JPS6042763A - Device and method for alignment - Google Patents

Abstract

PURPOSE:To execute and alignment even in case a defect exists in an alignment mark by detecting an object provided with plural alignment marks, and executing a control by an alignment part mark which has been discriminated as correct. CONSTITUTION:When an automatic alignment is started, the number of pulses and a pulse interval of each part mark group are measured by a laser scanning. When the measurement is ended, the contents of pulse number counters 107-110 are brought to access, and whether the contents of all the pulse number counters have become ''6'' or not is checked. In case the pulse number counter is not ''6'', namely, when there is a part mark which cannot be detected, a part mark whose position shift can be calculated is selected from the detected part mark by referring to a prescribed table. Subsequently, the shift quantity is calculated by selecting interval memories 117-120 corresponding to this part mark, whether it is within an alloable range of a prescribed shift quantity or not is discriminated, and when it is within the range, the alignment is ended. In this way, even when a defects exists in an alignment mark, the alignment can be executed.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a device for positioning an object having alignment marks, such as a semiconductor exposure device that prints an integrated circuit pattern of a photomask or a reticle (hereinafter referred to as a mask) onto a wafer. Regarding alignment mark detection.

Semiconductor exposure equipment has made remarkable progress in line with the development of semiconductor integration technology, and its mechanisms include equipment that performs close-contact or extremely close-up printing, equipment that performs 1-magnification printing using a lens or mirror system, and equipment that processes a reduced projection image onto a wafer in a stepwise manner. There is a wide variety of printing devices, so-called steppers.

One method of aligning the mask and wafer in a stepper is a TTL die-by-die alignment method in which the mask and wafer are aligned for each shot through a projection lens. Conventionally, in this type of equipment, this alignment method uses photoelectric detection to detect alignment marks provided at two or more locations in order to detect misalignment between the mask and wafer in the X direction, Y direction, and θ (rotation) direction. Positioning is performed by doing this. However, in the embodiments described below, one alignment mark is shaped like two alignment partial marks, but in addition, there are In the case of installing three alignment marks spaced apart from each other, a shape in which one alignment mark consists of one alignment partial mark may be used, or there are further variations.

After the above automatic alignment of the mask and wafer is completed, the mask is illuminated and the mask image is printed onto the wafer. However, during the process of etching the wafer and performing the circuit forming process, the alignment marks are Misalignment or partial loss is likely to occur, and if a portion of the alignment mark is distorted or missing, an appropriate signal cannot be obtained, making alignment impossible for that shot. Therefore, the exposure of that shot is given up and the process proceeds to the next shot, but the time spent searching for detection until it is determined that alignment is difficult is wasted, and there is also the inconvenience that the chips in that part of the wafer cannot be used. .

An object of the present invention is to realize alignment even when an alignment mark provided on an object to be aligned has a defect, using other alignment marks or portions without defects.

In order to achieve this purpose, there is provided a device that detects an object provided with a plurality of alignment marks consisting of one or more alignment partial marks, and automatically aligns the object based on the detected detection signals. means for discriminating between proper signals and improper signals;
Means is provided for performing alignment control based on the detection signal of the alignment partial mark corresponding to the signal determined to be appropriate.

Hereinafter, the present invention will be explained in detail according to the drawings,
Figure 1 (3) shows the main components of a computer (-) equipped with an alignment mark detection system. In the figure, 1 is a mask equipped with an integrated circuit pattern and alignment marks. It is held on a stage 24.

3 is a reduction projection lens, and 4 is a wafer having a photosensitive layer and an alignment mark. 5 is a wafer stage which holds '):f-/N-4. The mask stage 24 and the wafer stage 5 are arranged in the plane (X direction, Y direction) and in the rotational direction (θ
direction).

Then, for example, the wafer stage 5 is sent in the XY directions to perform a step-and-repeat motion, the mask stage 24 is moved in the XYθ directions to initialize the mask 1 in the apparatus, and then the mask stage 24 is moved in the XYθ directions. The mask 1 and the wafer 4 are aligned by moving in the XY directions and also moving the wafer stage 5 in the θ direction. However, alignment may be performed by moving the mask stage for both XYo, or conversely, alignment may be performed by moving only the Ueno 1 stage.

Also, 20 and 20' are alignment marks of mask 1, 2
1 and 2f are alignment marks on the wafer 4. In the figure, one picture of the alignment marks on the wafer 4 is shown for convenience, but in reality, it is assumed that the same number of sets as the number of exposed shots are arranged.

