Defect inspection device and its method

Description: TECHNICAL FIELD [Detailed description of the invention]

【0001】

[Technical field to which the invention belongs]

The present invention relates to a defect inspection apparatus and a method thereof for discriminating and inspecting defects such as scratches and particulate foreign substances generated in a flattening process by a polishing or grinding technique used in semiconductor manufacturing or magnetic head manufacturing.

【0002】

[Conventional technology]

Conventional techniques for discriminating and inspecting foreign matter adhering to a semiconductor wafer on which a circuit pattern is formed from the circuit pattern include JP-A-3-102248 (Prior Technique 1) and JP-A-3-102249 (Prior Technique 2). )It has been known. That is, in the prior arts 1 and 2, foreign matter on the semiconductor substrate is emphasized by oblique illumination and detected by the first photoelectric conversion element, and the edge of the circuit pattern which is the background on the semiconductor substrate is detected by epi-illumination. Foreign matter detection signal by emphasizing and detecting with the second photoelectric conversion element and dividing the foreign matter detection signal obtained from the first photoelectric conversion element with the detection signal obtained from the second photoelectric conversion element. It is described that the above-mentioned foreign matter is detected by emphasizing.

【0003】

Further, as a conventional technique for separating and inspecting a foreign substance adhering to the surface of a silicon wafer and a crystal defect existing on the surface, Japanese Patent Application Laid-Open No. 9-304289 (Prior Technique 3) is known. That is, the prior art 3 has a low angle light receiving system having an elevation angle of 30 ° or less with respect to the surface of the silicon wafer and a high angle light receiving system having an elevation angle larger than this, and emits laser light. The low-angle light-receiving system and the high-angle light-receiving system receive the scattered light obtained by irradiating the surface of the silicon wafer substantially perpendicularly, and those that are received only by the high-angle light-receiving system are crystal defects. It is described that those received by the low-angle light receiving system and the high-angle light receiving system are discriminated as adhering foreign substances and inspected.

【0004】

Further, as a conventional technique for distinguishing and inspecting foreign substances and scratches existing on the surface of a semiconductor wafer without misidentifying minute point-like recesses that do not hinder the creation of a circuit pattern, JP-A-11-142127 is published. (Previous technique 4) is known. That is, in the prior art 4, each of two different wavelength illumination lights is focused and irradiated to the same point on the surface of the semiconductor wafer at different low incident angles and high incident angles, and the collection thereof. The scattered light from the light spot is received separately for two wavelengths and photoelectrically converted, and the difference in intensity of each signal, that is, the scattered light intensity due to the illumination light with a low incident angle is weakened from the point-shaped recess. Therefore, it is described that foreign matter or scratches existing on the surface of the semiconductor wafer are inspected separately from the punctate recesses.

【0005】

[Problems to be Solved by the Invention]

By the way, CMP (Chemical Mechanical) is a typical flattening technology used for an object to be processed (for example, an insulating film) in semiconductor manufacturing or magnetic head manufacturing. Polishing). This CMP is a flattening technique in which free abrasive grains such as silica are sprayed on a polishing pad to polish the surface of the object to be processed. Further, as the flattening processing technology, a grinding processing technology in which fixed abrasive grains such as diamonds are embedded in a polishing pad and grinding processing is performed in the same manner may be used. In these polishing or grinding techniques, after polishing or grinding, scratches having various shapes, which are polishing or grinding scratches, may be generated on the surface of an object to be processed (for example, an insulating film on a semiconductor substrate (wafer)). is there. As described above, in semiconductor manufacturing and magnetic head manufacturing, if scratches having various shapes are generated on the surface of the object to be processed, etching becomes insufficient in the wiring formed on the scratches, which causes defects such as short circuits. Therefore, the polishing or grinding surface of the wafer after polishing or grinding is observed to monitor the occurrence of scratches having various shapes, and if it occurs frequently, the polishing or grinding conditions are reviewed according to the shape of the scratches. There must be. At the same time, if foreign matter is also generated, it causes defects such as insulation failure and short circuit of the wiring formed on the foreign matter. If foreign matter occurs frequently, it is necessary to take measures different from scratching, such as cleaning the device. In other words, in the polishing or grinding process for the object to be processed (for example, the insulating film on the semiconductor substrate), it is possible to separately monitor foreign matter and scratches having various shapes and take appropriate measures for each. You will need it.

【0006】

However, in any of the above-mentioned prior arts 1 to 4, when the object to be processed (for example, the insulating film on the semiconductor substrate) is polished or ground, scratches having various shapes generated on the surface thereof are formed. No consideration is given to discriminating and inspecting adhering particulate foreign matter.

