KR20090038531A - Method of auto-measuring defect in wafer - Google Patents

Abstract

A method for automatically checking a wafer defect is provided to perform a wafer defect check regardless of proficiency of an operator by automatically performing various defect checks. A cleavage surface of a vertical sample for measuring is parallel positioned with a lens of a microscope. A top and bottom moving region between a highest position(600) and a lowest position(605) is set in order to automatically focus the surface of the sample. The cleavage surface of the sample is imaged in an initial position(601). An image is obtained through a CCD camera. A resolution of the image is automatically analyzed. A resolution of the image measured in the lowest position is automatically analyzed. The image resolution of the initial position is automatically compared with the image resolution of the highest position and the lowest position. A lens of the microscope is moved to a position having a higher image resolution.

Description

How to automatically inspect wafer defects {Method of auto-measuring defect in wafer}

The present invention is a method for inspecting defects of a semiconductor wafer such as a silicon wafer, and more particularly, for evaluating the location, distribution, density, etc. of crystal defects present in the surface and bulk of the wafer. It is about a method.

As the degree of integration of semiconductor devices increases, the yield and reliability of semiconductor devices are greatly affected by the quality of the wafer on which the semiconductor devices are implemented. The quality of the wafer depends on how many defects occur throughout the so-called wafering process of fabricating the wafer during crystal growth.

These wafer defects can be divided into crystal defects occurring during ingot growth and defects caused by external contaminants. Among them, external contaminants such as dust are easily removed by an etching or cleaning process, but Crystal Originated Particles (COP), Flow Pattern Defect (FPD), Oxygen induced Stacking Fault (OiSF), Bulk Micro Defect (BMD), and Large DP (LDP) Crystal defects such as dislocation pit are not removed by a cleaning process or the like and affect the yield and quality of semiconductor devices. Therefore, the defects must be suppressed during the manufacturing process of the wafer. Therefore, it is very important in terms of yield management to verify and inspect the precise distribution, density, and the like of these defects before the semiconductor device is implemented on the wafer.

For example, the depth of the Denuded Zone (DZ) of oxygen-related process-induced defects typically formed after heat treatment using a wafer is within the bulk from the wafer surface in the backside direction. Depth to existing defects, BMD means dislocations, stacking defects, precipitates, etc. existing inside the wafer, typically after the heat treatment, the wafer is cleaved to {110} plane and selectively etched, at a magnification of 100 times or more It is a defect that is measured when the cleaved surface is observed. Therefore, the above defects are characterized in that they cannot be observed on the wafer surface. Therefore, to measure DZ depth and BMD density, prepare a vertical sample cut from the wafer, mount the cleaved surface parallel to the microscope lens, and then place the lens of the microscope at the point of the sample cross section and adjust the magnification. The DZ depth on the screen is measured and the density is measured by determining the number of BMDs.

As described above, the conventional method for measuring the DZ depth and the BMD density has many problems as follows because the operator observes the sample by using a microscope without any special equipment.

First, since the magnification of the microscope can be measured by varying the magnification from 50 times to sometimes 500 times higher, the DZ depth may be different if the magnification of the microscope and the conversion according to the magnification do not match.

Secondly, DZ depth or BMD density is measured at regular intervals moving from the center of the wafer to the edge, and then the final calculation results are represented using graphs and images. It is difficult to measure at a more precise interval since the movement interval is limited.

Third, the narrower the measurement interval, the longer the measurement time. In particular, since the work is performed by a person, the fatigue of the worker increases as time passes and the productivity also decreases. Since the work is performed continuously while looking at the same screen, the eye fatigue of the operator is increased, and when the DZ depth of many samples is measured, the reliability of the result may be deteriorated over time.

Fourth, in the case of 300 mm wafer, the thickness of the cross section is about 750 μm. When measuring the BMD density in this vertical direction, the microscope stage cannot be measured at precise intervals due to the limitation of movement.

Fifth, when repeated measurements are performed on the same point using the same sample, it is difficult to realize the reproducibility of the result because the measurement cannot be performed at the same point as the initial measurement point even if the fine adjustment is performed.

Sixth, in addition to the depth from the surface to the first defect, there are also cases where the depth to the second or third defect needs to be measured, which requires more analysis time because the work is done by humans, and many defects have a constant depth. If you're flocking to, it's hard to get accurate results.

On the other hand, grown-in defects such as OiSF and LDP that occur during ingot growth are defects present on the wafer surface. Since these defects are so small to be observed with the naked eye, a microscope is used to measure the distribution and density on the wafer surface at a certain magnification or higher.

These defects do not have any special equipment manufactured to measure them, and they are often evaluated using a general-purpose microscope. In order to more easily observe, in order to obtain a density in a specific direction in addition to the distribution of defects, a special die with a certain width is placed on the sample and the density is measured by using a microscope thereon.

However, as with DZ depth and BMD density measurements, these defects are also difficult to measure precisely because they are measured by a microscope, and as time passes, the fatigue of the operator increases, resulting in poor productivity and reliability of the results. I have a problem.

The present invention is to solve the above problems, the problem to be solved by the present invention is to provide a wafer defect inspection method that can automatically inspect various crystal defects of the wafer.

