Defect review apparatus and method for wafer

This application is based upon and claims the benefit of priority of the prior Japanese Patent Application No. 2008-302216, filed on Nov. 27, 2008 the entire contents of which are incorporated herein by reference.

The disclosures relate to an apparatus and a method for reviewing a defect based on two kinds of results of defect detection conducted under different detecting conditions.

Defects of a semiconductor wafer are detected by a detecting device using an oblique illumination optical system, a detecting device using an epi-illumination inspecting device, SEM (scanning electron microscope) appearance inspecting device, or the like. A process is performed to determine whether or not certain inspection data acquired by these detecting devices are identical. Such a process of determining identity greatly reduces the number of objects to be inspected.

Using a dark-field, bright-field, or similar defect detecting method, the position of a defect is specified and an SEM image of the area where a defect is detected is acquired, thereby reducing the total area of regions to be observed by the SEM. The method in which a defect detected by a defect detector is observed (and analyzed) using the SEM is called “A SEM review.”

Defect detection using various methods yields a quantity of defect detection data about each semiconductor wafer. An SEM review of the defect detection data makes it possible to categorize the defects thus detected according to defect type, such as, for example, lower layer pattern failure, an STI scratch, an inter-layer insulation film scratch, the presence of foreign matter, tungsten residue, etc.

Therefore, there is great demand for technology that can perform an SEM review more efficiently in a shorter time.

The embodiments relate to a defect review apparatus includes a storage device which stores data about a defect of an inspection target object; a first imaging device which captures an image located in a position on a surface of the inspection target object, the position being specified by information regarding the position of the inspection target object which has been input; and a control device which controls the first imaging device.

The storage device stores: first defect detection data including a defect number labeling the defect of the inspection target object detected by a first defect detection process, and information regarding the position of the defect; and second defect data including a defect number labeling the defect of the inspection target object detected by a second defect detection process, and information regarding its position.

The object and advantages of the invention will be realized and attained by means of the elements and combinations particularly pointed out in the claims.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not restrictive of the invention, as claimed.

FIG. 1 is a block diagram of a defect review apparatus according to an embodiment;

FIG. 2 is a flowchart for a defect review method according to the embodiment;

FIG. 3 is a detailed flowchart for step SA 2 of the defect review method according to the embodiment;

FIG. 4 is a detailed flowchart for step SB 7 of the defect review method according to the embodiment;

FIG. 5 is plan view illustrating a movable stage of the defect review apparatus according to the embodiment and an inspection target wafer placed on the movable stage;

FIGS. 6A to 6E are plan views of inspection target wafers illustrating a defect extraction process performed for defect alignment;

FIGS. 7A to 7C are tables illustrating examples of defect detection data;

FIG. 7D is a table illustrating examples of combined data.

FIG. 8A to 8C are plan views of inspection target wafers illustrating examples of the distributions of defects registered in the defect detection data;

FIG. 8D is a plan view of an inspection target wafer illustrating an example of defects registered in combined-data;

FIG. 9A to 9C are tables illustrating examples of defect detection data stored in the defect detection data storage sections; and

FIG. 10A to 10C are plan views of inspection target wafers for illustrating the order of capture of SEM images in a modification of the embodiment.

FIG. 1 illustrates a block diagram of a defect review apparatus 20 according to an embodiment, and a wafer to be inspected. The defect review apparatus 20 includes an input section 21 , a control device 22 , an output section 24 , a storage device 30 , SEM 25 , and an optical microscope 29 . The input section 21 receives defect detection data 42 A, 42 B, and 42 C. The defect detection data 42 A is acquired by subjecting a semiconductor wafer 40 to be inspected (hereinafter referred to as inspection target wafer) to a defect detection process 41 A under a given defect-detecting condition. Defects thus detected and the positions of these defects are registered in the defect detection data 42 A. The position of each defect is expressed by, for example, an XY coordinate system (also called “a target-object coordinate system”) defined on the surface of the inspection target wafer 40 . The defect detection data 42 B and 42 C are acquired by subjecting the inspection target wafer 40 to defect detection processes 41 B and 41 C respectively.

The defect detection processes 41 A to 41 C adopt bright-field defect detection, laser scattering (dark-field) defect detection, and the like and methods and conditions for these detections differ from one another.

