KR100765491B1 - Method for automatic alignment of wafer - Google Patents

Abstract

A method for automatically aligning a wafer is provided to improve the stability and reliability of alignment by using transform equations of matrix and coordinates. An align error is measured to align a reference pattern and a wafer pattern mounted in a wafer measuring step(S440). A horizontal transform matrix and a translation transform matrix are calculated according to the calculated align error, and a correcting value of the pattern is determined by using coordinates transform of the matrix(S450). A wafer is aligned by utilizing the determined correcting value(S460). The above steps are repeated so that the align error of the wafer satisfies the allowable range(S470,S480).

Description

Method for automatic alignment of wafer

1 and 2 are schematic diagrams for illustrating the concept of automatic alignment in the dicing process of the wafer,

3 is a block diagram showing a system of automatic alignment according to an embodiment of the present invention;

4 is a flowchart sequentially illustrating a method of automatically aligning wafers according to the present invention;

5 is an exemplary view showing a cutting direction of a blade;

6 is an exemplary view showing a cutting line of a wafer;

7 is an exemplary view showing a pattern displayed on an observation screen of a control computer;

8 is an exemplary view showing each center point by enlarging the center of the upper surface of the turntable;

9 is an exemplary view showing a distance relationship between a center point of a wafer and a pattern.

The present invention relates to a method for aligning an object on which a pattern is formed on a surface, such as a wafer or a printed circuit board (PCB) of the present invention, and more particularly, to an automatic wafer alignment method using coordinate transformation of a matrix.

The semiconductor manufacturing process has a high degree of automation compared to other fields because the relationship between yield and yield is important. In the past, the semiconductor production line, which was formed mainly by manpower, has recently shifted to facilities in order to lower production costs and defect rates.

One of the biggest problems in the automation of semiconductor production is the problem of automatic alignment of semiconductor wafers, in which the wafers of the previous step are mounted on the process means for the next step, and then the exact position of the work is adjusted. I'm at work.

However, the above work requires a lot of time and manpower because the processing precision of the semiconductor is a few nano to several microns, requires a lot of time and manpower, thereby increasing the semiconductor manufacturing cost.

The conventional wafer alignment process moves the transfer mechanism to a reference position while observing a wafer adsorbed by vacuum or the like on a table under a microscope. In general, the alignment operation moves the axis in the x-y-θ three directions on a plane, and repeats until the mark positions at both ends of the wafer surface match the reference positions.

Cutting a semiconductor wafer in accordance with a plurality of chip designs formed on the wafer is referred to as a dicing process, which will be described in detail with reference to FIGS. 1 and 2 to the wafer 110 mounted on the rotating table in the dicing process. When the dicing blade 120 is in contact with the dicing direction 130, the cutting line 140 is formed on the wafer surface.

However, if the wafer is not aligned, an error occurs between the sawing line 150 and the cutting line 140 of the wafer, which is the line of the wafer to which the dicing process is actually performed. Done.

Therefore, a process of matching the actual cutting line 140 and the sawing line 150 of the wafer by appropriately adjusting the mechanism so that the pattern 170 observed at both ends of the observation points 160a and 160b is parallel to the blade is required.

As described above, in order to align the wafer, in the "wafer alignment method of the sampling equipment (domestic patent, publication number 10-2002-0036086 (2002.05.16))", a cross mark is placed on the bottom of the wafer to reduce the alignment time of the wafer. A method of engrave and aligning a wafer is disclosed.

In addition, in the "semiconductor wafer alignment method (domestic patent, publication number 2001-0026895 (2001.04.06))", the distance between the patterns is measured at a plurality of points, and moved to the position showing the result closest to the reference pattern among the measurement patterns. A method of aligning a wafer with the average value measured is disclosed.

However, the methods disclosed in the two published patents are difficult to apply to the dicing process in which dust and cutting water are ejected as a method for the photo masking process.

Therefore, in order to align the wafers of the semiconductor, it can be applied to the dicing process, a method that can automatically align the wafers by measuring a reliable alignment error and a simple way to avoid overloading the system in calculating the alignment error There is a need for a method that can be applied to an alignment system by calculation to align wafers.

