JP2009222462A - Flaw inspection device and flaw inspection method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a flaw inspection device capable of rapidly and accurately inspecting what process of a plurality of sample manufacturing processes a flaw or stain is caused in without relying on the discretion of a worker, and a flaw inspection method. <P>SOLUTION: The flaw inspection device has a step for observing a sample using a multiwavelength microscopic apparatus 1 to display its observation image on an observation image display device 3, a step for collating the observation image displayed on the observation image display device 3 with the image based on the sample data of a memory device 4 by an image recognizing device 5, a step for detecting the flaw or the like of the sample by collating the observation image with the image based on the sample data and a step for judging what process of a plurality of the sample manufacturing processes a flaw or the like is caused in. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a defect inspection apparatus capable of inspecting the presence or absence of scratches or dirt in the depth direction of a multilayer structure in a semiconductor substrate such as silicon and knowing in which process the scratch or dirt has occurred, and the inspection apparatus The present invention relates to a defect inspection method using.

Conventionally, for example, a confocal microscope is used as a defect inspection apparatus for inspecting defects of a sample such as a semiconductor wafer or a chip (see Patent Document 1). The confocal microscope can form an image (all-in-focus image) in a state where a sample having a difference in elevation is completely focused on any part.
In the confocal optical system, the reflected light enters the photodetector only when the focal point is perfectly located. In other words, only the reflected light of the in-focus portion passes through the pinhole and enters the photodetector. For this reason, the brightness of the image is maximized when the image is in focus, and the brightness decreases when the image is out of focus. The confocal microscope uses this confocal optical system, and an omnifocal image in which the entire screen is in focus can be obtained even with an uneven sample. A normal optical microscope has a finite depth of focus, so it is difficult to focus on the entire image. However, an omnifocal image is focused on the entire image.

Defect inspection equipment for semiconductor wafers and chips, as represented by confocal microscopes, uses the properties of each spectrum of red, green, and blue, but its use is just observing defect defects. Scratches or dirt are determined based on the judgment of the person. Conventionally, red, green, and blue spectra are individually reflected, and a composite image that is focused in the depth direction is created using the reflected image to make the color image clearer.
Patent Document 1 discloses a three-dimensional shape measuring apparatus and a measuring method capable of measuring the three-dimensional shape of layers other than the surface layer. This three-dimensional shape measuring apparatus includes a wavelength switching mechanism that switches the wavelength of illumination light from a light source, a stage drive mechanism that scans the focal position with respect to the sample along the optical axis, and confocal reflected light reflected by the sample. A CCD camera that is detected via an optical system, a CCD camera that scans the focal position, and a processing device that performs processing based on detection light detected by the CCD camera are provided. The processing device calculates a focal point position based on the detection light detected by the CCD camera and constructs a three-dimensional shape. By switching to a different wavelength by the wavelength switching mechanism, a three-dimensional shape of the upper surface of the first layer and the upper surface of the second layer is constructed.
JP 2006-350078 A

When the conventional defect inspection apparatus as described above is used, it is left to the discretion of the operator to determine the detection of scratches and dirt and in which process the scratches and dirt are generated. With such means, it takes time to make judgments and identifications, and there are many judgment errors. In addition, the operator also collected the types of scratches and dirt and the generation process, and the time and labor were tremendous.
The present invention has been made to solve the above-described problems, and it is quick and accurate without leaving to the discretion of an operator which one of a plurality of processes for manufacturing a substrate is caused by scratches or dirt. Provided are a defect inspection apparatus and a defect inspection method capable of inspection.

  One aspect of the defect inspection apparatus of the present invention includes a multi-wavelength microscope apparatus using a light source having red, green, and blue spectra for observing a sample, an observation image display apparatus that displays an observation image of the sample, and sample information stored in advance. And a storage device that stores good and defective samples, and an image recognition device that collates the observation image displayed on the observation image display device with an image based on the sample information of the storage device; From the comparison result, it is detected which spectral component of the light of each spectral component has a defect in the observed image and recorded as a defective or non-defective product in the storage device. It is characterized by comprising an arithmetic processing unit for determining in which process it is formed.

