JPS62277742A - Registration of template - Google Patents

Abstract

PURPOSE:To conduct the theta alignment of a wafer and the registration of a template with constant high precision in a short time by memorizing a pattern at a first position on a roughly positioned reference wafer and comparing said pattern at a second position each separated by integer times as large as x and y indices in the x and y directions from the first position. CONSTITUTION:x and y stages are moved so that the position of (X3, Y3) are brought under a camera first, and a pattern at the position of (X3, Y3) is memorized to a picture recognizer 14 as the pattern of a template. The x and y stages are each shifted by integer times as large as ix and iy in the x and y directions, a section (X4, Y4) in the vicinity of the center of a wafer 4 is brought under the camera, said template and a pattern positioned under the camera at that time are compared by the device 14 at the section (X4, Y4), and (X4, Y4) is used as the position of the registration of the template when consistency is higher than a reference value. The x and y stages are each transferred by integer times as large as ix and iy, a section (X5, Y5) in the vicinity of a left end is brought under the camera, and a pattern is compared similarly by the device 14 and (X5, Y5) is employed as the position of the registration of the template.

Description

Detailed Description of the Invention 3. Detailed Description of the Invention [Industrial Applications] The present invention relates to a template registration method for wafer alignment.

[Prior art] A prober is a device that mechanically measures the electrical characteristics of a semiconductor integrated circuit fabricated on a silicon wafer using a probe needle to input/output terminals (buds) on a wafer chip through a wafer process. This is a device that tests chips by contacting them and transmitting and receiving signals with the tester, and marks chips that are determined to be defective as a result.

Chips are arranged in an orderly manner vertically and horizontally on the wafer. Each chip has the same pattern and pads in the same location. To inspect such a wafer using a blow bar, when the probe card is fixed at a position where all of the probe needles can contact the pads for one chip,
Even if other chips are brought under the probe needles, all probe needles must come into contact with the pad. In this case, in order to easily bring the other chips under the probe needle one after another (step movement), it is necessary to align the wafer with respect to the It is necessary to orient the wafer so that the vertical and horizontal alignment directions of the chips manufactured in the process match. Furthermore, in order to bring the probe needle into contact with all chips, it is necessary to know how the chips are arranged on the wafer.

The operation to satisfy the above requirements is called alignment.

Conventionally, in this type of alignment, the wafer that serves as the reference for alignment, for example, the first wafer in a lot, is manually aligned and the data (template) that serves as the reference for alignment is taken from that wafer, and the data (template) that serves as the reference for alignment is taken from that wafer. Alignment was automatically performed based on that data. This manual alignment for the first wafer includes only θ alignment, and this θ alignment has two stages: coarse θ alignment and fine θ alignment.

Rough θ alignment has been performed by detecting the edge of the wafer at several appropriate locations using a capacitance sensor or the like, or by a mechanical method.

Fine θ alignment, or so-called wafer θ alignment, in which the running of the xY stage in the The scanning was performed by an operator who operated a joystick or the like to rotate the chuck while viewing the scanning using a microscope or the like. Then, from the wafers whose θ has been matched as described above, the operator moves the stage by operating a joystick while viewing the pattern on a CRT, etc., and determines whether or not the wafer is passing as a template, and selects the number of templates required for automatic alignment. was registered.

[Problem to be solved by the invention] However, in this method, the θ of the first wafer
Alignment was difficult and time consuming, and its accuracy depended on the operation of the operator. In other words, since the alignment accuracy depends on the alignment accuracy of the first wafer aligned by a person, there is a drawback that alignment accuracy is not constant. Another disadvantage is that it takes time to register a template, and it is difficult to find a pattern suitable as a template.

SUMMARY OF THE INVENTION An object of the present invention is to provide a template registration method that can perform wafer θ alignment and template registration related to template registration in a short period of time and with constant high precision, in view of the above-mentioned drawbacks of the conventional example.

[Means and effects for solving the problems] The present invention has the following features:
A template registration method in a position detection device for detecting the relative position of a roughly aligned wafer by a pattern matching method using an image recognition device to align the wafer with higher precision, the method comprising: storing a pattern at a first location on the coarsely aligned reference wafer;
The pattern is compared at a second location that is an integral multiple of the X index in the X direction and/or an integral multiple of the X index in the is calculated to match the θ of the wafer, and in this state a template to be used for alignment of other wafers is registered.

In this way, by formulating a template registration method using an image recognition device, automation becomes easy.