In order to solve the above problems, the wafer defect inspection method proposed by the present invention is a wafer defect inspection method of a microscope auto focus adjustment method, and automatically applies a dichotomy to a final focus position. That is, (a) setting the up and down moving area between the highest position and the lowest position of the microscope lens; (B) taking a sample at an initial position of the microscope lens to obtain an initial position image; (C) obtaining the highest position image and the lowest position image after the microscope lens is moved from the initial position to the highest position and the lowest position, respectively; (D) automatically comparing the initial position image clarity with the highest position image clarity and the lowest position image clarity, and applying the dichotomy to a distance where the sharpness is high; (E) photographing the sample at the moved position to obtain a moving position image; (E) automatically comparing the moving position image sharpness with the image sharpness before moving, and applying the dichotomy to a distance where the sharpness is high; And (g) repeatedly performing the steps (e) and (bar) sequentially to obtain a position where the focal point is best focused.

The present invention also proposes two inspection methods using such an auto focusing method, which includes a microscope for observing a sample surface, a sample support for mounting the sample, and operation of the microscope and a sample support. The wafer defect inspection apparatus includes a computer for controlling the monitor and a monitor for outputting the photographed image of the microscope and the processing screen of the computer.

The first method is to measure defects in wafer bulk such as DZ depth, BMD density, etc., comprising: (a) performing defect auto-measurement after microscopic autofocusing at the first measurement point of the sample; (b) automatically storing and sorting the measurement result value and the measurement image; (c) moving the sample at predetermined intervals to perform autodetection of defects after microscopy autofocus; (d) automatically storing and sorting the measurement result value and the measurement image; And (e) sequentially repeating steps (c) and (d) until the measurement at the last measurement point is completed.

The second method is capable of measuring defects on the wafer surface such as OiSF density, LDP density, etc., comprising: (a) performing automatic measurement of defects along the radial direction after microscope autofocusing at the first measurement point of the sample; (b) automatically storing and sorting the measurement result value and the measurement image; (c) rotating the sample by a predetermined angle to perform automatic defect measurement along the radial direction; And (d) repeating step (c) until the measurement at the last measurement point is finished.

Through the wafer defect inspection method according to the present invention, the DZ depth, BMD density measurement, OiSF and LDP density measurement, etc., performed by humans are automatically performed, so that not only quantitative arithmetic results but also qualitative calculation results including images can be accurately obtained. You can get it. Since various defect inspections are automatically performed, there is an advantage in that accurate wafer defect inspection is always possible regardless of the skill of the operator.

In the case of defect measurement in wafer bulk such as DZ depth and BMD density measurement, in the present invention, the microscope and the cross section of the wafer are made horizontal and then the automatic measurement is performed, and the microscope automatically focuses with the surface of the sample to focus. The measurement can then be made. In addition to the depth from the wafer surface to the first defect, if the second or third defect is specified by the operator prior to the first automatic measurement, the measurement can be made automatically in a short time, even when many defects are concentrated at a constant depth. Accurate results can be obtained.

When moving at regular intervals during automatic measurement according to the invention, the image of the measurement position obtained and the arithmetic quantitative result are automatically aligned and stored in the form of a file. The image obtained at the measurement position is likely to be mixed when measuring short intervals because a large amount of images may come out, but in this method, when the first measurement, only the sample information is input, the information and cross section of the measured sample are displayed on the upper left of the image. The location of the measurement in the sample, and the DZ depth and BMD density are automatically generated and displayed for easy organization of results and convenient use of data.

In addition, since the measurement interval can be made precisely from tens to hundreds of microns, a precise profile (radial profile) can be obtained. Observations can be made on a consistent basis, producing the same results regardless of who is measuring. That is, since the size of the defect selected for measuring the DZ depth can be arbitrarily adjusted through the recipe selection before the automatic measurement, the same result can be confirmed when the measurement is performed using the same recipe. In addition, the measurement area can be arbitrarily designated on the screen seen through the microscope.

In the case of defect measurement on the wafer surface, such as OiSF and LDP density measurement, the microscope automatically focuses on the surface of the sample, automatically recognizes the OiSF and LDP, and counts the number automatically. By rotating the sample at an angle within 1 °, a line profile can be obtained at various angles within a range of -150 mm to +150 mm (300 mm wafers), and measurements can be made at various angles in one direction. In addition to the results, the results can be obtained in various directions. Since the counting area and the magnification of the microscope can be arbitrarily adjusted, the result can be derived in various forms.

Hereinafter, exemplary embodiments of the present invention will be described with reference to the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, only this embodiment is to complete the disclosure of the present invention, those skilled in the art to which the present invention belongs It is provided to fully inform the scope of the invention, and the invention is defined only by the scope of the claims. Like reference numerals in the drawings denote like elements.

The wafer defect inspection method according to the present invention is a wafer defect inspection apparatus having a microscope for observing a sample surface (a wafer surface as well as a cleaved surface of a vertical sample) and a sample support for mounting a sample. It can be implemented using any possible form of wafer defect inspection apparatus that can be controlled by the. In particular, the present invention may be performed using a wafer defect inspection apparatus as disclosed in Korean Application No. 2007-0099078, and the following description will be given by way of example for carrying out the present invention using the wafer defect inspection apparatus disclosed in the above application. Those skilled in the art will recognize that they are not limited to the use of such devices.

1 is a schematic configuration diagram of a wafer defect inspection apparatus 100 that may be used in the practice of the present invention, and FIG. 2 is a schematic side view of the wafer defect inspection apparatus 100 shown in FIG. 1 and 2, respectively.