SUMMARY OF THE INVENTION The present invention solves the above problems, and created by the necessity as described above, the object of the present invention is to provide a wafer automatic alignment method that can automatically align the wafer using a pattern recognition method and the coordinate transformation of the matrix. have.

Other objects and advantages of the invention will be described below and will be appreciated by the embodiments of the invention. In addition, the objects and advantages of the invention may be realized by the means and combinations indicated in the claims.

In order to achieve the above object, the wafer automatic alignment method of the present invention sets a reference pattern as a reference for automatic alignment by fixing a reference wafer for setting a pattern to a rotating table, and stores the reference pattern in a control computer. Pattern setting step; A reference position setting step of photographing the reference wafer by a camera and placing the reference pattern at a reference point of the control computer screen to set a reference position of observation; A wafer mounting step of removing the reference wafer from the rotation table and mounting the wafer to the rotation table such that a pattern of the wafer to be aligned is on the observation screen of the control computer; An alignment error measuring step of measuring an alignment error for alignment of the reference pattern and the wafer pattern mounted in the wafer mounting step; Compensation value determination step of determining the compensation value of the pattern by calculating the horizontal transformation matrix (T) and the translational transformation matrix (R) of the pattern according to the alignment error measured in the alignment error measurement step using the coordinate transformation of the matrix ; A compensation value applying step of aligning the wafers by applying the compensation value determined in the compensation value determining step; And an alignment error checking step of repeatedly performing the alignment error measuring step, the compensation value determining step, and the compensation value applying step so that the alignment error of the wafer aligned in the compensation value applying step satisfies a preset allowable range. do.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, terms or words used in the present specification and claims should not be construed as being limited to the common or dictionary meanings, and the inventors should properly explain the concept of terms in order to best explain their own invention. Based on the principle that it can be defined, it should be interpreted as meaning and concept corresponding to the technical idea of the present invention.

Therefore, the embodiments described in the specification and the drawings shown in the drawings are only the most preferred embodiment of the present invention and do not represent all of the technical idea of the present invention, various modifications that can be replaced at the time of the present application It should be understood that there may be equivalents and variations.

Figure 3 is a block diagram showing a system of automatic alignment according to an embodiment of the present invention, Figure 4 is a flow chart sequentially showing the automatic wafer alignment method of the present invention.

As shown, the automatic wafer sorting method according to an embodiment of the present invention, the reference pattern setting step (S410), the reference position setting step (S420), wafer mounting step (S430), alignment error measurement step (S440), compensation value The determination step S450, the compensation value applying step S460, and the alignment error checking steps S470 and S480 may be included.

Referring to FIG. 3, a system for automatically aligning a wafer of the present invention will be described. A pattern is formed on a surface of the wafer 310 to perform alignment, and the present invention is a wafer 310. The present invention is not limited thereto, and may be applied to various products on a plate having a pattern formed on a surface such as a printed circuit board (PCB).

The wafer 310 is mounted on the rotatable rotating table 320 and fixed, and the one or more cameras 330 are installed in a direction perpendicular to the plane of the wafer 310, and are moved by the transfer means 332. It is movable in the y direction, and acquires an image of the surface of the wafer 310 and transmits the image obtained to the control computer 340, the camera 330 flows in the vertical direction and the surface of the wafer 310 You can focus.

In addition, an illumination control device 350 is provided to adjust the illumination to obtain a precise image of the surface of the wafer 310 to adjust the intensity of the illumination, the control computer 340 is the camera 330 The image of the surface of the wafer 310 obtained from the image can be received to recognize the pattern, and it is responsible for controlling the wafer alignment.

The wafer automatic alignment method of the present invention by the above system will be described with reference to the flowchart of FIG. 4.

The reference pattern setting step (S410) is a step of setting a reference pattern for automatic alignment of the wafer, the illumination to the wafer by the illumination control device 350, the camera on the surface of the wafer 310 After setting the optimum observation screen conditions such as the focus of 330, the reference wafer is placed on the rotary table 320 and fixed so as not to flow.

