT2 is inserted. As described above, the second reticle RT2 is configured as shown in FIG. 16B, and during the first shot, the male mark WSR of the reticle RT2 and the female mark WKRI of the wafer WF are
I, WSLl and WKLll are precisely aligned by the TTL alignment optical system As and then exposed, and the circuit pattern CP2 of the reticle RT2 is printed as CF3I on the circuit pattern CP11. Furthermore, the male mark WSR1 of the reticle RT2 is burned into the middle of the female marks WKR11 and WKL11 of the wafer WF, and is no longer usable. Therefore, reticle RT2 has female marks WKR2 and WKL for the next process alignment.
2 is provided at a position shifted one step upward as shown in FIG. 16B, and these marks are printed as new female marks WKR21, WKL21, etc. on FIG. 16O. In this way, new marks are sequentially provided and old marks are not used, which is preferable because it allows error-free reading without reducing reading sensitivity.

Also, in the illustrated example, the scribe area is saturated with two reticles for ease of understanding, but the number of reticles is usually 10.
A few sheets are sufficient, and a normal scribe area has a sufficient size considering the size of the chip area, the size of the mark, the sensitivity of the reading means, etc. Alternatively, the old mark may be used repeatedly. During special manual alignment, mark MAR2, M on reticle RT2
AL2 and marks MAR11 and MA burned in the previous process
Lll, etc. are performed by the low magnification TV monitor TV2.

Figure 18 shows the TV screen in Figures 14 and 15 in the X direction.
This shows how the pixel pJ is divided into M parts in the Y direction.
i indicates the pixel in the J-th row and the i-th row. The number of divisions M in the Y direction usually matches the number of horizontal scanning lines, and therefore, in order to divide into pixels, it is sufficient to perform sampling N times within the -horizontal synchronization signal section.

Therefore, addition in the X direction is Sx + -DATA (Pa) + DATA (Pa
)+......+DATA(P+N
), Sx 2 -DATA (P21)+
DATA (P22)+・・・・・・+DATA
(PzN), SxM=DATA
(PM + )+DATA (PM z )+...
... +DATA (PMN), addition in the Y direction is SY + -DATA (Pa) + DATA (P2
1) +・・・・・・+DATA (PM + ), S
Y2 = DATA (P error) + DATA (P22)
+...+DATA (PM2), SYM -
DATA (P+N)+DATA (P2N)
+...+DATA (PMN).

When the addition is completed, the X and Y direction integration memories contain Sx+, SX2 r +++++++sxM, and SYI, respectively.
.

S v z +..., SvM data is stored.

An example of the alignment mark is the cross pattern mark shown in FIG.
The concentration distribution becomes as shown in C).

(B) shows the addition result in the X direction, and (C) shows the addition result in the Y direction. The characteristics of the concentration distribution in FIGS. 19(B) and (C) are as follows:
As can be seen from the figure, the added density of the mark is in two stages. For these two-stage concentration distribution, the first
As shown in Figure 9 (C), two slice levels, e.g.
When 5L1 and X5L2 are provided, the binarization patterns become the patterns shown in FIGS. 19(D) and (E), respectively. Therefore, when the centers of these binarized patterns coincide, these become the center coordinates of the alignment mark.

The block diagram in Fig. 20 shows an example of an alignment mark detection circuit, where the block X surrounded by a broken line is a block that adds the density of pixels in the X direction, and the block Y is a block that adds the density of pixels in the Y direction. be.

In Fig. 20, 31V is a video amplifier, 32V is an analog-to-digital converter, and 33V is a latch circuit.The video signal sent from the TV camera control section is amplified by the video amplifier 31V and digitized by the analog-to-digital converter 32V. After that, it is stored in the latch 33V. The output data of the latch 33V is output to the addition block X in the X direction and the addition block Y in the Y direction. Block Y
34V is an adder that adds data in the Y direction, 3
5V is an addition output latch that latches the output data of the adder 34, 36V is a Y-direction integration memory that stores the data of the addition output latch 35V, and 37V is an addition input latch that latches the output data of the memory 36V.

In block X, 38V is an adder that adds data in the X direction, 39V is a latch that latches the output of the adder 38V, and 40V is an
This is a direction accumulation memory.