1 and 2, the wafer defect inspection apparatus 100 includes a pedestal 10, a first stage 20, a second stage 30, a rotating stage 40, a microscope 50, and a loading plate ( 60).

The pedestal 10 is a member on which the first stage 20, the second stage 30, the rotating stage 40, and the microscope 50 are installed. The pedestal 10 has a flat upper plate portion 11, and four legs 12 are coupled to the lower side of the upper plate portion 11 to support the upper plate portion 11. In addition, a lower side of the upper plate 11 is provided with a keyboard support 13 so that a keyboard (not shown) electrically connected to a computer (not shown) to control computer operation can be placed. The keyboard support 13 is slidably installed on the guide rails 14 provided on both lower sides of the upper plate portion 11, thereby allowing the keyboard (not shown) to be pulled out or pushed into the bulk.

The first stage 20 is installed on the pedestal 10. That is, the first guide rails 21 are slidably coupled to the pair of first guide rails 21 installed side by side in the X-axis direction on the pedestal 10. Accordingly, the first stage 20 can be reciprocated along the X axis direction. The first stage 20 moves by receiving a driving force from a linear motor (not shown), and can precisely control the movement amount by adjusting the linear motor.

The second stage 30 is installed on the first stage 20. That is, the first stage 20 is slidably coupled to a pair of second guide rails 31 installed parallel to each other along the Y axis direction perpendicular to the X axis. Accordingly, the second stage 30 can be reciprocated along the Y axis direction. The second stage 30 also moves by receiving a driving force from a linear motor (not shown), and can precisely control the amount of movement in the Y-axis direction by adjusting the linear motor. On the other hand, the receiving portion 32 is formed in the upper portion of the second stage 30 to be concave to accommodate the rotary stage 40.

The rotating stage 40 is formed in a substantially disk shape and is disposed in the receiving portion 32 of the second stage 30. That is, the rotating stage 40 is connected to the step motor 41 capable of precisely controlling the amount of rotation and is capable of forward and reverse rotation in the accommodating part 32 by the operation of the step motor 41.

An annular scale member 70 is attached to the outer side of the receiving part of the second stage 30, that is, the outer side of the loading plate 60. The scale member 70 is disposed surrounding the loading plate 60, and an angle scale of 0 ° to 360 ° is displayed on the upper surface thereof. Accordingly, when the loading plate 60 having the reference point is rotated together by the rotating stage 40, the degree of rotation can be visually confirmed by looking at the scale attached to the scale member 70.

The loading plate 60 is for supporting and fixing a sample to be inspected, and is fixed on the rotating stage 40 and rotated together with the rotating stage 40.

The microscope 50 is for magnifying and observing a defect of a sample, and is disposed spaced upward from the loading plate 60. That is, the pedestal 10 is provided with a vertical support 18, the microscope 40 is installed to be elevated in the vertical support 18 along the Z-axis direction. In addition, the microscope 50 has a built-in CCD camera (not shown) for acquiring a defect enlarged by the lens as an image.

In addition, the wafer defect inspection apparatus 100 includes a computer and a monitor (not shown). The computer is connected to the keyboard and the monitor, respectively. In addition, the computer is electrically connected to the first stage 20, the second stage 30, the rotating stage 40 and the microscope 50, respectively, to control their operation, in particular the wafer defect according to the present invention. The inspection method is automatically performed by being stored in the computer readable recording medium and being driven by the computer. That is, by mounting a recording medium in which the wafer defect inspection method according to the present invention is stored, the computer controls the degree of movement and rotation of the stages 20, 30, and 40. Perform image storage and processing acquired from the CCD camera built in 50). The monitor displays the requirements for inspecting the wafer, such as the wafer's X, Y-axis movement, rotation, images obtained from the CCD camera, positional coordinates of the defect, density, and the like.

As mentioned above, defects in the wafer are observed in two areas: the surface area of the wafer and the bulk area of the wafer. When observing the surface defects of the wafer, the wafer is placed on the loading plate 60 as it is and the surface is inspected by the microscope 50. When measuring defects observed in the bulk region, the vertical sample is described as described below. Prepare and place on the loading plate 60 and the cleaved surface is inspected with a microscope (50).

3 is a view for showing a vertical sample preparation method.

First, the wafer 200 subjected to the heat treatment is prepared, the surface of the wafer 200 is turned upward, and the wafer 200 is cut in the vertical direction 201. When the wafer 200 is further cut in the vertical direction 201a with a predetermined interval 202 based on the cut surface, that is, the cleaved surface 210, the length 212, the thickness 211, and the width 202 are used. Measuring piece sample 213 is made with. The thickness 211 of the made sample 213 is about 750 um based on a 300 mm wafer.

Figure 4 is a flow chart of the automated inspection method according to the present invention, particularly relates to the measurement of DZ depth and BMD density corresponding to the measurement in bulk.

In the measurement process shown in FIG. 4, the operator places the sample 213 manufactured by the above method on the loading plate 60 of the vertical sample support or the wafer defect inspection apparatus (100 of FIGS. 1 and 2) (S100). After that, the step of adjusting the focus at the first measurement point and the final measurement point (s110 to s150) should be performed.