Thereafter, the user manually aligns the reference wafer, and transmits and stores the pattern to be used for automatic alignment to the control computer 340, so that the reference patterns 512a and 512b of the reference wafer 510 as shown in FIG. Complete the setting of).

The control computer 340 may measure the distance by converting the distance per pixel into a real distance with respect to the image acquired by the camera 330.

In the reference position setting step (S420), the transfer means 332 of the camera 330 finds a position corresponding to the processing line on which the dicing process is performed, and defines two observation positions per line. do.

The two observation positions will be described below by defining the first observation point and the second observation point.

The point where each pattern of both ends of the reference wafer 510 is set as the reference position of observation to the reference point on the screen of the control computer 340, and two reference positions are set for each of the two patterns of one line will be.

Next, the reference wafer is processed by a dicing blade 520 performing a dicing process, and the dicing processing start position of the reference wafer and the parameters necessary for processing are measured and stored in the control computer 340. do.

In the wafer mounting step S430, the reference wafer 510 is removed from the turntable 320, and the patterns 612a and 612b of the wafer 610 to be aligned are displayed on the screen of the control computer 340. The wafer 610 is mounted and fixed to the rotary table 320 so as to come in.

The alignment error measuring step S450 is a step of measuring an error for aligning the reference patterns 512a and 512b and the patterns 612a and 612b of the wafer 610, and in the reference position setting step S420. When the patterns 612a and 612b observed at the set reference positions are displayed on the observation screen of the control computer 340 as shown in FIG. 7, the reference patterns 512a and 512b and the patterns 612a and 612b are displayed. We will measure the alignment error between

This alignment error is measured by observing the pattern 512a of the wafer on the observation screen 710 to measure the y-axis distance 722 with the horizontal reference line 720 and the x-axis distance 732 with the vertical reference line 730. do.

Hereinafter, the coordinate matrix on the plane of the reference pattern is defined as P _t , and the spatial coordinate matrix on the pattern of the patterns 612a and 612b is defined as P _c , and is expressed as Equation 1 below.

Where x _t1 (reference pattern x coordinate of the first observation point), x _t2 (alignment pattern x coordinate of the second observation point), y _t1 (reference pattern y coordinate of the first observation point), y _t2 (alignment pattern of the second observation point) y-coordinate), θ _t1 (reference pattern angle of first observation point), θ _t2 (alignment pattern angle of second observation point),

x _c1 (reference pattern x coordinate of first observation point), x _c2 (alignment pattern x coordinate of second observation point), y _c1 (reference pattern y coordinate of first observation point), y _c2 (alignment pattern y coordinate of second observation point) ), θ _c1 (reference pattern angle of the first observation point), θ _c2 (alignment pattern angle of the second observation point).

Here, since the error angle in the wafer alignment is generally 2 ° or less, the error amount observed by the control computer 340 is defined as δ, and Pc is approximated as shown in Equation 2 below.

The compensation value determining step (S450) calculates a horizontal transformation matrix (T) and a translational transformation matrix (R) of the patterns 612a and 612b according to the alignment error measured in the alignment error measuring step (S440). Coordinate transformations are used to determine compensation values of the patterns 612a and 612b.

Hereinafter, the principle of automatic alignment using a pattern will be described in detail.

Assuming that the point P on the two-dimensional plane moves along with the rotation and translational movement to the point P ', each coordinate may be defined by Equation 3 below.

If the point P is rotated about the rotation center C = [C _x , C _y , 0, 0], then the rotation transformation matrix (R) and translation transformation matrix (T) are defined. .

Here, if the translational and rotational movement amounts are Δx, Δy, Δθ, the horizontal transformation matrix T and the rotation transformation matrix R are defined as shown in [Equation 5].

By properly moving the rotary table 320 in the xy-θ direction based on the measured error, the position P _c on the plane of the patterns 612a and 612b can be made equal to P _t , which is the setting state of the reference pattern. have.

Since this process is composed of a rotational and translational motion, it can be expressed as [Equation 6] by substituting Equation 4 above.