There is no particular limitation on the number of bits of digital data in these circuits, but for example, the analog/digital converter 32V is 8 bits, the adder 34V, 38V and the memory 38V,
40V has a 16-bit configuration.

41V is a sequence and memory control circuit that controls read/write and chip selection of the memory 36V, and 42V is a memory control circuit that controls the memory 40V in block X.

43V is the sequence and memory control circuit 41V
This is a control register for the microprocessor MPtJ to control, and the input of the register is connected to the 44V data bus of the microprocessor. Furthermore, the microprocessor MPIJ can access memories 36V and 40V via this data bus 44V. Are 45V, 46V, 47V, and 48V buffers for that purpose? The buffers 45V and 47V operate when the microprocessor MPIJ writes data to the memory 36 and 40, and the buffers 46V and 48V operate when reading data. 49V is a clock circuit, 50V
, 51VG, and a memory write address circuit and a memory read address circuit that generate a write address and a read address of tXX direction integration memory/36V. 52V
is the address selector that switches between the memory read address and the write address, and 53V is the microprocessor MPt.
Address buffer when J accesses memory 36V?
The output of the address selector 52V is selected except when the microprocessor MPLJ accesses, and the output of the buffer 53V is prohibited. 54V is X
A memory address circuit 55V generates an address of 40V for the direction integration unit, and an address selector 55V switches between the address of the memory address circuit 54V and the address generated when the microprocessor MPLJ accesses the memory 40V. This is a television synchronization signal generation circuit that generates a horizontal synchronization signal, a vertical synchronization signal, a blanking signal, etc. for the television based on the clock of 56V and M49V.

57V and 511V are respectively connected to the data bus 44 of the microprocessor MPU.
The position display register 59V is a cross mark display circuit, and the microprocessor outputs the position of the alignment mark detected for TV alignment to the X position display register 57V and the Y position display register 58V. It is sent as a cross mark signal to the video input terminal of the television camera control section. In addition, the ninth
The data is sent to the CPU shown in the figure, and the wafer stage is moved to the mark identification position by a servo motor.

The functions of the above-mentioned television alignment detection circuit are: (1) integration of data in the X direction, (2) integration of data in the Y direction, and (2) display of an alignment mark on the television screen.

Of these, the data integration in the X direction and the data in the Y direction are performed by the adder 34.3 of the television alignment detection circuit.
8 performs addition and stores the added data in memory. The data addition is performed in units of one frame of the television signal, and if necessary, the addition may be completed with one frame, or the addition of a plurality of frames may be performed. In either case, during the addition, the memory 36V, 40V data bus' and address bus are electrically isolated from the microprocessor MPLI's data bus 44V and address bus, and the memory 36V address is transferred to the address selector 52v1 memory. The address t of 40V is connected to the address of the t address circuit 54V, and addition is performed under the control of the read/write signal and chip select signal generated from the sequence and memory control circuit 41V and the memory control circuit 42V. .

When the addition of a predetermined number of frames is completed, an addition end signal is generated from the sequence and memory control circuit 41V on the interwoven signal line INT.

After generation of this addition end signal, the microprocessor MPL
I accesses the memory 36V and the memory 40V and detects the position of the television alignment mark from the added data. When the microprocessor accesses the 36V or 40V memory, the memory address, read/write signal, chip select signal, etc. are naturally controlled by the microcomputer's control signals. Also memory 36V
Is the data buff? 46■, Is the data in memory 40V a buffer? 48V to the data bus 44V and read by the microprocessor.

This will be explained in more detail using the flowchart shown in FIG. When an addition/addition start command is issued by the microprocessor in step SV1, addition in the X and Y directions is started as described above. The microprocessor waits at step SV2 for completion of addition, and when the addition of a predetermined number of frames is completed, the process proceeds to step SV3. Step S
At V3, the microprocessor searches for the maximum and minimum values of the image density data stored in memory. When the maximum value and minimum value are found, the slice level X5L1 and WSR1 are then set in step SV4.

The slice level WSL1 is, for example, a value of 10% of the difference between the maximum value and the minimum value (referred to as the peak value) of the image density data.

Next, in step S5, the slice level X5L1 is compared with the contents of the memory, and XL1 and XRI are determined from the coordinates (memory address) where the comparison result is inverted. Similarly, in step SV6, a slice level X5L2 of 20% of the peak value is set, and in step SV7, XL2 and XR2 are determined in the same manner as step SV5.