In addition, the alignment mark on mask 1 and the alignment mark on the way are aligned so that the dimensions of both alignment marks do not change even if projection or back projection is used when a system other than the same-magnification projection system is used. It is assumed that the dimensions of the marks are changed, and here the dimensions are set so that the reduction magnification is obtained by dividing the dimension of the alignment mark on the wafer by the dimension of the alignment mark on the mask.

28 is a rotating polygon mirror that rotates around a rotation axis 29. The laser beam emitted from the laser light source 22 is converged by the lens 23 to a point 31 on the surface of the rotating polygon mirror 28. Lenses 32, 33, and 34 are lenses for relaying the beam, and 35 is a triangular prism. Since the apex of the laser beam coincides with the optical axis, one scan of the laser beam is divided into the first half and the second half, left and right. Reference numeral 39 denotes a prism block for converting laser light scanning in a direction within the drawing to scanning in a direction perpendicular to the drawing, and is configured in a shape as shown in FIG. 1(B), for example. 42 is a photoelectric detection system (43,44°45.4
6), 43 is a lens, 45 is a spatial filter, 46 is a condenser lens, and 47 is a photodetector. 48, 49, 50,
51 is a total reflection mirror, 52 is a prism, and 53 is an f-θ objective lens.

Further, 57 is a condenser lens, and 58 is a photodetector for detecting a synchronization signal. A part of the laser light incident on the semi-transparent mirror 42 is condensed by the condenser lens 57 and enters the photodetector 58. The photodetector 58 thus serves to detect the start and end points of the laser beam scan.

As you can see from the figure, the signal detection system consists of completely symmetrical left and right systems.If the operator side is the front side of the paper, the system shown with a dash is the signal on the right, and the system without a dash is the signal on the left. We will call it the detection system, and the same parts are given the same numbers with a dash.

The relay lenses 32, 33, and 34 form the origin of the deflection from the rotating polygon mirror 28 at the bell 56 at the aperture position 55 of the objective lens 53. Therefore, the laser beam from the sheet or slot 14 is scanned over the mask and wafer by the rotation of the rotating polygon mirror 28.

In addition, in the objective lens system, the objective lens 53, the aperture 955
, the mirror 51 and the prism 52 are movable in the XY directions by a moving means (not shown), and the measurement positions of the mask 1 and the wafer 4 can be changed arbitrarily. For example,
When the mirror 51 moves in the direction indicated by the arrow A in the figure, the objective lens 53 and the diaphragm 55 simultaneously move in the incoming direction, and the prism 5 moves in the direction shown by the arrow A in order to keep the optical path length constant.
2 also moves in the incoming direction by 1/2 the amount of movement of the mirror 51.

On the other hand, when moving in the Y direction, the entire optical system for position detection moves in the Y direction (direction perpendicular to the plane of the paper).

Since the objective lens system is movable, it is possible to detect marks in cases where the position of the alignment mark differs depending on the mask, or in cases where a mask setting mark is provided separately from the alignment mark.

FIG. M2 (A) is a diagram showing 7 alignment marks on the mask surface. Mark 20 (FIG. 1) on mask 1 consists of orthogonal bars 75 and 76, and similarly mark 20' consists of bars 75' and 76'.

Furthermore, when the marks 21 and 21' on the wafer are back-projected onto the mask 1 through the projection lens, the mark 21 consists of parallel bars 71, 72, 73° 74 that are perpendicular to each other; 72', 73'.

Consisting of 74'. Details of this alignment mark are described in Japanese Patent Laid-Open No. 53-90872.

Here, the alignment mark is divided into two by the axis of symmetry and is called as follows.

Bar 71.72.75 marks the first part bar 73.7
4.76 second part mark par 72', 73'
, 76' to the third partial mark par 71', 72', 7
5' marks the fourth part of the laser beam as shown in Figure 2.
As shown in Figure 0', the left detection system scans the hair alignment mark in the Y direction from the front of the device (bottom of the page), and the right detection system scans the minus Y one-way hair alignment mark from the back of the device (top of the page). (B) is an enlarged view of the left detection system mark in FIG. 2 (Al).

When the laser beam 60 scans the mark in Fig. 2 fB),
The photodetector 47, 4τ has 87 as shown in Fig. 2 fc).
Signals 1 to S76 are obtained. S71°S72. S73
. . are signals corresponding to marks bars 71, 72, 73, . . . , respectively.