Here, assuming that any compensation value of each axis is a = (a _x , a _y , a _θ , 0), the horizontal transformation matrix T and the rotation transformation matrix R are represented by the following equation (7). Can be.

Here, a _x (x-axis compensation value), a _y (y-axis compensation value), and a _θ (rotation compensation value).

In the wafer alignment process, a portion of the specimen is magnified and observed, and thus observation is required at at least two points to align one processing line 740. Therefore, the reference position Pt, the actual position Pc, and the center point C should be composed of two columns as shown in [Equation 8].

Where x _t1 (transformed pattern x coordinate of first observation point), x _t2 (transformed pattern x coordinate of second observation point), y _t1 (transformed pattern y coordinate of first observation point), y _t2 (second observation point The transformed pattern of the y coordinate), θ _t1 (the transformed pattern angle of the first observation point), θ _t2 (the transformed pattern angle of the second observation point),

x _c1 (converted pattern x coordinate of first observation point), x _c2 (converted pattern x coordinate of second observation point), y _c1 (converted pattern y coordinate of first observation point), y _c2 (conversion of second observation point) Pattern y-coordinate), θ _c1 (the transformed pattern angle of the first observation point), θ _c2 (the transformed pattern angle of the second observation point),

C _x (x coordinate of the center point), C _y (y coordinate of the center point).

The error observed in the general machine vision screen can detect the x-y-θ direction, but when the θ direction is detected, the amount of computation increases and the observation time and system load increase. In addition, since the angle is detected with a limited number of pixels with respect to the localized portion, the accuracy is low.

For example, if a pixel is vertically displaced on a 640 x 480 screen, the machine detects a precision of 9 ° / 100, but in a dicing system the accuracy of the axis of rotation is about 0.1 ° / 100. Therefore, the alignment error δ can measure only two pieces of information, a horizontal distance 722 and a vertical distance 732, in order to shorten the alignment time and improve accuracy. In this case, the alignment error is represented by Equation (9).

On the other hand, for the coordinate transformation matrix T and R by substituting [Equation 7] to [Equation 9] into [Equation 6], the compensation value as shown in [Equation 9] to [Equation 11] Can be calculated.

Here, the tan value of the _angle θ, which is the θ angle compensation value, and the x compensation value and y compensation value of the nth observation point (n = 1 or n = 2) are calculated by Equation 10, 11, and 12. , a _θ (theta angle compensation value), a _xn (the x-axis compensation value at the nth observation point), and a _yn (the y-axis compensation value at the nth observation point). (n = 1 or n = 2)

The compensation value applying step (S460) is a step of aligning the wafer 610 by applying the compensation value calculated and determined in the compensation value determination step, so that the rotation table 320 fits the compensation value in the xy-θ direction. To match the patterns 612a and 612b with the reference patterns 512a and 512b.

In Equation 10, 11, 12, n is the number of observation points, and has a value of n = 1 or n = 2. Ideally, the compensation values obtained from observation point 1 and observation point 2 coincide with each other. The average of each direction can be taken as a compensation value since it may be different.

That is, taking the average compensation value in each direction is expressed by the following [Equation 13].

As shown in FIG. 8, when the center portion 810 of the wafer 610 is enlarged and viewed in an alignment mechanism, the area center 812 of the wafer, the area center 814 of the turntable 320, and the motor may be formed. The centers of rotation 816 of the wafers do not coincide with each other.

Thus, measuring the exact position of the center of rotation 816 of the wafer by the motor is quite difficult and time consuming.

In order to solve this problem, the present invention provides a method for erasing the center of rotation 816 of the wafer. That is, the wafer 610 may be aligned by excluding the rotation center 816 of the wafer.

Specifically, referring to FIG. 9, the distance in the x direction between the pattern 612a and the pattern 612b of the wafer 610 may be generally expressed as a constant l, and a value large enough so that the influence of alignment error is small. Has

Since the position where the wafer 610 is placed is approximately symmetric with respect to the rotation center 910, the x coordinate mean of the pattern 612a and the pattern 612b is similar to C _x, and the difference between the x coordinates is ℓ and similar.