  Further, according to one aspect of the defect inspection method of the present invention, a step of irradiating a sample with light having red, green, and blue spectra and displaying an observation image by the light of each spectrum component; The step of collating with the image based on the sample information stored as a defective product, and the observation image by light of which spectral component among the spectral components is defective by collating the observation image with the image based on the sample information And detecting as a defective product or a non-defective product, and determining in which step of the sample manufacturing process the defect is formed.

  With the above configuration, the present invention enables a quick and accurate inspection without leaving the operator's discretion as to which of the plurality of processes for manufacturing a substrate is a scratch or a stain.

The present invention is characterized in that a multi-wavelength microscope apparatus using a light source having red, green, and blue spectra is used for defect inspection, and an observation image of a sample can be clearly observed. For example, there is provided a defect inspection apparatus and method capable of quick and accurate inspection, which is performed without leaving to the discretion of an operator which one of a plurality of processes for manufacturing a semiconductor wafer or chip is caused by scratches or dirt. It is to provide.
Hereinafter, embodiments of the invention will be described with reference to examples.

First, an embodiment will be described with reference to FIGS.
1 is a block diagram of a defect inspection apparatus described in the embodiment, FIG. 2 is a flowchart illustrating a defect inspection method using the defect inspection apparatus shown in FIG. 1, and FIG. 3 is a defect inspection apparatus shown in FIG. FIG. 2 is a diagram for explaining the proper use of multiple wavelengths in a multi-wavelength microscope apparatus of a sample (here, a semiconductor chip is used) where scratches or dirt (hereinafter referred to as scratches) are applied to the surface of a semiconductor substrate and are observed at all wavelengths. 4 is a cross-sectional view of a semiconductor chip in which a scratch or the like is applied to the surface of a semiconductor substrate and observed only with an R wavelength spectrum component in the observation with the multi-wavelength microscope apparatus of the defect inspection apparatus shown in FIG. FIG. 6 is a cross-sectional view of a semiconductor chip in which scratches and the like are applied to the surface of a semiconductor substrate and observed only with a B wavelength spectrum component in the observation with the multi-wavelength microscope apparatus of the defect inspection apparatus shown in FIG. FIG. 7 is a diagram for explaining how to use multiple wavelengths in the multi-wavelength microscope apparatus of the defect inspection apparatus shown in FIG. 1, where a scratch or the like is applied to the passivation film, and a cross-sectional view of a semiconductor chip observed at all wavelengths, FIG. FIG. 8 is a cross-sectional view of a semiconductor chip observed with only a red spectral component (hereinafter referred to as an R wavelength spectral component) in the observation with a multi-wavelength microscope apparatus of a defect inspection apparatus, with scratches or the like being applied to the passivation film. FIG. 9 is a cross-sectional view of a semiconductor chip in which scratches and the like are applied to the passivation film and observed only with a blue spectrum component (hereinafter referred to as a B wavelength spectrum component). It is a figure explaining the multi-wavelength selective use in the multi-wavelength microscope apparatus of the defect inspection apparatus to show, except the uppermost layer (passivation film) and the uppermost layer ( During a cross-sectional view of the semiconductor chip was observed scratches on the insulating film).

As shown in FIG. 1, the defect inspection apparatus of this embodiment is a multi-wavelength microscope apparatus 1 using a light source having red, green, and blue spectra for observing a semiconductor chip transferred by a wafer or chip tray automatic transfer apparatus 2. An observation image display device 3 that displays an observation image of the semiconductor chip, a storage device 4 that stores sample information in advance and stores non-defective and defective samples, and an observation image display device 3. In addition, the image recognition device 5 that collates the observation image with the image based on the sample information in the storage device 4, and based on the collation result, the observation image by the light of which spectral component among the light of each spectral component is scratched and stained. In the process of manufacturing the sample, scratches and dirt are recorded as defective or non-defective products are recorded in the storage device 4 And a or the determining processing unit 6 which is.
In this defect inspection apparatus, the semiconductor chip is irradiated with light having red, green, and blue spectrums, and the observation images of the light of each spectral component are collated with images based on sample information stored in the storage device in advance. Detecting whether there is a scratch or the like in the observation image of the spectral component light, and further determining in which step of the plurality of steps for manufacturing the sample the scratch or the like it can.