First, the focus of the microscope 50 is manually adjusted at the first measurement point of the sample (s110), and then the loading plate 60 is moved to the last measurement position of the sample (s120). At this time, the first stage 20 and the second stage 30 are used. Next, focusing of the microscope 50 at the last measurement point of the sample is manually performed (s130). Based on the focus image on the computer (monitor) screen, the horizontality of the sample and the shaking of the left, right, front and back are adjusted (S140). When the final check of the sample state is performed (s150), if the level and the shaking degree satisfy predetermined conditions (Yes), the process proceeds to the next inspection step, step s160, otherwise (No). Returning to step s110 or s120 to repeat the above process.

Since the sample manufactured for the purpose of observing defects in the bulk is cut after the wafer and the cleaved surface is observed, the horizontality of the sample itself may be different for every sample. Of course, it is not necessary to focus at every measurement position in the measurement process, and even if there is a slight step, it may be covered by the auto focus adjustment function described later. In the above steps, it is only necessary to match the sample level with the left, right, front and back positions.

Next, after the final confirmation of the sample state (s150), the DZ depth automatic measurement is performed after the auto focus adjustment to be described later at the initial measurement point (s160). Before that, the operator may have a process of inputting sample information. Next, the DZ depth result and the measured image are automatically stored and aligned (s170). When the operator inputs the sample information, the measurement position of the sample is automatically generated during the movement, and the DZ depth value at that time may be automatically written. Then, the loading plate 60 is moved by moving the first stage 20 or the second stage 30 by a predetermined interval (s180). After the auto focus is adjusted again at the moved position, that is, the next measurement point, the DZ depth automatic measurement is performed (s190). Thereafter, as in the previous step S170, the DZ depth result and the measured image are automatically stored and aligned (s200). If it is determined that the measurement is finished until the measurement at the last measurement point (s210), and if the measurement is not completed until the measurement at the last measurement point (“NO”), the process returns to step s180 to move to the next measurement point to measure the process. If it is determined that the measurement is finished until the measurement at the last measurement point (“YES”), the DZ depth measurement is terminated, and in this embodiment, it is automatically moved back to the first measurement point for the next BMD density measurement (s220). ).

Then, a recipe selection for measuring the BMD density (s230) is performed, which corresponds to inputting defect information on which image is to be determined as a defect. Subsequently, after adjusting the auto focus at the first measurement point, this time, the BMD density is automatically measured (s240). Just as the DZ depth result and the measured image are automatically stored, the BMD density and the measured image are automatically stored and aligned (s250). Then, the loading plate 60 is moved by a predetermined interval (s260). The BMD density measurement is performed after the auto focus is adjusted again at the moved position, that is, the next measurement point (s270). Thereafter, as in the previous step S250, the BMD density and the measured image are automatically stored and aligned (s280). If it is determined that the measurement is finished until the measurement at the last measurement point (s290), otherwise, the process returns to step 260 to move to the next measurement point, and the measurement is repeated. (S300).

The actual DZ depth and BMD density values measured by the above-mentioned method will be described first with reference to the drawings.

FIG. 5A graphically illustrates the arithmetic quantitative results of measuring DZ depths at intervals of 10 mm from the center of the wafer to the edge direction using a 300 mm wafer.

“1P DZ-Test 1” in circles and solid lines in FIG. 5A is a result of measuring the depth from the surface to the first defect existing in the bulk direction, and “1P DZ-Test 1-Repeat” in triangles and dotted lines. Indicates the results of repeated measurements using the same sample. As a result of observing through this graph, it is possible to confirm the identity of the result according to the repeated measurement.

FIG. 5B is a photograph measured at a center point (center, 0 mm) measured initially among the results of FIG. 5A.

It can be seen that the depth 300 from the surface to the first (first) defect is correctly recognized on the screen shown in FIG. 5B. In addition, the information "1P-0 mm / DZ-9.9 um" shown in the upper part of FIG. 5B means the sample information-measurement position / DZ depth value, and the sample information "1P" ( If only 310 is input, the measurement position 311 of the next sample is automatically generated during the movement, and the DZ depth value 312 at that time is automatically filled in, so that the result can be easily checked even with the image alone.

FIG. 5C shows the image results of the DZ depths obtained by repeating this process by moving 10 mm intervals from the center of the wafer to the edge direction (312, 313). In fact, except for the sample loading time, it took about 6 seconds to move the measurement 10 points apart after measuring one point.

The time is measured one point and moved by a predetermined interval (10mm in the above conditions), and then automatically focus the image on the screen using the automatic focusing function described later, calculates the result value (DZ depth), the image This is the time taken to fill in both the sample information and the measured value information at the top of.

Therefore, even if the analysis is performed by a skilled worker, if the measurement interval is more precisely progressed, the time required for analysis can be greatly reduced. The result indicates an evaluation result of designating the entire image (FIG. 5B) displayed on the screen as one measurement area, and a smaller area may be designated as one measurement area therein.

Next, FIGS. 6A to 6C show the measured values after defining the depth from the surface of the wafer to the third defect using the same sample proceeded from FIGS. 5A to 5C as DZ depth.

FIG. 6A is a graph illustrating arithmetic quantitative results of measuring DZ depths at intervals of 10 mm from the center to the edge direction using a 300 mm wafer as in FIG. 5A. In FIG. 6A, the “3P DZ-Test 1” represented by circles and solid lines is a result of measuring the depth from the surface to the third defect existing in the bulk direction. The results of repeated measurements using the same sample are indicated. Depth results up to the third defect shown in the graph can be confirmed that the same is obtained even during repeated measurements.