The bright-field defect detection will first briefly be explained. First, while an inspection target wafer is scanned, an image of a pattern on the wafer surface is picked up by an imaging device such as a camera. Images in areas where the same pattern is formed are compared and, if any difference is detected between them, the difference is recognized as a defect. Examples of a method for comparing images include a random mode (chip comparison mode) and an array mode (cell comparison mode).

In the random mode, the image of a chip currently in focus is compared with images of adjacent chips on both sides. If any difference is detected between the image of the chip currently in focus and those of the adjacent chips on either side, the difference is recognized as a defect in the chip currently in focus. In the array mode, where a number of cells of the same pattern exist in one chip as in a memory cell, the image of a cell currently in focus is compared with images of the adjacent cells on both sides. If any difference is detected between the former and the latter images, this difference is recognized as a defect of the cell currently in focus.

Next will briefly be described the dark-field defect detection. A laser is emitted onto an inspection target wafer from an oblique direction, and light scattering from this wafer is detected by a camera or the like. Image signals produced by light scattering from a normal pattern are ignored but image signals produced by light scattering from a foreign matter are extracted to detect the foreign matter.

For example, the dark-field defect detection may be adopted in the defect detection processes 41 A to 41 C. The incident angles of laser among these defect detection processes 41 A to 41 C are set so as to differ. Light-field defect detection may be used in the defect detection processes 41 A to 41 C. In this case also, the brightnesses and wavelengths, and the like of light sources among these defect detection processes 41 A to 41 C are set so as to differ. Alternatively, the defect detection processes 41 A to 41 C may adopt dark-field and light-field defect detections in combination.

The defect detection data 42 A to 42 C supplied to the input section 21 are stored in defect detection data storage sections 31 A to 31 C respectively in the storage device 30 through processes performed by the control device 22 . Incidentally, there may additionally be four or more defect detection data storage sections. The control device 22 controls the SEM 25 and the optical microscope 29 .

The SEM 25 includes a movable stage 26 and an electron beam source 27 . The movable stage 26 moves in two dimensions while holding an inspection target wafer 40 . An observing coordinate system (UV coordinate system) is defined for the movable stage 26 . The SEM 25 transmits to the control device 22 an SEM image located in a position specified by coordinates in the observing coordinate system. The optical microscope 29 transmits to the control device 22 an optical image specified by the coordinates in the observing coordinate system.

The storage device 30 further includes a combined-data storage section 34 , an extractable area storage section 35 , and an image data storage section 36 . The functions of these will be described below.

The output section 24 displays defect review results, SEM images, and optical images.

FIG. 2 illustrate a flowchart for a defect review method according to the embodiment. In step SA 1 , under different detecting conditions the defect detection processes 41 A to 41 C are performed for an inspection target wafer 40 , and defect detection data 42 A to 42 C are acquired. Here, “the defect detection processes are performed under different detecting conditions” means not only that the defect detection processes are performed under different conditions using the same detecting device but also that the defect detection processes are performed using different detecting devices as in light-field defect detection and dark-field defect detection. In the embodiment, defect detection is conducted under three different conditions, thereby acquiring three sets of defect detection data, 42 A to 42 C. However, defect detection may be conducted under at least two different conditions, thereby acquiring at least two sets of defect detection data. Also, defect detection may be conducted under four or more different detection conditions, thereby acquiring four or more sets of defect detection data.

The defect detection data 42 A to 42 C acquired are input to the SEM review apparatus 20 . The defect detection data 42 A to 42 C inputted are stored in the defect detection data storage sections 31 A to 31 C respectively.

In step SA 2 , the inspection target wafer 40 is placed on the movable stage 26 of the SEM 25 . After being placed there, the inspection target wafer 40 is aligned with the target object coordinate system defined for the inspection target wafer 40 . Specifically, the relative positional relation of the wafer 40 with the observing coordinate system defined on the movable stage 26 is measured.