【0007】

In addition, the dimensions of scratches having various shapes are as small as 0.2 μm to 0.4 μm in width W and about 100 nm even if the depth D is very deep from several nm. An operator visually reviews using a microscope to discriminate scratches and foreign substances having various shapes, which requires a large amount of review time. As a result, countermeasures against scratches or particulate foreign matter are delayed, and a large amount of wafers continue to be polished in poor conditions, resulting in great damage to profits.

【0008】

An object of the present invention is that when an object to be processed (for example, an insulating film on a semiconductor substrate) is polished or ground by CMP or the like in semiconductor manufacturing or magnetic head manufacturing in order to solve the above problems. It is an object of the present invention to provide a defect inspection apparatus and a method thereof capable of discriminating and inspecting scratches and the like having various shapes generated on the surface and particulate foreign matter adhering to the surface.

【0009】

Another object of the present invention is a scratch having various shapes generated on the surface of an object to be processed (for example, an insulating film on a semiconductor substrate) when it is polished or ground by CMP or the like. A semiconductor substrate that does not have the above-mentioned defects can be manufactured with high reliability and efficiency by making it possible to perform 100% inspection or sampling inspection at a sufficient frequency to discriminate between and the adhered particulate foreign matter. It is an object of the present invention to provide the manufacturing method of the semiconductor substrate.

【0010】

Another object of the present invention is to discriminate and inspect concave defects such as thin film-like foreign substances and scratches having a small height and depth from convex defects of high particulate foreign substances. Therefore, it is an object of the present invention to provide a method for manufacturing a semiconductor substrate, which enables the semiconductor substrate to be manufactured without the above-mentioned defects with high reliability and efficiency.

【0011】

[Means for solving problems]

In order to achieve the above object, the present invention presents a stage on which an object to be inspected is placed and a portion on the surface of the object to be inspected placed on the stage in the normal direction or in the vicinity thereof with respect to the surface. An epi-illumination system that epi-illuminates illumination light consisting of UV light or DUV light from a direction with a desired light beam, and oblique illumination of illumination light consisting of UV light or DUV light at a location on the surface of the object to be inspected with a desired light beam. Of the illumination optical system having the oblique illumination system and the first reflected light generated from the epi-illuminated portion by the epi-illumination system of the illumination optical system, the light is directed toward a high angle with respect to the surface of the object to be inspected. Of the first high-angle scattered light and the second reflected light generated from the portion obliquely illuminated by the oblique illumination system of the illumination optical system, the second high-angle scattered light toward the high angle is collected. The first and second brightness signals (S (i)) that receive the first and second high-angle scattered light imaged by the high-angle imaging optical system and the high-angle imaging optical system that form an image. A detection optical system having a photoelectric conversion means for converting to T (i)), and a first brightness signal S (i) and a second brightness signal T (i) converted by the photoelectric conversion means of the detection optical system. ), A comparative determination unit that discriminates the defect i on the object to be inspected into a concave defect (scratch or thin film-like foreign matter) and a convex defect (particle-like foreign matter). This is a defect inspection device and a method thereof.

【0012】

Further, according to the present invention, in the epi-illumination system of the illumination optical system in the defect inspection device, the object to be inspected is not exposed to the epi-illumination light so as not to generate stray light from the high-angle condensing optical system. It is characterized in that it is configured to irradiate the surface.

【0013】

Further, in the present invention, the detection optical system in the defect inspection apparatus further shields a specific light image due to the first reflected light from the Fourier transform surface of the first reflected light emitted from the portion. It is characterized by having.

【0014】

Further, the present invention is characterized in that, in the comparison determination unit in the defect inspection apparatus, the correlation is a ratio (T (i) / S (i), S (i) / T (i)).

【0015】

Further, according to the present invention, in the comparison determination unit of the defect inspection device, data corresponding to the size of the defect calculated from the first luminance signal S (i) and the second luminance signal T (i) is obtained. Based on this, the concave defect is configured to be discriminated from scratch and a thin film foreign substance (in the present invention, the thin film foreign substance is also defined as a concave defect because the thickness is very thin).

【0016】

Further, according to the present invention, in the comparison determination unit of the defect inspection device, data corresponding to the size of the defect calculated from the first luminance signal S (i) and the second luminance signal T (i) is obtained. Based on this, it is characterized in that it is configured to discriminate between large and small particulate foreign substances, which are convex defects.

【0017】

Further, the present invention is configured so that the comparison determination unit in the defect inspection apparatus can recognize whether the discriminated convex defect is generated in the circuit pattern region or outside the circuit pattern region. It is characterized by that.