As described above, the kneading pattern shown in FIGS. 16(D) and (E), that is, the coordinates XL1 and XR1.

XL2 and XR2 can be determined. T to step Sv8
Calculate (XR2-XL2)/2.XRI-X[1
)/2, and if they are almost equal, it is determined that the detected coordinates are alignment marks and the process proceeds to step SV9, and if the comparison values are significantly different, it is determined that they are not alignment marks. and step 5vi
Proceed to o. If the process proceeds to step 5V1G, for example, the slice level setting value may be changed and measurement may be performed again, or the process may proceed to a process of searching for an alignment pattern assuming that there is no alignment pattern within the screen. Similarly, the Y coordinates YLI, YRl, YL2, and YR2 can also be determined.

The advantage of the embodiment shown in FIG. 16 is (2) Random noise is averaged by the addition, and the S/N ratio is improved. ■Position detection in the X direction and Y direction can be performed independently, making detection easier. For example, the memory capacity for storing 0 image data becomes smaller.

Hereinafter, the operation of the present invention will be explained according to the flowcharts shown in FIGS. 22-25.

First, in step SS1 of FIG. 22, initial settings of all devices are performed. - To give an example, the memory RAM is zero cleared, the entire TTL alignment optical system As is moved in the Y direction, the objective lens IIR (L) and the objective mirror 12R (L) are moved in the X direction, and the lens P
Reticle reference mark RKR (L) installed on O
and the objective mirror 12R.
(L) is set at 45 degrees so that the laser beam can irradiate the mark position, the reticle stage, wafer stage, and blade are set to their initial states, and various other initial settings are performed. In step SS2, the reticle RT
is adsorbed to the reticle chuck RC by vacuum suction, and in step SS3, the laser shotter BS is opened to prepare for alignment of the reticle RT. Next, in the scoop SS4, the entire optical system AS is moved in the Y direction by a pulse motor (not shown), and the objective lens 11R (L) is moved.
The objective mirror 12R(L) is moved in the X direction by a pulse motor (not shown), and the presence of the reticle set mark R8R(L) on the reticle RT is detected by the detector 18R(L). Mark R8R detected in step S85
(L) and the distance from the predetermined reference point are detected! 118R(L
), and each pulse motor PX of the reticle stage R8 is moved by the distance measured in the next step 5S61.
, PY, and Pθ to set the set mark R on the reticle RT.
8R'(L) is moved near the reference mark RKR(L). At the same time, objective lens 11R (L) and objective mirror 1
2R(L) is returned to the position facing the reticle reference mark RKR(L) in preparation for fine alignment.

In step 887, the reticle reference mark R on the lens ρθ is
KR(L) and reticle reference mark R8R of reticle RT
The amount of deviation in the left and right X and Y directions from (L) is detected 118R (
L). It is determined in step SS8 whether the average value of each measured value is within the tolerance value, and if it is within the tolerance value, the process proceeds to the next step 5S10, and if it has not yet reached the tolerance value, it is determined again in step SS9. Drive each pulse motor PX, PY, Pθ of reticle stage R8,
Step SS7.8 is repeated to move reticle stage R8 until the value is within the tolerance. If the CPU determines that the allowable value has been reached, the process proceeds to step 5810. In step 5S10, the exposure area of the reticle RT is set, and first, as shown in FIG. 14A, the opening area of the blade 8L is set so that the central circuit pattern part CP and the left and right scribe areas SCR (L) are exposed. .

Next, in step 811, the first wafer WF is attracted to the wafer chuck WC of the wafer stage WS. The wafer WF transported here is a wafer that has never been exposed to light, and therefore the alignment mark 7 has not been printed yet.

