JPH09283581A - Superposition error measuring method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method of determining the measuring range of the edges of measuring marks which is the premise to accurately measure the superposition error of circuit patterns at high reproducibility in the production process of a semiconductor element. SOLUTION: This method comprises taking a microscope image of two different-size square measuring marks superposed on a semiconductor wafer, converting this image into a digital image, processing the image to determine the positions of the etches of the marks and measuring the superposition error of the marks. About each edge of the measuring mark image, max. cross correlation coefficients in the directions x and y in an x-y orthogonal coordinate system are computed (S02) to automatically recognize each edge (S05) and the measuring range for each edge is automatically determined (S07).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of inspecting the overlay accuracy of a circuit pattern formed on a semiconductor wafer for each step in a photolithography process for manufacturing a semiconductor integrated circuit.

[0002]

2. Description of the Related Art In recent years, as semiconductor devices have become highly integrated, it is becoming more and more necessary to accurately superimpose circuit patterns for each process in a photolithography process for manufacturing devices. For that purpose, a measurement mark for measuring the overlay error is provided on the wafer separately from the circuit pattern,
For example, a square measurement mark (hereinafter referred to as a first box mark) formed in the first step and a square measurement mark (hereinafter referred to as a second box mark) formed in a second step that is smaller than the first box mark. It is measuring the overlay error of the microscope images of the two box marks). If the overlay error, that is, the value of the distance between the centers of the two box marks is within the allowable range, the semiconductor wafer is put into the next process as a good product, and if it exceeds the allowable range, it is excluded from the manufacturing process as a defective product. Was.

Conventionally, an overlay measuring machine has been used to measure the overlay error. The microscope image of the vernier mark is observed by an operator's eye, and the overlay error is measured visually by the microscope image. There are a manual method and a method of using a light-receiving element to capture the difference in the intensity of specular reflection light near the edges of two box marks, automatically recognizing the edge position by image processing, and automatically measuring the center-to-center distance. It was In order to determine the edge position, first, determine the measurement range shown by the shaded area in FIG. 2 around each edge,
The exact position of each edge is detected based on the measurement range.

[0004]

In the conventional method described above, in the case of the manual method, it is easy to make a difference in recognizing the scale of the microscopic image of the vernier mark depending on the operator or the appearance of the microscopic image. , The measurements were inaccurate. On the other hand, in the case of the automatic measurement method, the contrast of specular reflection light is particularly weak when the edge is not an acute angle, and it is difficult to accurately recognize the edge even if image processing is performed, so there is a problem in determining an appropriate measurement range. As a result, the measurement of the distance between the centers of the two box marks may be inaccurate. In order to manually determine the measurement range in this automatic measurement method, the operator needs some experience and skill, but the reproducibility of the measurement range is still poor, or the measurement range of the opposite edges of the box mark becomes asymmetric. There was an inconvenience.

Therefore, the present invention is based on the automatic measurement method.
An object of the present invention is to provide a method for recognizing an edge of a box mark image even if the contrast of specular reflection light from the edge is weak and determining a measurement range. As a result, finally, it becomes possible to measure the overlay error accurately and with good reproducibility.

[0006]