【0018】

Further, the present invention is characterized in that the comparison determination unit in the defect inspection apparatus includes a display means for displaying information on the discriminated defects.

【0019】

Further, the present invention is characterized in that the comparison determination unit in the defect inspection apparatus includes a display means for displaying information regarding the relationship of the first luminance signal for discriminating defects.

【0020】

Further, the present invention is characterized in that the comparison determination unit in the defect inspection apparatus includes a display means for displaying information regarding the relationship of the second luminance signal for discriminating defects.

【0021】

Further, in the present invention, the comparison determination unit in the defect inspection apparatus is provided with the relationship between the first luminance signal and the second luminance signal converted by the photoelectric conversion means of the detection optical system on the horizontal axis and the vertical axis. It is characterized by having a display means for plotting and displaying on a correlation diagram represented by a logarithmic value.

【0022】

Further, according to the present invention, in the illumination optical system of the defect inspection device, a portion on the surface of the object to be inspected that is epi-illuminated by the epi-illumination system and a portion that is obliquely illuminated by the oblique illumination system are detected within the field of view of the optical system. It is characterized by being configured differently in.

【0023】

Further, in the present invention, the stage on which the object to be inspected is placed and a portion on the surface of the object to be inspected placed on the stage are exposed to UV light or DUV from the normal direction or a direction in the vicinity thereof with respect to the surface. An epi-illumination system that epi-illuminates illumination light consisting of light with a desired light beam and an illumination light having a wavelength different from that of the illumination light consisting of UV light or DUV light at a location on the surface of the object to be inspected at a desired light beam. An illumination optical system having an oblique illumination system to illuminate, and a first reflected light generated from a portion illuminated by the epi-illumination system of the illumination optical system at a high angle with respect to the surface of the object to be inspected. Of the first high-angle scattered light toward the high angle and the second reflected light generated from the portion obliquely illuminated by the oblique illumination system of the illumination optical system, the second high-angle scattered light toward the high angle is collected. The condensing optical system and the first high-angle scattered light and the second high-angle scattered light condensed by the condensing optical system are separated by a wavelength-separating optical system and the wavelength-separating optical system. An imaging optical system that forms an image of each of the first high-angle scattered light and the second high-angle scattered light, and a first high-angle scattered light and a second high-angle image formed by the imaging optical system. A detection optical system having first and second photoelectric conversion means that receives each of the scattered light and converts each into a first brightness signal and a second brightness signal, and a first detection optical system. A comparison determination unit that discriminates defects on the object to be inspected based on the relationship between the first brightness signal converted by the photoelectric conversion means and the second brightness signal converted by the second photoelectric conversion means. It is a defect inspection apparatus and a method thereof, which is characterized by being provided with.

【0024】

Further, the present invention performs epi-illumination and oblique illumination with substantially the same luminous flux on shallow scratches and foreign substances generated on the surface of a polished or ground film by illuminating light consisting of UV light or DUV light with substantially the same luminous flux. Scattered light generated from shallow scratches and foreign matter caused by illumination and oblique illumination is received by the detector and converted into a luminance signal according to the intensity of each scattered light, and based on the correlation of these converted luminance signals. This is a defect inspection method characterized by discriminating between shallow scratches and foreign matter.

【0025】

Further, the present invention provides epi-illumination and oblique illumination with substantially the same luminous flux for illumination light composed of UV light or DUV light against flat thin-film foreign matter or foreign matter generated on the surface of a polished, washed or sputtering film. Is performed, and the scattered light generated from the thin-film foreign matter and the foreign matter due to the epi-illumination and oblique illumination is received by the detector and converted into a brightness signal according to the intensity of each scattered light, and these converted brightness signals are obtained. This is a defect inspection method characterized by discriminating between thin-film foreign matter and foreign matter based on the correlation between the above.

【0026】

BEST MODE FOR CARRYING OUT THE INVENTION

A defect inspection apparatus for stable operation of the flattening processing process used in the semiconductor manufacturing process, the magnetic head manufacturing process, and the like according to the present invention and an embodiment of the method will be described with reference to the drawings.