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

FIG. 1 is a schematic diagram of a stage of a prober according to an embodiment of the present invention. In the figure, 1 is an X stage that moves in the X-axis direction, 2 is a y-stage that moves in the X-axis direction, and 3 is a y-stage that moves in the X-axis direction.
4 is the wafer, 5 is the X-axis near pulse motor that moves X stage 1, 6 is the y-axis near pulse motor that moves X stage 2, and 7 is the analog distance to wafer 4. Capacitive sensor that outputs data, 8
is a camera.

FIG. 2 is a block diagram showing the configuration of a control system according to an embodiment of the present invention. In the figure, 11 is a CPU that performs calculations and control. 2 is a memory that stores programs and data that operate the CPU II, 13 is an ADC 114 that converts analog data from the capacitance sensor 7 into digital data, and compares the memorized image with the image output from the camera 8 to obtain the amount of deviation. 15 is an interface for operating the image recognition device 14 by the CPU 1l, 17.19 and 16.18
20.21 is the controller and driver for driving the linear pulse motors 5 and 6 by the CPU II, and 20.21 is the C
A display for PUII to display characters and its controller, 22 and 23 a panel with switches, LEDs, and joysticks and its interface, 2
4 is a θ-axis pulse motor that rotates the chuck 3 by θ, 25 is a Z-axis pulse motor that moves the chuck 3 up and down, 26 and 27 are a controller and driver for driving the pulse motors 24 and 25 by the CPU 11, and 28 is a CPU II and memory 12 and interface etc. 13; 15.1
This is a bus that connects 7, 19, 20, 21, and 22.

FIG. 3 is a sectional view of the wafer edge portion of the apparatus of FIG. 1.

First, a method for rough θ matching according to an embodiment of the present invention will be explained with reference to FIGS. 1 to 3. Rough θ matching is performed using the same principle as in the prior art, but all control is performed by the CPU II using a program stored in the ROM.

First, the CPU II is the X and y axis near pulse motor controller 17.19 and the x and y axis near pulse driver 16.
．． Drive control of the X stage 1 and the X stage 2 is started by moving the x and y axis near pulse motors 5.6 through 18. At the same time, the capacitance sensor 7 detects A
D Ct3=Start measuring the distance from the capacitance sensor 7 to the wafer 4 and wait for the edge of the wafer 4 to be reached.

Here, the edge is defined as a point 100 microns below the top surface of the wafer 4. That is, as shown in FIG. 3, the distance from the capacitance sensor 7 to the top surface of the wafer 4 is 500μ
In this case, the edge of the wafer 4 is where the capacitance sensor 7 indicates 600μ. Then, when the edge of this wafer 4 is detected, X at that time.

Store the position of X stage 1.2 on memory.

After performing this at appropriate locations all around the circumference, the center position of the wafer 4 and the angle formed by the direction of the orientation flat of the wafer 4 and the X axis are calculated from these data. Then, the chuck is rotated by that angle to align the direction of the orientation flat with the direction of the X-axis, completing the rough θ alignment.

FIG. 4 is a diagram showing the positional relationship between the wafer 4 and camera 8 during fine θ adjustment, and FIG. 5 is a flowchart for thorough θ adjustment. In Fig. 4, W is the moving direction of the X stage (X
a is the position of the pattern to be memorized by the image recognition device 14 (Xl, Yl), b is the position where the camera 8 will be when the X stage 1 is moved several indexes 6 after memorizing the pattern at position a, C is the position where there is a pattern that is the same as the pattern at position a, and α is the deviation vector (ΔX1+
Δy+), u is the direction of the scribe line calculated from this deviation vector α, and d is the direction from position a to X along direction U.
The displacement vector (ΔX2+ΔY2) output by the image recognition device 14 is the position where the camera 8 is located at the position where the camera 8 is located at the position d, where e is the position where the same pattern as the pattern a exists, V is a more accurate scribe line direction considering the deviation vector β, θ
1 is the angle formed by the direction U and X@, and θ2 is the angle formed by the direction V and the X glaze, which is the amount by which the chuck 3 is actually rotated.