6B also shows a photograph measured at the center point (0 mm) measured initially among the results of FIG. 6A. Sample information 410-measurement position 411 / DZ depth value 412 can be seen from the information " 3P-0 mm / DZ-17.2 um " appearing at the top of FIG. 6B. It can be seen that the depth to the third defect 320c, not to the second defect 320b, is regarded as the DZ depth 400. This result can be confirmed in the same manner in other measurement points of FIG. 6C.

In the above results, the measurement of the depth 400 up to the third defect 320c also took 6 seconds for one point measurement except for the sample loading time. Can be defined by the original operator.

7A to 7E show the results of measuring the BMD density.

First, FIG. 7A is a graph illustrating arithmetic quantitative results of measuring BMD density in a wafer at intervals of 10 mm from the center to the edge direction using a 300 mm wafer. In FIG. 7A, the graph “BMD test 1” represented by circles and solid lines represents an initial measurement result, and the graph “BMD test 1 -repeats” represented by triangles and dotted lines represents the results of repeated measurements using the same sample. From the results, it can be seen that the two values coincide within a small error range.

Next, FIG. 7B shows the image result at the center point in the measurement result of FIG. 7A. As described with reference to FIG. 5B, as in the result of measuring the DZ depth, when the operator inputs only the first "BMD" sample information, the BMD measurement position 500 and the measurement result 501 are automatically displayed on the upper left of the image. Will be written. In addition, an area 502 on the screen for measuring the BMD density may be arbitrarily designated. This result is a result of applying the same to other positions after designating a certain region 502.

FIG. 7C lists the image results of measuring the result of FIG. 7B from the initial measurement point (0 mm) to the final measurement point (140 mm).

7D illustrates a result of measuring the BMD density in the thickness 211 direction using the sample 213.

7A to 7C show the measurement results in the edge direction (0 mm to 140 mm) at the center of the sample 213, while FIG. 7D shows the surface of the wafer 200 at one point (for example, the center). This is the result of measuring the BMD density in the back direction. As a result, about 770 um thickness 211 of the sample 213 was divided into 15 equal parts to measure the density in each measurement area 503.

In FIG. 7D, the graph “1st V_BMD test” represented by circles and solid lines represents an initial measurement result, and the graph “1st V_BMD test _ repeated” represented by triangles and dotted lines represents a result of repeated measurements using the same sample. From the results, it can be seen that the two values coincide within a small error range, and it can be seen that the BMD density coincides with the front image of the sample 213 at a total of 15 equal points.

In particular, this function has the advantage that the measurement area can be arbitrarily adjusted according to the operator's reference for a thin sample. For example, the results show that the density is divided into 15 equal parts, but the densities can be assessed by dividing the area more closely, or by wider intervals.

7E shows the evaluation results at different points using the same sample. The graph “2nd V-BMD test” shown in circles and solid lines in FIG. 7E represents the initial measurement results, and the graph “2nd V-BMD test-repeats” represented by triangles and dotted lines shows the results of repeated measurements using the same sample. . As in FIG. 7D, the accuracy of the result according to the repeated measurement can be confirmed.

Next, the method and principle of automatically measuring the DZ depth and the BMD density in the present invention will be described in detail.

First, FIG. 8 illustrates a principle in which the lens of the microscope 50 automatically focuses on the cleaved surface 210 of the sample 213.

First, the cleaved surface 210 (cross section of the wafer) of the measurement vertical sample 213 is mounted in parallel with the lens of the microscope 50, and then the lens of the microscope 50 moves in the vertical direction (Z axis). In order to automatically focus the surface of the sample 213, the up and down moving area between the highest position 600 and the lowest position 605 is first set. Subsequently, when the lens of the microscope 50 photographs the cleaved surface 210 of the sample 213 at the initial position 601, this image is acquired as an image by a CCD camera, and the image of the acquired image is its sharpness. Is analyzed automatically.

When the image sharpness analysis at the initial position 601 is completed, the lens of the microscope 50 moves from the initial position 601 to the highest position 600 and the lowest position 605 of the preset Z axis, respectively, and acquires the measured images, respectively. Again, this image is acquired as an image through a CCD camera in the same manner as the above process, and then the sharpness is automatically analyzed.

After automatically comparing the initial position 601 image sharpness with the image sharpness of the highest position 600 and the lowest position 605, the lens of the microscope 50 moves by applying a dichotomy to the distance where the sharpness is high. First dichotomy, which is half of the distance from the initial position 601 to the lowest position 605 under the assumption that the image clarity of the lowest position 605 is better based on the image clarity of the initial position 601 in FIG. 8. The first moving position 604 moved through is shown.

Next, the image sharpness obtained at the first moving position 604 is automatically compared with the image sharpness obtained at the initial position 601, which is the previous position, and the dichotomy is applied to a place where the sharpness is higher as described above. Will move. In FIG. 8, an initial position 601 from the first moving position 604 is assumed to have a higher image sharpness of the initial position 601 between the image at the initial position 601 and the image sharpness at the first moving position 604. The second moving position 602 is transferred through the second dichotomy, which is half of the distance to).

The image sharpness obtained at the second moving position 602 is automatically compared with the image sharpness obtained at the first moving position 604, which is the previous position. It moves to the movement position 603.

Repeating the above process, that is, automatically comparing the moving position image sharpness with the image sharpness before moving, repeating the step of moving the microscope 50 lens by dividing the distance to the place where the sharpness is high, and finally focusing best. The auto focus adjustment is performed at the fit position (Z axis), and FIG. 8 illustrates the third moving position 603 as the autofocusing position.