In the next step 5S12, in order to identify the reticle number,
The objective mirror 12R or 12L is moved to the detection position of the reticle number RCN on the reticle number. There are usually several to 14 reticles RT used to manufacture one large-scale integrated circuit.
Since five sheets are prepared, if the reticle number RCN is encoded and provided at the time of creating each circuit pattern, the reticle number (type) can be automatically identified. Step S S 13
The reticle number RCN coded in is read by the detector 18L or 18R. 19R or 19L may be used as the illumination light source at this time. This is the first (first
Since this is the reticle, the process advances to step SS14. In step 3514, the laser shutter BS is closed;
Also, during exposure, the wafer stage W is changed from the 45° attitude to the vertical (two directions) so that the lower side of the objective mirror 12R(L) does not interfere.
The first shot (exposure) area of the wafer WF is set directly below the projection lens PO by moving the wafer WF by a predetermined amount in the X and Y directions using the SL thermometers 9XM and YM. This movement is carried out extremely accurately by a laser interferometer LZ. The wafer WF, whose first shot area is set directly below the lens PO, drives the pulse motor ZM to increase the speed of the θ2 stage so that the wafer WF reaches a focus detection level of the air sensors AGI to AG4 attached to the lens PO. It is moved upward (step SS16). After reaching the focus detectable level, each focus value is detected by the air sensors AG1 to AG4, each detected value is stored in the RAM, and an average value is calculated (step SS17). This average value is taken as the focus value of the first shot area, and according to this value, the θ2 stage is moved upward or downward by the pulse motor ZM and/or the piezoelectric element Pz as described above until it reaches the target focus value (step S S 1l
l). Next, the shutter ST of the exposure light source LP is opened and opened for a predetermined time to expose the first shot area of the wafer WF, and the TTL alignment female of the circuit pattern part CP of the reticle RT and the left and right scribe areas SCR (L) is opened and closed for a predetermined time. Marks WKRn and WKLn are burned (step S
S19). and/or manual alignment mark MAR1 if necessary.

MALI can also be burned. Then step 8820°2
2 and 24 are made, and if the second shot area is neither of the above, step 58251 is performed.
, proceed to 252. In steps 251 and 252, as described above, the focus is detected by the sensor corresponding to the next shot area, and the θZ steady is moved upward or downward by the motor ZM and/or the piezoelectric element PZ until the focus is reached. At the same time, wafer stage WS is moved to X by servo motors XM and YM.

The camera moves in the Y direction, and the next shot area moves directly below the lens PO. This movement is also performed extremely accurately by the laser interferometer LZ, and similarly precise step feeding, focus detection, and exposure are sequentially performed. Step 5S2
When a specific shot area predetermined with
), CRR(L) are exposed in step 882.
In step 1.23, the opening area of each blade BL is set.

Further, when transitioning from the specific shot area to the normal shot area, the opening area of the blade 8m is returned to the normal shot area (step S S 10 in FIG. 17 A1).

When the final shot area has been exposed, the process moves from step 5S24 to step 882B. In step 8326, the wafer stage WS is moved to a predetermined position in order to write the wafer number on the wafer WF, the laser shutter BS is opened for a sufficient time for writing, and the coded wafer number and/or lot number WCN is written at the edge of the wafer WF. section (see Figure 14). In step 5S27, the wafer stage WS is moved to the wafer ejection (receiving) position to eject the wafer, and at the same time, the θ2 stage is moved to the initial lowest position by the pulse motor ZM. Next, it is determined in step 5S28 whether the wafer WF being transported is the final wafer. This is done by comparing whether the number of wafers has reached the number instructed by the operator to the microprocessor from the console in advance, and if it is not the final wafer, step 5S1
Return to step 1 and continue the same process as above.

As described above, the circuit pattern and the alignment mark of the first ticle are printed for the predetermined number of wafers and lots, and then the process is completed. The first group of wafers on which the circuit pattern and alignment mark of this reticle have been printed are then sequentially exposed from the second reticle to the n-th (final) reticle by precisely overlapping the same shot area on the same wafer. That is, when the second reticle is brought in, step S
The same operations as described above are performed from S1 to SS13,
In step 5S13, it is detected that the reticle number is "2", so the process advances to step 8829. In step 5S29, the wafer WF is set directly below the air sensors AGI to AG4 attached to the lens PO, and as described above, in steps 5S3G and 31, the speed is increased to θ゛22 stage speed, focus detection and average value calculation are performed, and step S Proceed to S321-323. Step 33321
Now, the mirror R of the off-axis alignment optical system OA (
L) 18 was set to low magnification system, and R(L) 1 was compared with dark field.
Select 3B.