When using the multiwavelength microscope apparatus 1 to inspect the surface of a semiconductor substrate constituting a semiconductor chip as a sample, an observation image using a red spectral component is effective. The red spectral component is transmitted through the uppermost layer of the substrate surface such as silicon and reflected by the substrate surface. That is, the unevenness of the substrate surface can be displayed as an image. The uppermost film is a transparent passivation film 15 such as polyimide. Further, the blue spectral component is reflected by the surface of the uppermost transparent film. That is, the unevenness on the surface of the passivation film 15 can be displayed as an image. Therefore, if the silicon substrate is flawed and the passivation film immediately above it is not flawed, it can be determined that it is caused by a process (previous process) prior to the final process (post process) of the wafer processing process (FIG. 9). (See (a)). According to the present invention, it is possible to accurately perform defect inspection using the characteristics of each spectral component.
As already described, the red spectrum (R) component is transmitted through the uppermost layer of the substrate surface such as silicon and reflected by the substrate surface (see FIG. 9B). The blue spectrum (B) component is reflected on the surface of the uppermost transparent film (see FIG. 9A). The green spectrum (G) component has the properties of both a red spectrum (R) component and a blue spectrum (B) component, is transmitted through the uppermost transparent film, and is formed between the transparent film and the substrate surface. Reflected by any of the plurality of interlayer insulating films (see FIG. 9).

Next, a method of inspecting a sample (semiconductor chip) for defects using the defect inspection apparatus shown in FIG. 1 will be described with reference to FIGS.
The defect inspection method shown in FIG. 2 includes steps (1) to (14). A semiconductor chip as a sample is transferred from the wafer or chip tray automatic transfer apparatus to the multi-wavelength microscope apparatus 1 (step (1)). Next, the semiconductor chips are observed in the order of indexes stored in advance (step (2)). Thereafter, all wavelength components (R, B, G) are irradiated simultaneously. The state in which the semiconductor chip is irradiated with all wavelength components simultaneously is as shown in FIGS. According to the multiwavelength microscope apparatus 1, a clear image higher than that of a metal microscope can be captured and displayed by the CCD regardless of the depth direction.

3 to 8 show sectional views of the semiconductor chip. In these drawings, the description of the semiconductor element and the like is omitted, and the multilayer wiring portion on the substrate surface is described.
In the figure, a field oxide film 11 such as silicon oxide is formed on the surface of a semiconductor substrate 10, and a first layer wiring pattern 16 such as a metal wiring is formed on the field oxide film 11. The first layer wiring pattern 16 is covered with an interlayer insulating film 12 such as silicon oxide. A second layer wiring pattern 17 such as a metal wiring is formed on the surface of the interlayer insulating film 12. The second layer wiring pattern 17 is covered with an interlayer insulating film 13 such as silicon oxide. A third layer wiring pattern 18 such as a metal wiring is formed on the surface of the interlayer insulating film 13. The third layer wiring pattern 18 is covered with an interlayer insulating film 14 such as silicon oxide. The first layer wiring pattern 16 and the second layer wiring pattern 17 are connected by a connection wiring 19, and the second layer wiring pattern 17 and the third layer wiring pattern 18 are connected by a connection wiring 22. Further, the interlayer insulating film 14 is covered with a transparent passivation film 15 made of polyimide or the like as the uppermost layer.

When the semiconductor chip is irradiated with all the wavelength components at the same time, for example, the red spectral component (R wavelength spectral component) is transmitted through the transparent passivation film 15 and the interlayer insulating films 12 to 14 and reflected on the surface of the semiconductor substrate 10. The blue spectral component (B wavelength spectral component) is reflected by the surface of the transparent passivation film 15. The green spectral component (G wavelength spectral component) is transmitted through the transparent passivation film 15 and reflected by any one of the interlayer insulating films 12 to 14.
3, 4, and 5 have scratches 20 on the surface of the semiconductor substrate 10, and the semiconductor chips in FIGS. 6, 7, and 8 have scratches 21 on the passivation film 15.
The observation image is displayed on the observation image display device 3 and collated with the image in the storage device 4 by the image recognition device 5. As a result of collating the observation image of the semiconductor chip with the image of the storage device 4 by the image recognition device 5, the sample is classified into a semiconductor chip without scratches (OK) and a semiconductor chip with scratches (NG) (Step ( 3)).