【0027】

First, a defect inspection apparatus according to the present invention and a first embodiment of the method thereof will be described. As shown in FIG. 1, the present invention relates to a defect inspection device 100 that performs sampling or 100% inspection of products in the middle of a semiconductor manufacturing process. In the semiconductor production line, the production conditions are controlled by the computer 101 for process control, for example, via the network 103, or for each individual production apparatus (not shown). In the middle of the process, semiconductors are inspected by foreign matter inspection equipment, optical appearance inspection equipment, SEM inspection equipment, or manually. If an abnormality is found by inspection, it is reviewed with an optical review device, SEM review device, etc., and in some cases EDX (energy dispersive X-ray: energy dispersive X-ray) Perform a more detailed analysis using spectroscopy) to find out the cause of the abnormality. After that, the yield is improved by taking measures against the manufacturing conditions and manufacturing equipment that caused the abnormality. Further, data such as coordinates and dimensions of foreign matter and defects detected by the defect inspection device 100, and data such as types and categories of discriminated defects are managed online by the yield management system 102.

【0028】

As shown in FIG. 2, the defect inspection apparatus 100 according to the present invention has SiO on the Si wafer 21.₂When an interlayer insulating film (object to be processed) 22 such as a film is formed and CMP (Chemical Mechanical Polishing) is applied, the shallow scratch 23a generated on the wafer 10 and the foreign matter 24 are discriminated from each other. By the way, SiO₂Under the interlayer insulating film 22 such as a film, there is not necessarily a semiconductor substrate 21 such as a Si substrate, and a wiring layer may be present. In the CMP process, this SiO₂Polishing is performed to flatten the surface of the film 22. Therefore, the scratch 23a, which is a polishing scratch, is a SiO as shown in FIG. 2 (b).₂It occurs on the surface of film 22. Here, SiO₂Let t be the film thickness of the film 22, W be the width of the scratch 23a, and D be the depth. The approximate dimensions of the scratch 23a are such that W is about 0.2 μm to 0.4 μm. Further, the depth D is about several nm to about 100 nm even for a very deep object. As described above, the scratch 23a generated by CMP is characterized in that the depth is very shallow with respect to the width. Figure 2 (a) shows the dimensional parameters of the foreign matter 24. Here, the foreign matter 24 is modeled as a granular object having a diameter of Φ. The actual foreign matter 24 is not such a beautiful spherical shape, but the scratch 23a is very shallow with a depth D of several nm to several + nm with respect to the width W (about 0.2 μm to 0.4 μm), but the foreign matter (foreign matter (about 0.2 μm to 0.4 μm)). Particulate foreign matter) 24 shows that there is not an extremely large difference in width and height as in scratch 23a. The present invention focuses on the unique dimensional ratio of the scratch 23a.

【0029】

Next, a first embodiment of a surface inspection device such as a scratch for realizing the first embodiment will be described with reference to FIGS. 1 to 9. That is, in the first embodiment of the surface inspection device, as shown in FIG. 1, the position coordinates are measured, the traveling is controlled in the XY direction, and the stage 15 on which the wafer 10 to be inspected is placed, for example, Multiple light sources 2a consisting of light sources such as Ar laser, nitrogen laser, He-Cd laser, and excima laser with a wavelength of 488 nm (blue wavelength) (not limited to laser light sources) and outputting light of different wavelengths. , 2b, and an illumination optical system 1 composed of reflection mirrors 4a, 4b, 4c, a condenser lens 6, a beam splitter 7 that separates by wavelength, a photomal, a CCD camera, a CCD sensor, a TDI sensor, etc. A / D converters 16a and 16b that convert analog brightness signals output from each of the detection optical system 5 composed of photoelectric converters 8a and 8b and photoelectric converters 8a and 8b into digital brightness signals, and the A / Based on the arithmetic processing unit 9 composed of the storage units 17a and 17b and the comparison arithmetic unit 18 that temporarily store the digital brightness signals obtained from each of the D conversion units 16a and 16b, and the position coordinates measured from the stage 15. It is composed of a stage controller 14 that controls the traveling of the stage 15 and an overall control unit 30 that controls the stage controller 14 and further controls the arithmetic processing unit 9 and receives the inspection result obtained from the arithmetic processing unit 9. As the light sources 2a and 2b, it is preferable that the wavelength is as short as possible like an excimer laser light source in order to discriminate and detect minute foreign matter 24 and scratch 23 generated on the CMP insulating film 22. That is, the light source 2a is, for example, a laser light source that emits a laser beam of 488 nm or 365 nm, and the light source 2b is a DUV laser beam of a double wave (532 nm) or a fourth harmonic wave (266 nm) of a YAG laser, or a KrF excimer laser. It can be configured with a laser light source that emits light. Then, the UV light or DUV light emitted from the light source 2a is directly applied to the surface of the condenser lens 6. Instead, the wafer surface (the surface of the insulating film to which CMP is applied) is irradiated from the normal direction or its vicinity via the reflection mirror 4a and the reflection mirror 4c. This is called epi-illumination 12. Alternatively, the UV light or DUV light emitted from the light source 2b irradiates the wafer surface (the surface of the insulating film to which CMP is applied) from an oblique direction via the reflection mirror 4b. This is referred to as oblique illumination 11. In the first embodiment, epi-illumination and oblique illumination are realized by using two independent light sources 2a and 2b and a plurality of reflection mirrors 4a to 4c, respectively, but one light source 2b Then, an optical path switching mechanism (not shown) that switches the optical path of the UV light or DUV light emitted from the light source 2b between the mirror 4b and the mirror 4c may be used. In addition, the number of reflection mirrors and the presence or absence of an optical path switching mechanism do not matter.