Therefore, the detection signal S71. S75. S72. S73゜8
By measuring the intervals Wl, W2, W3, and W4 of 76.874, the amount of misalignment between the mask and the wafer can be detected.

X of the first partial mark and second partial mark of the left signal system
The amount of deviation △XL, △YL in the direction and Y direction is ΔXL=(Wl-W2-W3+W4)/4...Formula (
1) ΔYL= (-W1+W2+W3-W4)/4
... can be expressed by formula (2).

Also, the amount of deviation of the third partial mark and fourth partial mark of the right signal system, ΔXR, ΔYR, is ΔXR=(-W1'+W2
'+W3'-W4')/4 -131△YR=(-W1
'+W2'-W3'+W4')/4 = Equation (4) Therefore, the amount of deviation in the rotational direction △θ is △θ=(△YR-△YL)/(XR-XL)...Equation (5) However, XL and XR are the positions of the left mark and right mark from the center of the mask, and (XR-XL) is the distance between the left and right marks.

FIG. 3 is a diagram showing an example of shot arrangement on a wafer.

In the figure, the area indicated by P is the exposure pattern area exposed in one shot, and the number of steps T per exposure shot is 4.
It exposes each individual at the same time. At this time, the alignment mark is provided at the position indicated by L L' in the scribe line between the exposure shots. In other words, the first and second alignment partial marks are provided at the position indicated by L in the scribe line. , L' are provided with third and fourth alignment portion marks.

In this embodiment, even if one or two of the four alignment partial marks cannot be detected properly, if at least two partial marks can be detected, automatic alignment is performed based on the detection signal.

Table 1 shows combinations of partial marks. In the table, ○ marks are partial marks that could be detected, and X marks are partial marks that could not be detected due to defects.

Table 1 These can be summarized as follows.

1. Alignment is possible if three of the four partial marks can be detected.

2. Alignment is possible if the following two partial marks among the four partial marks can be detected.

3. However, alignment is not possible if only the following two partial marks can be detected. Therefore, when alignment is performed using only two partial marks that step the wafer to the next shot, the mask and wafer should be aligned in the X and Y directions so as to correct the positional deviations △ , but no correction in the rotational (θ) direction is performed. The reason why the rotational direction is not corrected is that the reliability of the moving stage feeding mechanism is high at present, and there is almost no error in the rotational direction during linear movement of about 1 to 2 steps. This is because when a step movement is performed from a corrected shot, if the X and Y directions are corrected, the θ direction is sufficiently included in the allowable range.

The amount of positional deviation ΔX and ΔY when alignable partial marks are combined can be determined from the following formula 0 1. First partial mark and second partial mark...the above equations (1) and (2) Partial mark and fourth part mark...The above formula (
1) and (2) 3. First partial mark and first partial mark △X=(Wl-W2+W3'-W4')/4...Formula (6) △Y=(-Wl+W2-W3'+W4') /4
...Formula (7) 4, second partial mark and first partial mark ΔX=(-Wr+W2'-W3+W4)/4...
Formula (8) △Y=(-W1'+W2'+W3-W4)/
4, ...Equation (9) Figure 4 is a block diagram of a circuit that determines whether the detection signals of the four partial marks are appropriate or not and stores information, and Figure 5 shows the operation of each part of the circuit in Figure 4. FIG. 2 is a diagram showing waveforms for explaining.

47 and 47' are alignment mark detection photodetectors indicated by the same numbers in FIG. 1, and the signal waveform of the alignment mark shown in FIG. A signal waveform is outputted from the photodetector 47' in the form of an alignment partial mark shown at f8) in FIG. Note that in FIG. 5A (Bl), the waveforms indicated by Gl, G2, G3, and G4 correspond to the first, second, third, and fourth partial marks, respectively.

On the other hand, 58 and 58' in Fig. 4 are photodetectors for detecting synchronization signals indicated by the same numbers in Fig. 1, and their outputs have the waveforms shown in Fig. 5 (F) and (G). .

The left detection system synchronization signal (F) and the right detection system synchronization signal ((
Wu is input to the control timing generation circuit 101. As described later, the control timing generation circuit controls the entire detection circuit shown in FIG. 4 under the control of a microprocessor (not shown), and the control timing generation circuit 101 connects the analog switch via a control line 102. Control 103.

The control line 102 is, for example, a left synchronization signal (F), and the analog switch 103 selects the left detection system and right detection system signals based on this signal, and outputs the composite waveform shown in FIG. 5(C). do.