In order to achieve the above-mentioned object, the invention according to claim 1 of the present invention is for measuring two quadrangles of different sizes formed by superposing on a semiconductor wafer. Observe the mark with a microscope, take a microscope image of the measurement mark, convert to a digital image, perform image processing, determine the position of each edge of the measurement mark,
In the method of measuring the overlay error of the measurement mark, x is measured for each edge of the image of the measurement mark.
Automatically recognizing each of the edges by calculating the maximum cross-correlation coefficient in the x-direction and the y-direction in y rectangular coordinates
The feature is that the measurement range corresponding to each edge is automatically determined.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 4 is a diagram showing a schematic configuration of a main part of an overlay measuring machine used for overlay error measurement according to an embodiment of the present invention. FIG. 4 shows a measuring stage 3, a microscope 4, an image processing section 5, a system control section 6,
Stage drive control unit 7, focus position detection unit 8 and light source unit 9
Consists of Further, the measuring stage 3 is mounted with the wafer 10 thereon, and the X direction and the Y direction (in the figure, the direction perpendicular to the paper surface).
The XY stage 3a is linearly movable, the Z stage 3b is linearly movable in the Z direction, and the θ stage 3c is rotatable about the Z axis. The image processing unit 5 includes a CCD camera 5a for taking a microscope image of a box mark, an A / D converter 5b for converting the microscope image into a digital image, and an image processing device 5c for processing the digital image.
The image data processed by the image processing device 5c is transferred to the system control unit 6 by a LAN, for example. The stage drive control unit 7 includes an XY drive control unit 7a that controls the movement of the XY stage 3a, a Z drive control unit 7b that controls the movement of the Z stage 3b, and a θ drive control unit 7c that controls the rotation of the θ stage 3c. . The XY drive control unit 7a and the θ drive control unit 7c perform movement control of the XY stage 3a and rotation control of the θ stage 3c, respectively, according to commands from the system control unit 6. The focus position detection unit 8 sends the detected value to the Z drive control unit 7b, and performs movement control of the Z stage 3b in the Z direction, that is, focus adjustment. Further, the light emitted from the light source unit 9 illuminates the wafer 10 through the semi-transmissive mirrors 11 and 12 and the microscope 4, and the reflected light from the wafer 10 sequentially passes through the microscope 4 and the semi-transmissive mirrors 12 and 11 to generate a CCD. An image is taken by the camera 5a. For the light source of the light source unit 9, for example, a metal halide lamp is used.

Next, the procedure of overlay error measurement according to the present invention will be described. The wafer 10 is placed on the measuring stage 3 after the eccentricity amount is calculated and the rotation is corrected by a non-contact pre-alignment unit (not shown). The wafer 10 is moved in the XY directions by the XY stage 3a of the measuring stage 3 so that the box mark is located at the center of the field of view of the microscope 4, and is moved in the Z direction for focusing. The wafer 10 is subjected to so-called global alignment using two pre-registered box marks on the wafer, and precise rotation correction of the wafer 10 is performed by the θ stage 3c. As a result, the edge of the box mark becomes parallel to the X and Y directions of the XY stage 3a.

The microscope image of the box mark is photographed by the CCD camera 5a, digitized by the A / D converter 5b, and then stored in the storage medium of the image processing apparatus 5c. This image data is sent from the image processing device 5c to the system control unit 6
Then, the arithmetic processing is performed according to the sequence described below. FIG. 1 is a flow chart showing a procedure for automatically determining a measurement range in overlay error measurement according to an embodiment of the present invention.

In S01, the digitized images of the box marks 1 and 2 shown in FIG. 2 are read from the system controller 6. Positioning is performed such that the center of the box mark 1 is the origin of the xy coordinates, and the x and y axes are parallel to the X and Y directions of the XY stage 3a. S0
2, the cross-correlation coefficient M (τ) between the line profile of x = 0 shown in FIG. 4 and one line profile of x ≠ 0 is calculated by the following equation. The following formula is x = 0
The vector can be written as follows, where R is the vector of x and S is the vector of x ≠ 0. The subscript τ is the amount of deviation of x ≠ 0 with respect to x = 0.

[0011]

[Equation 1]

For one line of x ≠ 0, the maximum value of the cross-correlation coefficients is set as the maximum cross-correlation coefficient. However, when the cross-correlation coefficient M (τ) <0, all are regarded as M (τ) = 0. The maximum cross-correlation coefficient is calculated for each line where x ≠ 0, and the maximum cross-correlation coefficient value is plotted for each x line. In S03, a histogram of maximum cross-correlation coefficient values is created. In general, an image contains noise, and thus this process is performed to remove noise.