If the semiconductor chip is not damaged (OK), it is recorded as “good” in the storage device 4 by the arithmetic processing device 6 (step (4)). Furthermore, the semiconductor chip recorded in the storage device 4 as “good” by the arithmetic processing device 6 is displayed on the observation image display device 3 together with the observation image as one of good products (step (5)).
The arithmetic processing unit 6 determines, for example, which process in the sample manufacturing process the coordinate information of the wafer or chip tray, statistical processing information on the presence / absence of a sample scratch (for example, NG chip / wafer lot parameter), scratches, etc. It has information to be identified, and the observation image display device 3 determines and displays which process causes scratches and the like.

  On the other hand, if the semiconductor chip has scratches or the like (NG), it is determined in which step of the semiconductor chip the scratches or the like have occurred by the processes after step (6). First, observation is started by irradiating the semiconductor chip with only the B wavelength spectrum component (see FIGS. 5 and 8). When only the B wavelength spectrum component is observed, only the irregularities of the passivation film 15 are projected, so that the scratches 20 on the semiconductor substrate 10 cannot be confirmed. If there is no scratch or the like (OK), the observation is started by irradiating only the R wavelength spectrum component to the semiconductor chip (step (7)).

When the scratch or the like is confirmed, the arithmetic processing unit 6 records this semiconductor chip in the storage device 4 as “defective product” (step (8)). Then, the semiconductor processing device 6 counts the semiconductor chips as one of “defective products” and displays it on the observation image display device 3 together with the observation image. At the same time, the arithmetic processing unit 6 determines that the surface of the semiconductor substrate is scratched (see FIG. 4) and displays on the observation image display device 3 that this scratch or the like is caused by the “pre-process” in the semiconductor chip manufacturing process ( Step (9)). As shown in FIG. 4, when the semiconductor chip is irradiated with only the R wavelength spectrum component and observed, only the irregularities on the surface of the semiconductor substrate 10 are projected. Therefore, if there is a scratch or the like here, it can be determined that it is caused by a “pre-process” in the semiconductor chip manufacturing process.
In step (6), if there is a scratch or the like (NG), first, the arithmetic processing unit 6 records “defective product” in the storage device 4 (step (10)). Next, the arithmetic processing unit 6 counts the semiconductor chips as one of “defective products”, and displays it on the observation image display device 3 together with the observation image. At the same time, the arithmetic processing unit 6 determines that the scratches or the like are applied to the uppermost passivation film 15 (see FIG. 8). The observation image display device 3 indicates that the scratches are caused by the “post process” in the semiconductor chip manufacturing process. It is displayed (step (11)).

As described above, the semiconductor chip is observed with all the wavelength components to recognize the scratches, and further, only the B wavelength spectrum component is irradiated on the semiconductor chip and observed to determine that the scratches are applied to the passivation film. Therefore, it is determined that it is caused by the “post-process” in the semiconductor chip manufacturing process. However, in such observation, for example, when a scratch or the like overlaps both the passivation film and the semiconductor substrate surface, the scratch or the like on the semiconductor substrate surface cannot be specified. In that case, it can specify by step (12)-(14).
Hereinafter, steps (12) to (14) will be described.
After step (11), only the R wavelength spectrum component is irradiated onto the semiconductor chip to start observation (step (12)). If it is determined from this observation that the semiconductor chip is not damaged, the semiconductor chip is present only in the scratch on the passivation film. At this time, the arithmetic processing unit 6 does not The observation image display device 3 is kept displayed as being caused by the “post-process” (step (13)).