At the same time step 33322, a low-magnification alignment mark WPR of the mechanically pre-aligned wafer WF is
(L)1 is set almost directly below the objective lens R(L)L. At the same time, in step S323, the θ2 stage is moved upward or downward from the focus average value detected in step 5831 until the target focus value is reached. Mark WPR(L)1 on wafer WF with objective ~
The operation of moving the lens R(L)L downward is performed using a predetermined constant. In step 5S33, the reference line KSL (cursor on the TV screen) and the alignment set mark WPR of the wafer WF are
(L) The amount of X and Y deviation from 1 is measured, and the amount of deviation is R
Stored in AM. Next, in step 5S34, one of a plurality of alignment modes A to C is selected, and accurate and high-speed alignment, stepping, and exposure are performed according to each alignment mode. Each alignment mode will be explained below.

In mode A, first, in step 5A11, the reference line KSL is
position and wafer WF alignment set mark W
Wafer stage WS is moved in the X and Y directions by servo motors XM and YM according to the value obtained by adding a constant to the amount of X and Y deviation from PR(L)1, and high-magnification alignment mark CR(L)11. of wafer WF is moved. 12 is the objective lens R (L)
Set it almost directly below L. At the same time, the θZ steady is rotated by the pulse motor θM according to the amount of deviation in the θ (rotation) direction calculated from the above X and Y deviation amounts (Step 5A1
2). In step SA2, the mirror R (L) 18 of the off-axis optical system OA is set to a high magnification system, and the reference mark TPR (L) and the high magnification alignment mark CR are set by this high magnification system.
(L) The amount of X and Y deviation from 11.12 is measured (step SA3). Further, in this step SA3, the amount of expansion and contraction of the wafer is also measured, and if it is within the allowable value, the value is distributed and added to each amount of deviation of X. X, Y in step SA4
It is determined whether the deviation is within the allowable value, and if it is determined that it has not yet reached the allowable value, the reference mark T'
Alignment mark CR (L) 11, 1 on PR (L)
The wafer stage WS is moved to X, Y, and θ by servo motors XM, YM and pulse motor θM so that
direction (step 5A5), return to step SA3, repeat the same procedure, and when the value falls within the tolerance, proceed to step 5A61.62. In step 5A61, the wafer stage WS is moved in the X and Y directions by the servo motors XM and YM according to the value obtained by adding a constant to the current position of the wafer WF, so that the first shot (exposure) area of the wafer WF is positioned directly below the projection lens PO. Set. At the same time step 5
At A62, the θZ steady of the wafer stage WS is moved upward by a predetermined amount by the pulse motor ZM. This is because the focal length of the projection lens PO is shorter than the focal length of the objective lens R(L)L. Below, as in the previous example, focus detection, average value calculation (step 5A7), θ
Perform Z steady movement (SA81), attitude change of mirror 12R (L) (SA82), exposure (SA 9), sequential focus detection, step movement (SA111.SA 11)
2) is repeated, and if it is determined that the exposure of the final shot has been completed (SAlG), the process proceeds to step 3A12. In step 5A12, wafer stage WS is moved by servo motors XM and YM to a position where wafer number WON, which was entered in the first step, can be detected.

For example, when detecting with TTL engineering system As, wafer number W
CN is set directly below the lens PO, and the encoded wafer number WCN is read by the detector 18R or 18L.

In addition to the laser source Is, a light source 19 is used as an illumination light source at this time.
R or 19L can also be used. Alternatively, it can also be read using an off-axis engineering system OA. In this case, the wafer stage WS may be moved so that the wafer number WCN$rJ is located directly below either the objective lens RL or LL. The read wafer number is stored in RAM. In step S A 14, the pulse motor ZM moves the θZ steady to the lowest position, and at the same time moves the wafer stage WS to the wafer ejection (receiving) position.
The process ends after the wafer is discharged.

If it is determined in step 5A15 that the final wafer is not yet completed, the process returns to step 5811 in FIG. 22, and the same procedure is followed thereafter.

In mode B in FIG. 24, first in step S81, for example, the shot area 13 in FIG.
Set by. At the same time, in steps 5B12 and 13, movement in the θ direction and Z direction is performed. Next, the laser shutter BS is opened (step 5B2), and X, Y by TTL is
Measure the amount of deviation. Here, as shown in shot area 13 in FIG. 13, the left and right deviations in the X direction are 1XL1, XRI
, the same Y direction YLI, YRI, each average value is 8
1 X-(Xi +XR1)/2°81 Y-(YLI +YR1)/2 is calculated and stored in the RAM (step 583). Next, a second designated shot (for example, FIG. 13, 19) is set directly below the projection lens PO, and a similar average value S2
, 82 Y is determined (step SB4.5). Next, in step 5B61, the X, Y average deviation flsIX, S2 X, 81 Y,
S2 From Y, determine the amount of deviation in the X, Y, and θ directions of the entire wafer (global) using the following formula.