FIG. 3 illustrate a flowchart for step SA 2 in detail, and FIG. 5 is a plan view illustrating the movable stage 26 and inspection target wafer 40 held thereon. The inspection target wafer 40 is held on the moveable stage 26 so that the X-axis of the target object coordinate system and the U-axis of the observing coordinate system are substantially parallel to each other. On account of this, the Y-axis of the target object coordinate system and the V-axis of the observing coordinate system are also located substantially parallel to each other. To place the inspection target wafer 40 on the movable stage 26 , the original positions of the target object coordinate system and observing coordinate system are co-aligned.

In step SB 1 , SEM images of specific patterns in three chips 44 A, 44 B, and 44 C located along a diameter parallel to the Y-axis are acquired by the SEM 25 while the movable stage is moved in the direction of the V-axis. For example, the images of patterns corresponding to the respective corners of the chips 44 A to 44 C are acquired. The wafer coordinates of these specific patterns are already known. The chips 44 A and 44 C are selected from near each end of each diameter, and the chip 44 B is selected from the middle of these diameters or from near the middle.

In step SB 2 , SEM images of specific patterns in three chips 44 D, 44 B, and 44 E located along a diameter parallel to the X-axis are acquired by the SEM 25 while the movable stage is moved in the direction of the U-axis. The chips 44 D and 44 E are selected from near each end of the diameter. The chip 44 B is identical to the one selected in the previous step, SB 1 .

In step SB 3 , the specific pattern images thus acquired are analyzed, and thereby the coordinates of the specific patterns in the five chips, 44 A to 44 E, are obtained in the observing coordinate system. Subsequently, a determination is made whether any displacement between the coordinates in the target object coordinate system acquired from the corresponding patterns and those in the observing coordinate system falls within a permissible range or not. The permissible range is set to, for example, 1 μm. If the displacement between coordinates falls within the permissible range, step SB 7 is executed, and if the displacement exceeds the permissible range, step SB 4 is executed.

In step SB 4 , a determination is made whether a correction process for the observing coordinate system has been carried out or not. If the determination is made that the correction process has not been carried out, a correction process for the observing coordinate system is performed in step SB 5 . After this correction process, the flow returns to step SB 1 and the image acquiring process is performed using the corrected observing coordinate system. If it is determined in step SB 4 that a correction process for the observing coordinate system has been carried out, the equipment such as the movable stage 26 may have been in an abnormal state. Accordingly, a warning is provided in step SB 6 urging an operator to check for the abnormal state.

In step SB 7 executed as a result of a determination made in step SB 3 that the displacement between the coordinates falls within the permissible range, a defect alignment process is performed.

FIG. 4 illustrates a flowchart for the defect alignment process. FIG. 6A illustrates the distribution of a defect 45 registered in, for example, defect detection data 42 A. There may be a case where defects are close together locally. Such defects 45 C which are close together locally are called “a cluster defect.” A cluster defect is determined as follows: if the number of defects, for example, within a very small area exceeds a given upper limit, the defects within the very small area are recognized as a cluster defect.

In step SC 1 , cluster defects 45 C are removed from an extraction target. In FIG. 6B , cluster defects 45 C removed from the extraction target are indicated with white circles.

In step SC 2 , as illustrated in FIG. 6C , one defect in the vicinity of the periphery of the inspection target wafer 40 is extracted from the defects 45 from which the cluster defects 45 C are removed, and this defect is labeled as a first defect DP 1 . Here, “the vicinity of the periphery” is defined as an annular area 46 that has the periphery of the inspection target wafer 40 as its outer edge. The width of the annular area 46 is stored in the extractable area storage section 35 illustrated in FIG. 1 . The width is, for example, 5 μm.

As a first defect DP 1 , any one defect 45 may be extracted from the annular area 46 . It may be a defect 45 closest to the periphery. If no defects are found within the annular area 46 , the presence or absence of a defect is determined while the annular area 46 is broadened 1 μm at a time. If there are defects within the broadened annular area, any one defect is extracted from among them as a first defect DP 1 . If no defects have been detected within the annular area 46 , the outermost defect among defects 45 located in an area further inside than this annular area may be extracted as the first defect DP 1 .

In step SC 3 , as illustrated in FIG. 6D , a defect 45 located closest to the intersection 48 where a first straight line 47 extending from the first defect DP 1 as its start point and passing through the center of the inspection target wafer 40 meets the periphery of the inspection target wafer 40 is extracted as a second defect DP 2 . Alternatively, any one of the defects whose distance from the intersection 48 is equal to or shorter than the distance stored in the extractable area storage section 35 may be extracted.