As such, when the microscope 50 automatically repeats the series of processes described above under the control of the computer, the position (Z axis) having the highest sharpness can be found by the principle of dividing the distance. The preset highest position 600 and the lowest position 605 are fairly short distances, and the process of FIG. 8 shows that the time required for adjustment from the initial position 601 to the final autofocus position 603 is very short, about 3 seconds. It was done in time.

Next, the principle of recognizing a defect through an image displayed on the screen will be described.

First, when the image (automatically taken from FIG. 5B) obtained by the lens of the microscope 50 is analyzed by the histogram according to the intensity of the contrast, the background of the sample 213 having the highest contrast ratio within the area visible on the screen (FIG. 5B) 360) and the actual defect (350 of FIG. 5B) can be separated. In addition, since the minimum and maximum values for the defect size are set in advance, the image is obtained by adjusting the auto focus, and when the maximum length of the defect candidate shape is within the set size range, it is recognized as a defect and detected. At this time, the minimum value and the maximum value of the defect size can be changed.

The principle of automatically recognizing the distance from the surface to the defect as DZ depth is as follows.

Since the upper black region 330 of FIG. 6B is not the region represented by the actual sample 213, the black region 330 is black. A detailed description of this is shown in FIG. 9.

When the sample support 701 is equipped with a sample 704 having a constant length 700 and a width 703, the lens of the actual microscope 702 is a sample 704 when observed at a constant magnification using the microscope 702. The area outside the width 703 of the sample 704 in focus with the surface of) is in black because it cannot be focused. Therefore, in the image seen on the screen, the depth from the surface to the defects present in the bulk direction is the initial starting point of the cross section of the contrast difference value displayed through the histogram, which is changed from dark value to bright value (340 in FIGS. 5B and 6B). Recognition, the initial starting point and the location of the first defect can be recognized as the DZ depth and can be expressed as a quantitative value.

Next, how to automatically recognize the depth up to the first defect (1P-DZ depth) and the third up to the third defect (3P-DZ depth) on the screen will be described. The description below describes only the first and the third with reference to the experimental results, but in practice it can be measured by designating defects in various positions such as the second, fourth, and fifth.

For example, if you define DZ depth as the depth to the third defect and then measure it automatically, based on the principles described above, the depth from the surface to the first defect, the depth to the second defect, and the depth to the third defect are Will be measured. Since the distance to each defect measured in the above process is automatically stored and calculated, the rank of the value is checked by itself so that the depth to the designated defect can be measured automatically. That is, the depth from the surface to the BMD initially found in the bulk direction can be detected as the depth of the defect, or the depth to each BMD found sequentially after the first time can be detected as the depth of the defect.

The following describes the principle of measuring BMD density.

The BMD density should be separated from the background where the actual BMD candidate is present and the background where it is not in the image appearing on the screen in the same manner as the principle used in the DZ. Therefore, after analyzing the screen obtained through the CCD camera in the same manner as above, using the contrast ratio, the background of the highest contrast ratio and the actual BMD candidate can be separated. Actually, the part expressed in BMD is possible because it appears in black unlike the surroundings. After separating BMD candidates in the screen in this way, if the minimum and maximum values of the preset BMD size are used, if the maximum length of the detected BMD candidate shape in the region is within the minimum and maximum range, the BMD candidate is recognized as BMD. The count can be counted each time to determine the number. In the case of BMD density measurement, the measurement area (502 in FIG. 7B) and the number of BMDs are displayed on the screen in terms of density.

10 is a flowchart of another automatic inspection method according to the present invention, and particularly relates to OiSF density and LDP density measurement corresponding to surface measurement. The two defects do not exist in the same wafer because the pretreatment proceeds in advance for defect observation. However, only the algorithm for recognizing defects is different, and the measuring methods are identical. Therefore, the following measuring procedures can be applied in the same way.

In the wafer defect inspection apparatus 100 as shown in FIGS. 1 and 2, the rotation stage 40 may determine the rotation angle within 1 °, and does not proceed unless included in the recipe selection. Because OiSF and LDP are defects that exist on the wafer surface, unlike DZ depths or BMD densities, the autofocus function does not need to be redone every measurement point (area) if the initial autofocus is performed. In this device, the autofocus can be set to be performed every few steps. Therefore, the autofocus function does not need to be performed in every step to save time unless there is a special case.

First, to complete the sample for evaluation after the heat treatment or etching process (s800). An evaluation sample is placed on the sample support (the loading plate 60 in the apparatus shown in Figures 1 and 2) (s810). Next, the measurement magnification and the measurement position are selected (s820), and measurement is started along the radial direction (s830). At this time, the sample is moved by moving the first stage 20 or the second stage 30 along the radial direction. When the measurement in the radial direction is completed (s840), it is determined whether or not the sample support is rotated, and if it is to be rotated (Yes, that is, not to the final measuring point), it is rotated by a predetermined angle (s860) to measure the next. Returning to the start step (s830), the above process is repeated, and if there is no need to rotate (No), the defect evaluation is completed (s870).

11 is a view related to the rotation of the sample support mentioned in the above process.