Since the rotation compensation value a _θ is a small value of 2 ° or less, the tan aθ term can be approximated to aθ and the cos aθ term can be approximated to 1.0.

Also, since the difference of each y coordinate is mostly defined on the horizontal plane, the difference of each y coordinate is approximated to 0.0.

Applying the approximated value to [Equation 10], the following [Equation 14] is obtained.

Applying Equation 14 approximated as described above to Equation 13, the compensation value is calculated as in Equation 15.

The compensation value calculated as described above may substantially change the direction of the observation screen for various reasons in designing the instrument.

In this case, it can be solved by the screen direction switching matrix D _v .

The redirection matrix has a diagonal element all 1 or -1 and the rest is filled with 0 and has a value of 1 if it matches the left-hand rule direction and -1 for the opposite direction.

The redirection matrix is expressed by Equation 16 below.

When the direction change matrix Dv of Equation 16 is arranged in connection with Equations 2 and 6, Equation 17 is arranged.

In addition, even when the direction of the transfer mechanism is reversed, the actual compensation value can be calculated by Equation 16 and Equation 17 as described above.

The alignment error checking steps S470 and S480 may include measuring the alignment error, determining the compensation value, and applying the compensation value such that the alignment error of the wafer aligned in the compensation value applying step satisfies a preset allowable range. Repeat this step to reduce the alignment error.

In general, the alignment error can be reduced by compensating the alignment compensation value β at the current position P _c , but the method of precisely reducing the error is to repeat the process of observation and compensation.

In addition, it is difficult to accurately match the center of the wafer to the center of rotation, and there is a difference in the state of recognition marks in the production line.

In other words, assuming that the i-th iteration process is performed, the compensation process may be expressed by Equation 18 in consideration of the convergence coefficient η.

Here, ^{i + 1} P (i + 1 th ordered pattern matrix), ⁱ P (i th ordered pattern matrix), η (convergence coefficient), i (number of alignments), and β (alignment compensation value).

Since this process is repeated until the alignment error falls below the tolerance ε, the condition that the repeating process continues is the same as the condition of Equation 19 below.

That is, when the alignment error δ is larger than the preset tolerance ε, the alignment error checking steps S470 and S480 are repeatedly performed, and the alignment error δ is smaller than the preset tolerance ε. In this case, wafer automatic alignment is terminated.

Here, in performing the repetition so that the alignment error (δ) may come in the allowable tolerance (ε), an error message is displayed when the repetition number is set to exceed the set repetition number (S490), and the repetition process may be terminated. .

As mentioned above, although this invention was demonstrated by the limited embodiment and drawing, this invention is not limited by this, The person of ordinary skill in the art to which this invention belongs, Of course, various modifications and variations are possible within the scope of equivalents of the claims to be described.

As described above, the following effects are created in the wafer automatic alignment method of the present invention.

First, the stability of the alignment is improved and the reliability is improved by using the transformation equation of the matrix and coordinates by recognizing the pattern of the wafer, and the process of alignment is simplified and the image processing calculation amount is reduced, compared to the method using the conventional observation image, The time for sorting can be shortened.

Second, by applying the image and the information of the instrument in conjunction, and by obtaining the image of both ends of the wafer, it can be easily applied to a large size wafer.

Third, by using a formula that excludes the rotation center by approximating the calculation of the rotation center of the wafer, a complicated process of measuring the rotation center is omitted.

Fourth, the reliability of alignment is improved by repeating the wafer until the alignment operation falls within the allowable tolerance, and can be easily applied to the process automation by connecting the front and rear processes with the loader.