This synthesized waveform fC) is amplified by a video amplifier 104, and then processed by a binarization circuit 105 at an appropriate slice level to form digital waveforms S71, S75 .
It is converted to 872... Therefore, the above-mentioned bar intervals Wl, W2... are 871 and S75°S75 and 8
The interval is 72...

Next, the edges of the digital waveform (D) are detected by the edge detection circuit 106, and the waveform shown in FIG. 5(E) is generated.

That is, the rising edge of S71 is converted into the rising edge 9 of 571R, and the falling edge of S71 is converted into the rising edge of 571F. Therefore, for example, in order to increase the bar interval W1, the pulse interval (S75R+875F)/2- (871R+871F
)/2 should be measured.

Edge pulse (E) is (respectively number counter 107°10
8. Input as clock ψ pulses of 109 and 110.

First, the control/timing generation circuit 101 generates a partial mark selection signal (.eta.(I)U) (K) shown in FIG. 5 based on the synchronization signal (F) (Gl).

These signals fH) fl) (J) (8)) are, respectively,
1st. 2nd゜3rd. This is a signal for selecting the fourth partial mark, and the number counters 107, 108, 109, 110 are
Controlled by the signal (Group (1) (J) (10).

The number counter 107 counts the number of pulses of the first partial mark group, the number counter 108 counts the number of pulses of the second mark group, the number counter 109 counts the number of pulses of the third mark group, and the number counter 110 counts the number of pulses of the fourth mark group.

Pulse interval 5751 (,,575F,571R,571
The method of measuring F is performed by using sampling pulses that are sufficiently shorter than these pulse intervals.

This sampling pulse is generated by the control timing generation circuit 101, and is generated by the signal line 111.
It serves as a clock input for interval counters 112.113114.115. This interval counter 112.113, -11
4.115 are the partial mark selection signals (11
(1) Sampling pulses are counted from the rising edge of the selection signal only in the interval U)fK).

On the other hand, edge pulse (E) is write timing circuit 1
16. The write timing circuit 116 uses the interval memory 117 and 118 based on the edge pulse (E).
．． 119.120 write signals are generated.

The write timing circuit 101 is controlled by the control timing generation circuit 101, and generates a write signal during measurement, but generates a read signal after the measurement is completed.

Interval Memo+J1i7゜118, 119, 120 is a 16-pit x 6 random access memory with data input/output of the rate type, and its read/write control is performed by the write timing circuit 116 described above. The chip select signal is sent from the control/timing generation circuit 101, and the address signal is sent from the number counter 1 during measurement.
By the output data of 07, 108, 109° 110, after the measurement is completed, the address bus 1 of the microcomputer is
35.

Therefore, after the measurement is completed, the pulse data shown in Table 2 is stored in each interval memory.
Interval memory 117°118.119,1 via 133
20 and the data is stored in buffers 126 and 12.
7,128,129, and are taken into the microprocessor via data bus 134.

On the other hand, the number of pulses of each partial mark stored in the number counters 107, 108, 109, 110 is also input to the microprocessor via buffers 122, 123, 124, 125.

When the number of pulses is 6, it is assumed that a partial mark is detected (○ mark in Table 1), and when the number of pulses is other than 6, it is assumed that a partial mark is not detected (x mark in Table 1).

Next, FIG. 6 shows a flow diagram illustrating the operation of the auto-alignment method of the present invention.

When automatic alignment starts at 601, first, at step 602, partial marks are measured by laser scanning.

This measurement is to measure the number of pulses and the pulse interval of each mark group, as explained in FIGS. 4 and 5.

When the measurement is completed, in step 603, the
The contents of the book number counters indicated by 110 are accessed, and it is checked whether the contents of all the book number counters are 6.

When the contents of all the number counters are 6, it is assumed that all four partial marks have been detected, and the contents of the interval memories 117, 118, 119, and 120 are read in step 604, and then in step 605, the above-mentioned The formula (11~
The amount of deviation is calculated using equation (5). In this case, the amount of deviation in the X direction and the Y direction is taken as the average value of the left and right marks.

On the other hand, if the number counter is not 6 in step 603, that is, if there is a partial mark that cannot be detected, the process branches to step 610, and the table based on Table 1 is referred to.
From the information on the detected partial marks, a partial mark whose positional shift can be calculated is selected. Next, in step 611, the interval memory corresponding to these partial marks is selected, its data is read 9, and the above-mentioned equation (6) (71 (8) (9
), and in step 612, the amount of deviation is calculated using a well-known method.