GX-(81X+S2
+YR1)/2)/ where K is the distance between the designated marks of the first, second, and second shots and is a constant. .

At the same time, in step 5B62, the expansion/contraction IPE=SI X-32X of the entire wafer due to thermal expansion or the like is determined.

In step 88, it is determined whether or not each of the obtained values is within the allowable value.
Determine by 7. If it is determined that the value is outside the allowable value, the process proceeds to step 3B81.82. In step 8881, a predetermined amount (medium K) is added to the previously calculated X and Y deviation amount, and to that value, a value obtained by equally distributing 11 PE of wafer expansion and contraction for each shot is added, and further calculated in step 8882. If the servo motor movement amount is included in the result, this value is also added, and according to the total value, the first specified shot area is again set directly below the projection lens PO by the servo motors XM and YM. In step 8882, the quotient obtained by dividing the amount of deviation in the rotational direction by the resolution of the pulse motor θM is set as the moving amount of the pulse motor θM, and if a remainder occurs, the remainder is set as the moving amount of the servo motors XM and YM. By controlling the movement of the servo motors XM and YM in the X and Y directions, θ correction can be performed as a result. Normally, the X-axis and Y-axis of wafer stage WS are orthogonal in principle, and there is no θ component. However, in actual mechanical design, this complete orthogonality cannot be expected and a θ component always occurs. Therefore, by measuring the θ component when the device is assembled and controlling the amount of movement of the X and Y motors when operating the device, it is possible to move the wafer stage in a direction that eliminates the θ component. . This principle is called so-called orthogonality correction. This principle is utilized in step 5B81.82, and it is extremely preferable that the minute angle that cannot be corrected by the pulse motor θM can be corrected as a result by adjusting the movement amount of the servo motors XM and YM. In step 889, it is determined whether or not the pair of shot measurements described above have been repeated a predetermined number of times, and if not completed, the process returns to step S83 and the above operations are repeated. By repeating this process, the wafer position gradually approaches the allowable value, and if it is determined in step SB7 that it is within the allowable value, the process proceeds to the next step SB12. When the specified number of times is completed, the laser shutter BS is closed (step SB10) and the TTL measurement is completed. In step 5811, it is detected that the YES signal has not been sent within the specified number of times in step SB7, and preparations are made to advance the alignment mode to another mode, for example C. In step 5B12, even if the global alignment of the entire wafer is completed in the previous step, if there is a rotational direction deviation for each shot, an exposure deviation will occur, so this mode is used to measure this.

Therefore, in step 5B12, first, YLl, YRl, YL2, Y-
θ direction deviation amount from R2 for each shot tan SO2-
(YLl-YRl)/to 1.

2 is calculated as tan SO2=(YL2-YR2)/. It is determined whether the calculated SO2 and SO2 are within the allowable range (
Step S B 13), and if the values are outside the allowable values, each value is an approximate value, that is, whether the rotational deviation is in the same direction and the same inclination. If the inclinations are uneven, it is determined that precise overlay exposure is difficult in this mode, and the mode is switched to mode C (step S B 19). Since exposure is possible when they have similar slopes, the average Sθ=(Sθ1+3θ2
)/2 (step 5816), the reticle stage R8 is driven by the pulse motor Pθ until it reaches the average Sθ, and the reticle RT is rotated (step S
B17). Next, the θ direction deviation peak S for each shot is again calculated.
Measure θ1' and Sθ2' (step 5B18) t,, average □ Determine whether Sθ' = (Sθ1' + Sθ2')/2 is within the allowable value □ (step 8819) L, if not, go to step 8817 Return and repeat the same action. If the value is within the allowable value in step S B 19, then Fig. 23A
Steps 5A61 and 62 of 2 are accessed, exposure and steps are executed in the same manner as in mode A, and wafer processing in mode B is achieved.

Next, alignment mode C will be explained.