In step SC 4 , as illustrated in FIG. 6E , defects located closest to the intersections 51 and 52 where a straight line 50 crossing the first straight line 47 at a right angle and passing through the center of the inspection target wafer 40 meets the periphery of the inspection target wafer 40 are extracted as a third defect DP 3 and a fourth defect DP 4 respectively. Alternatively, any one of the defects whose distance from the corresponding intersections, 51 and 52 , are equal to or shorter than the distance stored in the extractable area storage section 35 may be extracted.

Instead of the straight line 50 , a straight line that intersects, at a right angle, a diameter passing through the second defect DP 2 , and passes through the center of the inspection target wafer 40 , may be used. Also, instead of the straight line 50 , a straight line that intersects at a right angle a straight line passing through the first and second defects DP 1 and DP 2 may be used.

In the foregoing method, the first to fourth defects DP 1 to DP 4 are extracted from among the defects registered in the defect detection data 42 A. However, they may be extracted from among defects registered in another defect detection data. Alternatively, they may be extracted from among defects registered in any of a plurality of detection data.

In step SC 5 , SEM images of the first to fourth defects DP 1 to DP 4 are acquired. The observing coordinate system is corrected so that these defects are located in the respective centers of the SEM images.

If there is a false defect among the extracted first to fourth defects DP 1 to DP 4 and no defect is detected in the SEM image, another defect may be extracted from among defects in the vicinity of this false defect.

An actual defect is observed to correct the observing coordinate system, so that SEM images in which the defects are located almost in the center may be obtained for the defects other than the first defect DP 1 to the fourth defects. In addition, the first to fourth defects DP 1 to DP 4 may be distributed near the periphery of the inspection target wafer 40 and at substantially regular intervals along the circumference. This improves accuracy in correcting the observing coordinate system.

If the first to fourth defects DP 1 to DP 4 are detected from among the cluster defects 45 C, defects adjacent to the defects actually extracted are likely to be erroneously recognized as extracted ones. Such misrecognition may be prevented by removing the cluster defects 45 C before defect extraction.

After the defect alignment process is finished, step SA 3 illustrated in FIG. 2 is executed. In step SA 3 , the SEM images of defects are acquired based on information regarding the position of detected defects. The procedure for step SA 3 will now be described in detail.

FIGS. 7A to 7C illustrate examples of defect detection data 42 A to 42 C respectively. An identifier for a detected defect and information regarding its position are registered in each of the defect detection data 42 A to 42 C. For example, the positions of the defects DA 1 , DB 1 , and DC 1 of the defect detection data 42 A, 42 B, and 42 C are all defined by (x 1 , y 1 ). Therefore, these defects are recognized as the same defect. Thus, based on information regarding the position of the detected defects, the determination of identity of the defects is made. Conditions for determining identity of defects need not require that the positions of the defects be exactly the same. Even if the detected positions of the defects differ slightly, when the difference falls within the range of detection error or is substantially equal to the dimensions of the defect, a determination is made that these defects are the same.

If one defect is registered in at least in two of the defect detection data 42 A to 42 C, this defect is called “a duplicate defect.” Combined-data 42 M is created by combining the defect detection data 42 A to 42 C so that the duplicate defect is registered as a single defect. The combined-data 42 M is stored in the combined-data storage section 34 .

FIG. 7D illustrates examples of the combined-data 42 M. The defects DA 1 , DB 1 , and DC 1 of defect detection data 42 A, 42 B, and 42 C respectively are registered as one defect.

FIGS. 8A to 8C illustrate examples of the respective distributions of defects registered in the corresponding defect detection data 42 A to 42 C. FIG. 8D illustrates an example of the distribution of the defect registered in the combined-data 42 .

The control device 22 sequentially captures SEM images in positions specified by defect information regarding position registered in the combined-data 42 M. The captured SEM images are stored in the image storage section 36 illustrated in FIG. 1 . FIG. 8D illustrates an example of the defect order of capturing of the SEM images, indicated by arrows.