Branch 604 → 60 regardless of whether a partial mark is detected or not.
5 or the amount of deviation △X, △Y found at branch 610 → 612
It is determined in step 606 whether or not the amount of deviation is within a predetermined tolerance range for the amount of deviation.

If it is within the allowable range, the process branches to step 613 and the alignment ends.

If it is outside the allowable range, alignment is performed, but if four partial marks are detected at this time, driving in the X direction, Y direction, and θ direction is performed, and if four partial marks are not detected, the positioning is performed. performs driving only in the X and Y directions.

Therefore, in step 607, it is determined whether or not there is θ drive. If there is, θ drive is performed in step 608, and if not, the process branches to step 609. In either case, in step 609, the mask and wafer are moved relative to each other in order to correct the amount of deviation in the X and Y directions. This movement in the X direction, Y direction, and θ direction may be performed by either the mask stage or the wafer stage. After the stage drive is completed, the process returns to step 602 again, the mark interval is measured, and the loop from steps 602 to 609 is repeated until the mark interval is within the allowable range.

In addition, in the above embodiment, the number of bars detected was counted and the bar spacing was memorized, but it is also possible to divide the mode into a mode in which the number of bars is counted and a mode in which bar spacing information is taken in for appropriate partial marks. good. Furthermore, instead of scanning the alignment mark with a spot light and picking up the scattered light therefrom, an alignment mark image may be formed and scanned with a slit or with an image sensor. Furthermore, in addition to relatively aligning a mask and a wafer, the present invention can also be applied to aligning a wafer with respect to a device body in circuit testing equipment or the like.

As described above, according to the present invention, even if the alignment mark is distorted or missing and cannot be detected properly, the correction amount can be calculated by selectively combining other alignment marks or their partial information. , which is a correction drive, has the effect of realizing automatic alignment of parts where alignment has traditionally been delayed, or performs exploratory detection operations in order to obtain complete detection information, resulting in giving up on alignment. This has the effect of eliminating wasted time.

At times, the success rate of alignment can be dramatically improved, so when applied to a stepper, the number of acquired chips can be significantly increased and throughput can be improved.

[Brief explanation of drawings]

FIG. 1 (N is a front view of the embodiment of the present invention, (B) is a perspective view of the constituent members. FIG. 2 (5) is a plan view of the mask and Ueno alignment marks on the mask surface, (
B) is an enlarged view of the alignment mark, and (C) is an output signal diagram when the alignment mark is optically scanned. FIG. 3 is a plan view showing an example of shot arrangement. FIG. 4 is a circuit block diagram according to the embodiment. Figure 5 (4) ~
(6) is a signal waveform diagram. FIG. 6 is a flow diagram showing the alignment process according to the embodiment. In the figure, 1 is a mask, 3 is a reduction projection lens, 4 is a wafer, 5 is a wafer stage, 20.20' @21.2
r is alignment mark, 22 is mask stage, 4
7 and 47' are photodetectors, 58 and 58' are photodetectors for synchronizing signal detection, 71 to 76 and 7f to 76' are bars forming alignment marks, 71 and '12.75 are first partial marks, 73 and 74.76 are second part marks, 72' and 7
3′. 76' is the third part mark, 71', 72', 75'
is the fourth part mark, 107 to 110 are the number counters, 1
12-115 are interval counters, 116 are write timing circuits, 117-120 are interval memories, 122-133
is a buffer.

Claims (1)

[Claims] +1) In an alignment device that detects a plurality of alignment marks having alignment partial marks and automatically aligns an object based on the detected detection signals, an appropriate one of the detection signals is detected. An alignment device characterized by comprising means for discriminating between a signal and an inappropriate signal, and a means for performing alignment control based on a detection signal of an alignment partial mark corresponding to a signal determined to be proper. (2) The alignment apparatus according to claim 1, wherein the alignment apparatus is a semiconductor exposure apparatus that prints an image of a photomask on a wafer stepwise. (3) The alignment mark is 2. The alignment device according to claim 1, wherein the alignment partial marks are combined. (4) a step of detecting a plurality of alignment marks provided on the object to be detected, each having an alignment partial mark; and a step of identifying an alignment partial mark, among the alignment marks, on which prescribed detection is to be performed; An alignment method characterized by performing seven alignments of the object based on detection signals of alignment partial marks.