First, the sample to be measured is mounted on a sample support (loading plate 60 in the apparatus shown in Figs. 1 and 2). In the sample mounting, the position where the notch 230 of the wafer can be mounted is prepared in the loading plate 60, and when the reference to the user of the equipment, the notch 230 is placed in front of the viewer (240). In this state, if the density of defects is found in the radial direction of the first sample, the density in one direction may be obtained (240a). If the first operator defined the rotation angle as 45 °, the rotation stage 40 rotates by 45 ° after completion of the first measurement 240a (250). After the rotation, the secondary measurement result 250a can be obtained by obtaining the radial density in the same manner. Again, the rotation stage 40 is rotated 45 °, it is possible to obtain a third measurement result (260a).

This method has the advantage of allowing more accurate analysis to determine the density of defects because the radial density can be obtained from various angles. In this embodiment, the rotation of the sample support has been described in the clockwise direction, but the wafer defect inspection apparatus 100 as shown in FIGS. 1 and 2 actually performs the measurement while rotating in the counterclockwise direction, and the loading plate 60 Since the angular scale is displayed on the scale member 70 disposed to enclose the angular scale, it is also easy to check with the naked eye.

FIG. 12A shows a result 240a of measuring the density of OiSF without rotation using the method mentioned in the above procedure, and FIG. 12B is a result of measuring the density of LDP. The graph “1st OiSF test” represented by circles and solid lines in FIG. 12A represents the initial measurement results, and the graph “1st OiSF test-repeats” represented by triangles and dotted lines represents the results of repeated measurements using the same sample. The graph “LDP test” in circles and solid lines in FIG. 12B represents the initial measurement results, and the graph “LDP test-repeats” in triangles and dotted lines represents the results of repeated measurements using the same sample. As a result of repeated measurements in FIG. 12A and FIG. 12B, since the result values are almost identical to each other, the reliability of the result can be confirmed.

13A and 13B illustrate OiSF (FIG. 13A) and LDP (FIG. 13B) recognized by the wafer defect inspection apparatus 100 as shown in FIGS. 1 and 2 in the above process. The automatic recognition algorithm of the two defects will be described later.

Auto focusing on the surface of the sample is omitted because it is mentioned in relation to the DZ depth and BMD density measurements.

The following is a description of the algorithm that automatically recognizes the image of a defect that appears on the screen.

As shown in FIGS. 13A and 13B, both OiSF and LDP analyze the image automatically acquired by the lens of the microscope 50 in the histogram according to the intensity of the contrast, and the background of the sample having the largest contrast ratio within the region displayed on the screen. Real faults can be isolated. In the case of OiSF, since the minimum and maximum values of the defect size are set in advance, after obtaining the image by adjusting the auto focus, if the maximum length of the defect candidate shape is within the set size range, all defects are recognized and detected. . At this time, the minimum value and the maximum value of the defect size can be changed.

In the wafer defect inspection apparatus 100 as shown in FIGS. 1 and 2, the OiSF is recognized when the maximum length 900 and the minimum length 910 of the OiSF to be detected are two times or more, and the detected OiSF shapes If the maximum length of is in the set size range, it is counted.

In the case of LDP, the contrast intensity of the CCD camera image is analyzed by histogram in the same manner as above, and the background and the LDP, which are the highest contrast ratios, are separated. The minimum value and the maximum value size of the LDP are already set, and if the maximum length of the detected LDP candidate shape is within the set size range, the LDP is recognized and counted. In particular, in the case of LDP, only when a nucleus of lighter color than the surroundings such as the white dot 920 is constantly present in the defect, it is finally recognized as LDP. Therefore, OiSF and LDP are filtered according to various options, and only count defects that meet the final criteria.

In the above, the present invention has been described in detail with reference to preferred embodiments, but the present invention is not limited to the above embodiments, and various modifications can be made by those skilled in the art within the technical idea of the present invention. Is obvious.

1 is a schematic diagram of a wafer defect inspection apparatus capable of performing a wafer defect inspection method according to the present invention.

FIG. 2 is a side view of the wafer defect inspection apparatus shown in FIG. 1.

3 is a view for showing a vertical sample preparation method.

4 is a flowchart of an automatic inspection method according to the present invention.

5A to 5E show the results of measuring the DZ depths.

6A to 6C show results obtained after defining the depth from the surface of the wafer to the third defect using the same sample proceeded from FIGS. 5A to 5C as DZ depth.

7A to 7E show the results of measuring the BMD density.

8 shows the principle that the lens of the microscope automatically focuses on the cleaved surface of the sample.

9 is a view for explaining the principle of automatically recognizing the distance from the surface to the defect as the DZ depth.

10 is a flowchart of another automatic inspection method according to the present invention.

11 is a view showing a measurement state according to the rotation of the sample support.

12A is a result of measuring the density of OiSF, and FIG. 12B is a result of measuring the density of LDP.

FIG. 13A is a photograph of OiSF, and FIG. 13B is a photograph of LDP.