Fine θ alignment is also performed on the stage and chuck as shown in Figure 1, just like coarse θ alignment, but in order to detect θ deviation,
A camera 8 and an image recognition device 14 are used. The image recognition device 14 memorizes a certain pattern, compares it with other images, and outputs the degree of matching and the deviation in the x and x directions. Further, on the wafer 4, chips having the same pattern are regularly and periodically arranged in the vertical and horizontal directions. In other words, a pattern within one chip is located within the same chip within another chip. Therefore, if the scribe lines of the wafer are aligned in the x and
(referred to as index) integer multiple of quantity ix in the X direction or
Even if it moves in the X direction which is an integer multiple of the direction index (hereinafter simply referred to as X index) amount iy, the pattern is the same as before the movement. Here, the index is the sum of the chip size and the width of the scribe line.

Based on the above premise, Toru's zeroing procedure will be explained with reference to FIGS. 4 and 5. Below, regarding the position on the wafer 4, the center of the wafer 4 is the origin □, and the X stage 1
The moving direction of stage 2 is the X axis, and the moving direction of the X stage 2 is the y axis.
Expressed in an x,y coordinate system with axes.

Following the above-mentioned rough zeroing, when starting the characteristic zeroing, first, in step 101, xt X stage 1.2
is moved so that the image recognition device 14 memorizes the pattern on the right side of the wafer (positive side of the X-axis). At this time, let the position of the pattern be a (XI, Yl). Next, in step 102, relatively close to position a (an integer multiple of ix (M
x) Move the X stage 1 so that the camera 8 is located at the position III. Then, in step 103, the pattern captured by the camera 8 and the remembered pattern are compared to determine the amount of deviation α (ΔX+++Δy+) in the x and X directions.
Calculate. This deviation amount α is calculated between the X axis and the scribe line (
When converted into the amount of deviation θ1 in the θ direction from u), the following is obtained. In step 104, using this deviation amount θ1, (XI, Yl
) is relatively far to the left side of the wafer and is an integral multiple (N times) of ix in the direction of θ1 (X2.Y2)
Y2 = N −ix−sin θ, +Y
.

Calculate.

x, X stage 1. so that the camera 8 comes to this position d.
2 is moved (step 105), step 106
, compare the pattern captured by camera 8 with the remembered pattern, detect the amount of deviation β (Δx2 + Δy2),
Furthermore, in step 107, the amount of deviation in the θ direction between the X axis and the scribe line (V) is calculated from this amount of deviation β.

In step 108, by rotating the chuck with this deviation amount θ2, the X direction and the scribe line (V) are
The fine 0 adjustment is completed by matching the direction of .

Figure 6 shows wafer 4 and camera 8 in the case of template registration.
FIG. In the same figure, (X+'
, Y+') is the position that comes after the position (xt, Yl) of the pattern stored with the fine zero adjustment from earlier is rotated by θ2. (X3. Y3) is the position of the pattern to be stored as a template, and (Xl').

yt') in the x and X directions by integral multiples of ix and ty, respectively. This is to ensure that (X3.Y3) has the same pattern as (XI'+Yl') passed as a template.

Next, a method for registering a template will be explained using FIG. 6, including the modification of the index size. Here, the template is registered at three locations near the center of the wafer 4 and near both ends on the X axis, but the template pattern is registered only at the first location, and the other two locations are Only positions of the same pattern are registered.

First, move the x, X stage 1.2 so that the position (X3.
Let 4 remember it. Next, move the x and X stages 1.2 in the
4 times in the X direction), the center of the wafer 4 (
X4. Make sure that Y4 'I is below camera 8.

Here, the image recognition device 14 compares the above template with the pattern currently under the camera 8, and if the degree of coincidence is higher than the reference value (X4. Y4) is set as the template registration position. Furthermore, X stage 1 and y stage 2
are integer multiples of ix and iy, respectively (the X-axis direction is n.

-n4 times and m5-m4 times in the X-axis direction) so that the left end (XS, YS) is below the camera 8. Here, as in the case near the center, the image recognition device 14 compares the patterns, and if the degree of matching is higher than the reference value, the position is determined as a template registration position.

After the above operations are completed, the coordinates (XS, Y3), (X
4. By storing Y4) and (XS, YS) on the RAM as template registration positions, the registration of the template and the registration of its registration position are completed.

In addition, when performing index correction, the above (XS,
YS), the next X-direction index is calculated from the X-direction deviation amount Δx5 output by the image recognition device 14, and the X-coordinate of the template registration position X4'=X
Find 3+n4 ・ix' x5 '=x, +ΔX5. Next, the x and y stages 1.2 are moved by an integral multiple of ix and iy, respectively, so that the position (XS, ya) far from the center of the wafer in the X direction is below the camera 8.