First, in step 5011, the servo motor moves the first shot area of the wafer WF directly below the reduction projection lens PO according to the value obtained by adding a constant to the X and Y deviation amount between the position of the reference line KSL and the pre-alignment set mark WPR(L)1. XM,
Set by YM.

At the same time, the θ2 stage is rotated by the pulse motor θM according to the value calculated from the above X and Y deviation amounts (step 5C12), and the θ2 stage is moved by the pulse motor ZM.
(step 8013). Next, the air sensor AG1 detects the aorcus in the first shot area.
～AG4 detects and calculates the average value, and pulse motor ZM and/or piezoelectric element P until the target focus value is reached.
z to move the θ2 stage upward or downward (step 5C2). Next, the objective lens 11R(L) and the objective mirror 12R(L) are moved to a position facing the reticle male mark WSR(L)n-1 of the reticle RTn (SC3
) to perform single-width TTL alignment. Next, the laser shutter BS is opened (SSC4), and the mark WSR(L)n-1 on the reticle is irradiated with the laser light from the laser source 1S using the objective mirror 12R(L). Step SC5
As is well known, the laser scan is started and the first X is formed between the male mark WSR(L)n-1 of the reticle RTn and the female mark WKR(L)n-1,m of the wafer WF.

Measure the amount of Y deviation. Step SC determines whether the first deviation amount is within a first tolerance value, for example, 0.1μ.
Judge with 6. Here, the amount of deviation in the X and Y directions of each shot of the wafer with respect to the reticle is XL, YL, and XR on the left and right, respectively.
, YR, the average amount of deviation is given by SX-(XL+XR)/2°SY(YL+YR)/2-. The amount of deviation tansθ in the direction of rotation
As in the previous example, G; i tanSθ-(YL-YR)/L. Here, L is a constant distance between the left and right marks WK (S) R-WK (S) 1 of each shot. In step SC6, the average value SX of each deviation bottle,
If both SY are within the allowable values, alignment is completed and the laser shutter BS is closed (step 5C12), and the process proceeds to the next step. If even one of the average deviation amounts SX and SY is outside the allowable value, the process proceeds to step SC7 to perform alignment.

Note that various values such as 0.3μ, 0.5μ, etc. can be specified as this tolerance value from the console. In step S01, Sθ is calculated from the above YL and YR. This calculated S
When the θ2 stage is rotated in the direction of eliminating the deviation according to θ, the amount of deviation occurs again in the x and Y directions. This is because the center of the wafer and the center of each shot are different.

This second X.

Since the Y deviation peak can be calculated in advance, it is calculated in step SC8. In step SC9, the calculated X and Y deviation amount is set to a second tolerance value, for example, 3μ.
It is determined whether or not it is within the range, and if it is within the range, proceed to steps 80101 and 102; otherwise, proceed to step 5.
Q Go to 111-113. step s cloi,
102 is within the allowable value, so the first and second X,
According to the sum of Y deviation amounts, the male mark WSR(L)n-1 of the reticle RTn is the female mark WKR(L)n-1 of the sea urchin AWF.

The reticle stage R8 is moved in the X and Y directions by the pallet motors PX and PY so that it is in the middle of the reticle stage R8. At the same time, the pulse motor θM is driven by Δθ to rotate the θZ steady. In step 5C111, the male mark WSR(L)n-1 of the reticle RTn is changed to the female mark WK of the sea urchin AWF according to the first X, Y deviation because it is outside the allowable value.
Pulse motor P so that it is in the middle of R(L)n-1, m
X.

PY moves reticle stage R8 in the X and Y directions. At the same time, in step S C112, the second X.

Mark WSR' (L) to be placed in the same manner as described above according to the amount of Y deviation.
n-1 is a female mark WKR (L) n-1.

Wafer stage WS is moved in the X and Y directions by servo motors XM and YM so that it is in the middle of m. At the same time, in step 5C113, as in step 5C102, the θ2 stage is rotated by Δθ by the pulse motor θM. In this way, by selecting and aligning the reticle and wafer according to the tolerance values, high-speed alignment and high overlay accuracy can be achieved at the same time. In other words, driving by a pulse motor has high precision but requires a long driving time, while driving by a servo motor has high speed but is insufficient in terms of accuracy, and the reticle side inherently has a longer travel distance than the wafer side. Considering the short distance, etc., the reticle stage is precisely driven by a pulse motor because the movement distance for correction is short when it is within the allowable value, and the movement for correction when it is outside the allowable value. Since the distance is long, it is preferable to drive the wafer stage at high speed with a servo motor. Furthermore, since the reticle stage side is also driven for correction at this time, the accuracy is sufficiently outstanding. The above operations are repeated until the result falls within the first tolerance value in step 806. When the high-speed, high-precision alignment is completed in this manner, the process proceeds to step 5012 as described above, and then to step 5C13.