<Explanation of symbols for the main parts of the drawings>

10 ... Base 20 ... Stage 1

30 ... 2nd Stage 40 ... Revolving Stage

50 ... microscope 60 ... loading plate

70 ... scale member 100 ... wafer defect inspection device

200 ... wafer 210 ... cleaved surface

213 ... sample 230 ... notch

Claims (13)

As a wafer defect inspection method of a microscope auto focusing method, (A) setting up and down moving regions between the highest position and the lowest position of the microscope lens; (B) taking a sample at an initial position of the microscope lens to obtain an initial position image; (C) obtaining the highest position image and the lowest position image after the microscope lens is moved from the initial position to the highest position and the lowest position, respectively; (D) automatically comparing the initial position image clarity with the highest position image clarity and the lowest position image clarity, and applying the dichotomy to a distance where the sharpness is high; (E) photographing the sample at the moved position to obtain a moving position image; (E) automatically comparing the moving position image sharpness with the image sharpness before moving, and applying the dichotomy to a distance where the sharpness is high; And (G) repeatedly performing the steps (e) and (bar) sequentially to obtain a position where the focal point is best focused. A wafer having a microscope for observing a sample surface, a sample support for mounting the sample, a computer for controlling the operation of the microscope and the sample support, and a monitor for outputting a photographed image of the microscope and a processing screen of the computer. Using a defect inspection device, (a) performing defect auto-measurement after microscopic autofocusing at the first measurement point of the sample; (b) automatically storing and sorting the measurement result value and the measurement image; (c) moving the sample at predetermined intervals to perform autodetection of defects after microscopy autofocus; (d) automatically storing and sorting the measurement result value and the measurement image; And (e) sequentially repeating steps (c) and (d) until the measurement at the last measurement point is completed. A wafer having a microscope for observing a sample surface, a sample support for mounting the sample, a computer for controlling the operation of the microscope and the sample support, and a monitor for outputting a photographed image of the microscope and a processing screen of the computer. Using a defect inspection device, (a) performing the autodetection of defects along the radial direction after microscopy autofocusing at the first measurement point of the sample; (b) automatically storing and sorting the measurement result value and the measurement image; (c) rotating the sample by a predetermined angle to perform automatic defect measurement along the radial direction; And (d) repeating step (c) until the measurement at the last measurement point is finished. The method of claim 2, wherein before step (a), Manually adjusting the focus of the microscope at the first measurement point of the sample; Manually adjusting the focus of the microscope at the last measurement point of the sample by moving to the last measurement position of the sample; And Wafer defect inspection method further comprising the step of adjusting the horizontal and the left, right, before, after shaking of the sample based on the focus image on the monitor screen. The method according to claim 2 or 3, wherein if the sample information is input before the step (a), the measurement position of the sample is automatically generated during the movement on the monitor screen, and the measured value at that time is automatically written. Wafer defect inspection method characterized in that. 3. The method of claim 2, further comprising designating an area on the monitor for measuring the defect. According to claim 2 or 3, wherein the microscope autofocus adjustment, (A) setting up and down moving regions between the highest position and the lowest position of the microscope lens; (B) taking the sample at the initial position of the microscope lens to obtain an initial position image; (C) obtaining the highest position image and the lowest position image after the microscope lens is moved from the initial position to the highest position and the lowest position, respectively; (D) automatically comparing the initial position image clarity with the highest position image clarity and the lowest position image clarity, and applying the dichotomy to a distance where the sharpness is high; (E) photographing the sample at the moved position to obtain a moving position image; (E) automatically comparing the moving position image sharpness with the image sharpness before moving, and applying the dichotomy to a distance where the sharpness is high; And (G) repeatedly performing the steps (e) and (bar) sequentially to obtain a position where the focal point is best focused. The method of claim 2, wherein the defect measurement target is at least one of a depth of a defect layer (DZ) and a bulk micro defect (BMD) density. The method of claim 2, wherein before step (a), The method may further include selecting a recipe regarding the defect information, wherein the defect measurement target is a BMD, and an image automatically acquired by the lens of the microscope is analyzed by a histogram according to the intensity of the contrast, and the area within the area illuminated on the monitor screen. The BMD candidate is divided into a background and a BMD candidate of the sample having the highest contrast ratio, and the BMD candidate is recognized as a BMD if the maximum length of the BMD candidate is within a minimum value and a maximum value of the defect size set in the recipe. Wafer defect inspection method characterized in that the detection. 10. The method of claim 9, wherein the image automatically acquired by the lens of the microscope is analyzed by histogram according to the intensity of the contrast, and the position of the dark value from the surface of the sample to the light value in the bulk (bulk) direction to the surface of the sample Wafer defects, characterized by detecting the depth from the surface to the first detected BMD in the bulk direction as the depth of defect-free, or detecting the depth to each BMD sequentially found after the first as the depth of defectless method of inspection. The method of claim 3, wherein the target of the defect measurement is at least one of an oxygen induced stacking fault (OiSF) density and a large dislocation pit (LDP) density. The method of claim 3, wherein before step (a), The method further comprises a step of selecting a recipe relating to the information on the defect, wherein the object of the defect measurement is OiSF, and the image automatically acquired by the lens of the microscope is analyzed by a histogram according to the intensity of the contrast, and the area within the area illuminated on the monitor screen. When the region of the highest contrast ratio is separated into the background of the sample and the OiSF candidate, and the maximum length of the OiSF candidate is more than twice the minimum length and falls within the minimum and maximum values for the size of the defect set in the recipe. And detecting and detecting the OiSF candidate as OiSF. The method of claim 3, wherein before step (a), The method further comprises a step of selecting a recipe relating to the information of the defect, wherein the object of the defect measurement is LDP, and the image automatically acquired by the lens of the microscope is analyzed by a histogram according to the intensity of the contrast, and the area within the area illuminated on the monitor screen. In the LDP candidate, the region band having the highest contrast ratio is separated into the background of the sample and the LDP candidate. And detecting the LDP candidate as LDP when the color nucleus is present.

Images