Claims (10)

A reference pattern setting step of setting a reference pattern as a reference for automatic alignment by fixing a reference wafer for setting a pattern to a rotating table, and storing the reference pattern in a control computer; A reference position setting step of photographing the reference wafer by a camera and placing the reference pattern at a reference point of the control computer screen to set a reference position of observation; A wafer mounting step of removing the reference wafer from the rotation table and mounting the wafer to the rotation table such that a pattern of the wafer to be aligned is on the observation screen of the control computer; An alignment error measuring step of measuring an alignment error for alignment of the reference pattern and the wafer pattern mounted in the wafer mounting step; Compensation value determination step of determining the compensation value of the pattern by calculating the horizontal transformation matrix (T) and the translational transformation matrix (R) of the pattern according to the alignment error measured in the alignment error measurement step using the coordinate transformation of the matrix ; A compensation value applying step of aligning the wafers by applying the compensation value determined in the compensation value determining step; And And an alignment error checking step of repeatedly performing the alignment error measuring step, the compensation value determining step, and the compensation value applying step so that the alignment error of the wafer aligned in the compensation value applying step satisfies a preset allowable range. Wafer automatic alignment method characterized in that. The method of claim 1, In the reference positioning step, the camera moves in the x direction and the y direction of the wafer to acquire an image of the surface of the reference wafer, and transfers the acquired image to the control computer to obtain the pattern of the reference wafer from the control computer. Wafer automatic alignment method characterized in that the recognition. The method of claim 1, In the reference position setting step, the reference position is a wafer automatic alignment method, characterized in that for setting the reference position by setting two observation points of the first observation point and the second observation point. The method of claim 3, Coordinate transformation of the matrix is a transformation according to the following equation.

P _t (transformed coordinate matrix), T (translational transformation matrix), R (rotational transformation matrix), P _c (transformation coordinate matrix), C (rotational transformation matrix of rotation transformation). The method of claim 4, wherein Wherein the T (translational transformation matrix) and R (rotational transformation matrix) are a translational transformation matrix and a rotational transformation matrix as in the following equation.

Here, a _x (x-axis compensation value), a _y (y-axis compensation value), and a _θ (rotation compensation value). The method of claim 5, Wherein P _t (transformed coordinate matrix), P _c (transformed coordinate matrix), and C (rotational center matrix of rotation transformation) are matrices represented by the following equations.

Where x _t1 (transformed pattern x coordinate of first observation point), x _t2 (transformed pattern x coordinate of second observation point), y _t1 (transformed pattern y coordinate of first observation point), y _t2 (second observation point The transformed pattern of the y coordinate), θ _t1 (the transformed pattern angle of the first observation point), θ _t2 (the transformed pattern angle of the second observation point), x _c1 (converted pattern x coordinate of first observation point), x _c2 (converted pattern x coordinate of second observation point), y _c1 (converted pattern y coordinate of first observation point), y _c2 (conversion of second observation point) Pattern y-coordinate), θ _c1 (the transformed pattern angle of the first observation point), θ _c2 (the transformed pattern angle of the second observation point), C _x (x coordinate of the center point), C _y (y coordinate of the center point). The method of claim 6, The tan value of the θ angle compensation value a _θ and the x compensation value and y compensation value of the n th observation point (n = 1 or n = 2) are calculated by the following equation. .

Here, a _θ (theta angle compensation value), a _xn (the x-axis compensation value at the nth observation point), and a _yn (y-axis compensation value at the nth observation point). (n = 1 or n = 2) The method of claim 7, wherein In the determining of the compensation value, the rotation compensation value a _θ of the wafer is approximated by a tan _θ value a _θ and cos a _θ value 1.0. Characterized in the wafer automatic alignment method. The method of claim 7, wherein For changing the direction of the observation screen of the control computer Define the turn matrix as

Wafer automatic alignment method characterized in that for calculating the compensation value by the coordinate transformation of the following equation.

Where D _v (direction transition matrix) and δ (alignment error). The method of claim 7, wherein In the alignment error checking step, the convergence coefficient η is set by repeatedly performing the alignment error measuring step, the compensation value determining step, and the compensation value applying step, and the convergence speed is adjusted by the following equation.

And the error checking step is repeated until an alignment error (δ) falls within an allowable tolerance (ε). Here, ^{i + 1} P (i + 1 th ordered pattern matrix), ⁱ P (i th ordered pattern matrix), η (convergence coefficient), i (number of alignments), and β (alignment compensation value).

Images