In step 5C13, the objective lens 11R(L) and objective mirror 12R(L) are moved backward to predetermined positions so as not to interfere with exposure, and the objective mirror 12R(L) is vertically changed in attitude.

Next, the shutter ST is opened and closed for a predetermined period of time to perform exposure (step 5C14). When the exposure is completed, the final shot area is determined in step 5cis, and if it is not the final shot area, the process proceeds to steps SC161 to SC164. In step 161, the wafer stage is moved to the next shot area, and at the same time the θ2 stage is moved up (down) until the focus detection value is reached (step 162), and the objective mirror 12R (L) is moved to the next shot area. Mark WSR (L
) n-1 to a position opposite to mirror 12.
Change the attitude of R(L) to 45° (step 5C163
), and return the reticle stage R8 to the X and Y positions stored at the time of the first shot. When these operations are completed, the process returns to step SC3, and precise overlay exposure is performed by so-called die-by-die alignment until the final shot is completed. Once the final shot is completed, step 5
A determination is made at C15, and the process returns to step S A12 to repeat the same operation to complete the processing for 10 bits.

This device also has a manual alignment mode, and the above-mentioned special manual alignment mode and other manual alignment modes can be executed from any step by interrupt processing. May be used where possible. When performing alignment using the TTL optical system As shown in FIG. 5, the light sources 19R and 19m are turned on or a diffuser plate DF is inserted in the optical path of the laser 1S.

In addition, when using the or-axis optical system OA, the light source R11
, Lll are turned on, and the selection of dark field and bright field is roughly made by selecting f' (13A, R13B, L13A, and 113B).

Further, each alignment mark is selected as shown in FIG. 15.

[Effects] As described above, the present invention can greatly contribute to extremely high overlay accuracy, high productivity (high speed), high flexibility, miniaturization, etc.

[Brief explanation of the drawing]

Fig. 1 is a sectional view showing an outline of an apparatus according to an example of the present invention, Fig. 2 is an overview of the blade, Fig. 3.4 is a sectional view and plan view of the reticle stage, and Fig. 5 is an overview of the optical system. 6.7 is an overview and cross-sectional view of the wafer stage, 8.9 is a two-direction drive block diagram and an X (Y) direction drive block diagram of the wafer stage, and 10.11 is a 9-dimensional diagram of the wafer stage. 12 is an overview diagram showing an example of the off-axis optical system in FIG. 5, FIG. 13 is a top view of the wafer, FIG. 14 is a block diagram of the entire device, and FIG. 15A ,
B and C are diagrams showing each of the television monitors, Fig. 16A and B
16C and 16D are diagrams illustrating the state of exposure to a wafer. 17A, B, C, and D are diagrams showing the aperture relationship between the reticle and the blade. 18. 19 is a diagram illustrating an example of dividing a television screen, FIG. 19 is a diagram illustrating addition and slice levels, FIG. 20 is a control block diagram thereof, FIG. 21 is a flowchart for explaining its operation, and FIG. 22 81-S5 , FIG. 22 81-S5, FIG. 22 81
-S5 and FIGS. 25-25 01-C3 are flowcharts explaining the operation of each alignment mode. ■0...Exposure optical system, As...TTL alignment optical system, RT...Reticle, R8...Reticle stage, Po...Reduction projection lens system, OA...Off-axis alignment Optical system, WF...
・Wafer, WS...Wafer stage, LZ...Laser interferometer. Figure 1 Figure 2 Figure 31! i Figure 18N Figure 19

Claims (1)

[Scope of Claims] 1. An alignment device that operates the pulse motor in the quotient when the amount of movement is divided by the resolution of the pulse motor, and operates other fine drive means in the remainder. 2. The alignment device according to claim 1, wherein the fine driving means includes a piezoelectric element. 3. The alignment device according to claim 1, wherein the fine drive means includes a servo motor.