In step SA 4 , a determination is made if any of the captured SEM images shows no defects. If there is an SEM image in which no defects have been detected, an optical image in the position where no defects have been detected is captured in step SA 5 , and then step SA 6 is executed. If defects have been detected in all the SEM images, step SA 5 is skipped and step SA 6 is executed. The optical image captured in step SA 5 is stored in the image data storage section 36 illustrated in FIG. 1 .

Also, in optical image capturing as in the SEM image capturing, control is exerted so as not to capture duplicate optical images of the duplicate defect.

In step SA 6 , the type of defect detected is determined based on the SEM image and the optical image. For instance, if no defects are found in an SEM image while a defect is found in an optical image, the defect is determined as one formed in a lower layer. If a defect is found in an SEM image, the defect is determined as a defect in an upper layer. If no defect is found in either an SEM image or an optical image, a determination is made that this defect is not an actual defect resulting from the process but is a false defect resulting from the defect detection process.

Further, according to the geometrical shape, size, pattern and the like of an SEM image or an optical image, the type of defect may be categorized as a lower layer pattern failure, a lower layer foreign matter, an inter-layer insulation film scratch, an upper layer foreign matter, embedded tungsten residue, and the like.

FIGS. 9A to 9C illustrate the contents of the defect detection data storage sections 31 A to 31 C. A defect type, a pointer indicating the place where an SEM image is stored, and a pointer indicating the place where an optical image is stored, are associated and stored for each of defects registered in the defect detection data 42 A to 42 C.

A defect type, an SEM image, or an optical image is output to the output section 24 at the request of an operator. Optimal conditions for the defect detection process may be estimated from a number of defect detection data in which the defect types are registered. For example, optimal conditions for the defect detection have fewer false defects and fewer detection failures of defects in an upper layer.

Referring to FIGS. 10A to 10C , next will be described a modification of the embodiment described above. In this example, specifically, the procedure for step SA 3 illustrated in FIG. 2 differs from that in the embodiment described above.

FIG. 10A to 10C illustrate examples of the respective distributions of defects registered in the corresponding defect detection data 42 A to 42 C. As illustrated in FIG. 10A , SEM images in the positions of defects registered in the defect detection data 42 A are acquired sequentially. Defects in white circles in FIG. 10A indicate that SEM images of them are captured.

As illustrated in FIG. 10B , SEM images in the positions of defects registered in the subsequent image detection data 42 B, are captured sequentially. At this time, a determination is made whether a defect of which an SEM image is to be captured is a duplicate or not. If the defect of which an SEM image is to be captured is a duplicate, an SEM image of the defect is not captured. An SEM image of a defect that is identical to one that has been captured in the process illustrated in FIG. 10A is assigned to the defect registered in the defect detection data 42 B. If the defect of which an SEM image is to be captured is not a duplicate, an SEM image in the position of the defect is captured.

As illustrated in FIG. 10C , SEM images in the positions of defects registered in the next defect detection data, 42 C, are captured sequentially. In this case also, a determination is made whether each defect of which an SEM image is to be captured is a duplicate or not. SEM images are captured only where these are non-duplicates. FIG. 10C illustrates the case where defects DC 1 to DC 8 are all duplicate images. No SEM images of the defects DC 1 to DC 8 registered in the defect detection data 42 C are captured.

The embodiment described above and the modification avoid capturing a duplicate SEM image corresponding to a defect determined as a duplicate. Thus, the time required for defect review can be shortened. In addition, avoiding repeated emission of electron beams onto the same place reduces the damage on inspection target wafers.

In the embodiment described above, an example of a defect review for a semiconductor wafer is used. However, defect review may also be performed for other inspection target objects such as a TFT substrate used in a liquid crystal display apparatus.

All examples and conditional language recited herein are intended for pedagogical purposes to aid the reader in understanding the invention and the concepts contributed by the inventor to furthering the art, and are to be construed as being without limitation to such specifically recited examples and conditions, nor does the organization of such examples in the specification relate to a showing of the superiority and inferiority of the invention. Although the embodiment(s) of the present inventions have been described in detail, it should be understood that various changes, substitutions, and alterations could be made hereto without departing from the spirit and scope of the invention.