Similarly, the image recognition device 14 compares the patterns and if the degree of matching is less than the reference value, the amount of deviation in the X direction ΔY6
Calculate the index in the X direction, and calculate the y coordinate of the template registration position Y4'=Y
Find 4+m4 ・iy'Y,'=Y, +m, ・iy'.

After the above operations are completed, the coordinates (XS, Y3), (X4'
.

Y4') and (XS', YS11 are stored in the RAM as template registration positions, and the index amounts ix' and iy' after correction are also stored in the RAM. This allows for index correction and its correction. Registration of the template and its registered position is completed, and from now on, when stages 1 and 2 are moved, it becomes possible to move using the corrected index amount.

[Modifications of Embodiments] The present invention is not limited to the above-described embodiments, and can be implemented with appropriate modifications. For example, in the above embodiment, the capacitance sensor 7 is used for rough θ adjustment, but an optical sensor may be used instead. Alternatively, instead of using the sensor 7, a method of mechanically aligning the direction of the wafer 4 by bringing a rotating object into contact with the edge of the wafer 4 may be used. Furthermore, an image recognition device can also be used for rough θ matching. This is done by making the magnification of the camera 8 smaller than when using coarse θ adjustment, and comparing a wide range with the image recognition device 14.
This is a method in which θ is adjusted according to the amount of deviation in x and y output by .

The fine θ adjustment is performed only once in the above description, but depending on the reliability of the θ rotation, the θ rotation may be performed again based on the position confirmation after the θ rotation and the result.

In addition, although we did not mention saving the data of registered templates and their registered positions in the above, it is important to save such data in backup RAM, etc., like other data, so that it will not be lost even after the power is turned off. Keep it
If data such as the template can be sent from the backup RAM to the image recognition device 14 when the power is turned on again, alignment can be performed without registering the template again. Furthermore, if the template is stored in an external storage device such as a floppy disk so that it can be read out as needed, alignment can be performed at any time without registering the template again.

Furthermore, if the matching degree between the template and the pattern is poor as a result of the comparison by the image recognition device 14 described above, it is possible that the chip is different from the pattern stored by the monitor chip, etc. The above comparison may be performed again after moving.

[Effects of the Invention] As described above, according to the present invention, fine θ alignment, template registration, and index amount correction can be performed using the image recognition device as described above for a wafer having a repeating pattern with a constant period in the vertical and horizontal directions. Since this is formulated using , the θ alignment and template registration can be performed with high precision and in a short time, and furthermore, by automating these processes, manual operations can be made unnecessary.

[Brief explanation of the drawing]

FIG. 1 is a diagram of a stage of a prober according to an embodiment of the present invention, FIG. 2 is a block diagram showing the configuration of a control system of an alignment system according to an embodiment of the present invention, and FIG. 1 is a cross-sectional view of the wafer edge portion of the apparatus, FIG. 4 is a diagram showing the camera position above the wafer during θ alignment, FIG. 5 is a flowchart showing the procedure for θ alignment of the wafer, and FIG. 6 is: FIG. 7 is a diagram showing the position of a camera on a wafer in the case of template registration. 1: x stage, 2: y stage, 3: chuck, 4: wafer, 5: x-axis linear pulse motor, 6: y-axis linear pulse motor, 7: capacitance sensor, 8: camera, 11: CP U, 12: Memory, 13: ADC. 14: Image recognition device, 15, 23: Interface, 17, 19.21.24: Controller, 1[i, 1
B, 25: Driver, 20: Display, 22: Panel, 26: 'θ-axis pulse motor, 27: Z-axis pulse motor, 28: Bus.

Claims (1)

[Claims] 1. Template registration in a position detection device for detecting the relative position of a roughly aligned wafer by a pattern matching method using an image recognition device and aligning the wafer with higher precision. A method comprising: first storing a pattern at a first location on a coarsely aligned reference wafer; Compare the pattern at a second location twice as far away, calculate the θ deviation of the wafer from the amount of deviation in the x and y positions at this time, adjust the θ of the wafer, and use this state for alignment of other wafers. A template registration method characterized by registering a template. 2. The second location first compares the patterns at a third location that is relatively close to the first location and is distant in the x direction by an integer multiple of the x index, and determines the amount of deviation in the x and y positions at this time. 2. The template registration method according to claim 1, wherein the location is calculated as a location that is an integer multiple of the x index in that direction by obtaining a relatively coarse θ shift from the location.