In addition, when scratches or the like are observed in the semiconductor chip by this observation, scratches or the like on the passivation film due to the “post process” are originally recognized on the semiconductor chip. In addition, scratches or the like due to the “pre-process” are recognized at the position on the semiconductor substrate. Therefore, the arithmetic processing unit 6 causes the observation image display unit 3 to display that there are scratches of “pre-process” and “post-process” at the same place (step (14)).
Through the above steps, it is possible to perform an accurate inspection with few detection errors without leaving to the discretion of the operator which one of a plurality of processes for manufacturing a semiconductor chip is caused by a scratch or the like. In this embodiment, the B wavelength spectral component and the R wavelength spectral component are used as the light source used in the multiwavelength microscope apparatus. However, for example, other components such as a green spectral component (G wavelength spectral component) are grasped and used. I can do it.
In the defect inspection apparatus of this embodiment, since an image recognition apparatus is added to the multi-wavelength microscope apparatus, defect inspection can be automatically performed, and the types of scratches can be made into a database and collated therewith. Scratch types and scratch generation processes can be easily counted and monitored.

The block diagram of the defect inspection apparatus demonstrated in an Example. The flowchart figure explaining the defect inspection method using the defect inspection apparatus shown in FIG. FIG. 2 is a cross-sectional view of a semiconductor chip (sample) that is observed at all wavelengths with scratches or dirt on the surface of a semiconductor substrate for explaining the use of multiple wavelengths in the multi-wavelength microscope apparatus of the defect inspection apparatus shown in FIG. In the defect inspection apparatus shown in FIG. 1, a semiconductor chip (sample) sectional view in which scratches or dirt is applied to the surface of a semiconductor substrate and is observed only with a red spectrum component (R wavelength spectrum component). In the defect inspection apparatus shown in FIG. 1, a semiconductor chip (sample) sectional view in which scratches or dirt is applied to the surface of a semiconductor substrate and is observed only with a blue spectrum component (B wavelength spectrum component). FIG. 2 is a diagram for explaining how to use multiple wavelengths of the defect inspection apparatus shown in FIG. 1, and a semiconductor chip (sample) cross-sectional view in which scratches or dirt are applied to the passivation film and observed at all wavelengths. In the defect inspection apparatus shown in FIG. 1, a semiconductor chip (sample) sectional view in which scratches or dirt are applied to the passivation film and observed only with a red spectrum component (R wavelength spectrum component). In the defect inspection apparatus shown in FIG. 1, a semiconductor chip (sample) sectional view in which scratches or dirt are applied to the passivation film and observed only with a blue spectrum component (B wavelength spectrum component). FIG. 2 is a diagram for explaining how to use multiple wavelengths of the defect inspection apparatus shown in FIG. 1, and is a cross-sectional view of a semiconductor chip (sample) for observing scratches or dirt on the uppermost layer and other than the uppermost layer.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Multi-wavelength microscope apparatus 2 ... Wafer and chip tray automatic conveyance apparatus 3 ... Observation image display apparatus 4 ... Memory | storage device 5 ... Image recognition apparatus 6 ... Arithmetic processing apparatus 10 ... Semiconductor substrate 11 Field oxide film 12, 13, 14 Interlayer insulating film 15 Passivation film 16, 17, 18 ... Wiring pattern 19, 22 ... Connection wiring 20, 21, ...・ Scratches or dirt

Claims (2)

A multi-wavelength microscope apparatus using a light source having red, green, and blue spectrum for observing a sample, an observation image display apparatus that displays an observation image of the sample, sample information is stored in advance, and non-defective and defective samples are stored. A storage device that stores non-defective products, an image recognition device that collates the observation image displayed on the observation image display device with an image based on the sample information of the storage device, and the light of each spectral component from the collation result Which spectral component of the observation image by light is detected as a defect and recorded as a defective or non-defective product in the storage device, and at which step of the sample manufacturing process the defect is formed. A defect inspection apparatus comprising an arithmetic processing unit for determining. Irradiating a sample with light having red, green, and blue spectrum, displaying an observation image by the light of each spectrum component, and comparing the observation image with an image based on sample information stored in advance as good and defective products And comparing the observed image with an image based on the sample information to detect which spectral component of the spectral component is defective in the observed image and record it as a defective or non-defective product And a step of determining in which step of the sample manufacturing process the defect is formed.

Images