In S04, a level at which the maximum cross-correlation coefficient can be treated as the same is taken out from the above histogram, and the graph shown in FIG. 3 is obtained. FIG. 3 shows the normalized maximum cross-correlation coefficient for each x-line. That is, when the graph is traced from x = 0 in the x direction, the maximum cross-correlation coefficient takes the same high level while being inside both of the images of the box marks 1 and 2 shown in FIG. The maximum cross-correlation coefficient drastically changes at the edge of the mark 2, and when it reaches the outside of the box mark 2, the maximum cross-correlation coefficient becomes lower and takes the same level. Furthermore, the maximum cross-correlation coefficient rapidly changes to zero at the edge of the box mark 1, and the level of zero is maintained even outside the box mark 1.

In S05, the edge of the box mark is recognized from the graph of FIG. That is, the portion where the maximum cross-correlation coefficient changes rapidly corresponds to the edge. S0
In FIG. 6, a pair of edges existing at positions substantially symmetrical with respect to the center of the graph of FIG. 3, that is, x = 0 is recognized, and the edge position is determined. That is, the values corresponding to the edges 1a and 1b of the first box mark 1 shown in FIG.
The values corresponding to 2b are A2 and B2, respectively.

In S07, the measurement range is automatically determined for each edge. This measurement range is determined to have an appropriate width depending on the steepness and linearity of the edge and the conditions around the edge, for example. The wider the measurement range, the better the reproducibility of the edge position, but the noise increases probabilistically, so it is necessary to select an appropriate width. Above S01 to S
Each operation up to 07 is similarly performed for the y line,
Finally, as shown in FIG. 2, box marks 1 and 2
The measurement range (indicated by the shaded area in FIG. 2) corresponding to all the edges is determined.

After that, the positions of all the edges of the box marks 1 and 2 are accurately detected based on the above-mentioned measurement range,
Thereby, the overlay error amount is measured in the following procedure. First, the edge positions of the two box marks are measured, then the midpoint positions of the two box marks in the x direction are calculated, and the distance between the midpoints is the overlay error amount Δx in the x direction. Similarly, also in the y direction, the overlay error amount Δy is obtained. Therefore, the overlay error amount of the two box marks can be obtained from Δx and Δy.

[0017]

As described above, according to the present invention, the edge of the box mark provided for measuring the overlay error amount can be accurately and automatically recognized, and the measurement range for each edge can be automatically determined. Even if the edge profile is not clear, a proper measurement range can be obtained. In addition, the operator does not need to be skilled. As a result, the overlay error amount Δx in the x direction, the overlay error amount Δy in the y direction, and the overlay error amount of two box marks can be measured with high accuracy and reproducibility.

Further, even when the edge of the box mark and the moving direction of the XY stage do not coincide with each other to some extent, the edge position is recognized by the correlation calculation, so that an appropriate measurement range can be determined.

[Brief description of drawings]

FIG. 1 is a flowchart showing a procedure for determining a measurement range according to an embodiment of the present invention.

FIG. 2 is a conceptual diagram showing two box marks and a measurement range according to the embodiment of the present invention.

FIG. 3 is a graph showing changes in the normalized maximum cross-correlation coefficient in the x-line direction according to the embodiment of the present invention.

FIG. 4 is a schematic configuration diagram of a main part of an overlay measuring machine for overlay error measurement according to an embodiment of the present invention.

[Explanation of symbols]

 1 ... Box mark made in the first step 2 ... Box mark made in the second step 3 ... Measuring stage 4 ... Microscope 5 ... Image processing unit 5a ...・ CCD camera 5b ・ ・ ・ A / D converter 5c ・ ・ ・ Image processing device 6 ・ ・ ・ System control unit 7 ・ ・ ・ Stage drive control unit 8 ・ ・ ・ Focus position detection unit 9 ・ ・ ・ Light source unit 10 ・..Wafer

Claims (1)

[Claims] 1. A microscope is used to observe two square measurement marks of different sizes formed on a semiconductor wafer by superimposing, and a microscope image of the measurement mark is photographed and converted into a digital image. A method of performing processing to determine the position of each edge of the measurement mark and measuring the overlay error of the measurement mark, wherein for each edge of the image of the measurement mark, the x direction and the y direction in xy orthogonal coordinates A method for measuring an overlay error, which comprises automatically recognizing each edge by calculating a maximum cross-correlation coefficient in a direction, and automatically determining a measurement range corresponding to each edge.