【0030】

Further, in the illumination optical system 1, a location on the wafer surface where the epi-illumination light 12 is epi-illuminated by the epi-illumination system and a location on the wafer surface where the oblique illumination light 11 is obliquely illuminated by the oblique illumination system are detected. The wavelengths of the epi-illumination light 12 and the oblique illumination light 11 can be made the same by making them different in the field of view of the system 5. However, in this case, it is necessary to install the light receiving surfaces of the photoelectric converter 8a and the photoelectric converter 8b so as to correspond to the difference in the irradiation location on the wafer surface.

【0031】

As described above, the illumination optical system 1 does not directly irradiate the surface of the condenser lens 6, but is in the normal direction or in the normal direction with respect to the CMP surface in which the insulating film 22 on the wafer 10 is CMP-applied. It suffices if two systems of illumination (epi-illumination and oblique illumination) 11 and 12 are realized from a near direction and an oblique direction (an angle of about 30 ° or less) close to the horizontal plane of the wafer. In the case of the epi-illumination 11, as shown in FIG. 1, the pseudo epi-illumination may be as close to the vertical direction as possible.

【0032】

Next, the detection procedure will be described. The detection is performed twice for one wafer 10 by switching the illumination direction. Specifically, first, the CMP of the insulating film 22 on the wafer 10 without directly irradiating the surface of the condenser lens 6 with the epi-illumination light 12 composed of UV light or DUV light emitted from the light source 2a. Irradiate the surface. Then, the positively reflected light component generated from the insulating film 22 was removed without generating stray light reflected from the fine surface roughness of the surface of the condenser lens 6 and the extremely fine foreign matter adhering to the surface. In this state, only the ultra-shallow fine scratches 23a generated by the CMP on the insulating film 22 and the scattered light (low-order diffracted light component) emitted from the foreign matter 24 are collected by the condenser lens 6 and passed through the beam splitter 7. For example, light is received on the light receiving surface of the photoelectric converter 8a composed of a CCD, a TDI sensor, or the like. Then, the output of the photoelectric converter 8a is A / D converted by the A / D conversion unit 16a to obtain the luminance value S (i) for each defect i, and then is temporarily written to the storage unit 17a.

【0033】

At the same time, oblique illumination light 11 emitted from the light source 2b and composed of UV light or DUV light having a wavelength different from that of the light source 2a is irradiated to the same coordinate position as the epi-illumination light 12 on the wafer surface. By controlling the movement of the stage 15, the overall control unit 30 switches the irradiation direction using an optical path switching mechanism (not shown) to make the oblique illumination light 11 the same as the epi-illumination light 12 on the wafer surface. Irradiation may be performed in the position coordinate system.

【0034】

Then, with the specularly reflected light component generated from the insulating film 22 removed, the scattered light (low) emitted from the ultra-shallow fine scratch 23a generated by the CMP and the foreign matter (particulate foreign matter) 24 on the insulating film 22. Only the next diffracted light component) is focused by the condenser lens 6 and received through the beam splitter 7 by, for example, the photoelectric converter 7b. Then, the output of the photoelectric converter 7b is A / D converted by the A / D conversion unit 16b to obtain the luminance value T (i) for each defect i, and then is temporarily written to the storage unit 17b.

【0035】

Next, the comparison calculation unit 18 detects the detected luminance value S (i) for each defect i by the epi-illumination 12 stored in the storage unit 17a and the detection for each defect i by the oblique illumination 11 stored in the storage unit 17b. Calculate the ratio R (i) to the brightness value T (i). If the calculated luminance ratio R (i) is larger than the preset threshold value (judgment reference value: discrimination line 20 shown in FIG. 5), the comparison calculation unit 18 determines that the foreign matter 24, and if it is smaller, the extremely shallow fine scratch 23a. Determine and output to the overall control unit 9. As described above, since the scratch 23a generated by the CMP is extremely shallow and fine, when the surface of the condenser lens 6 is irradiated with the epi-illumination light 12, the weak stray light generated from the surface of the condenser lens 6 is also, for example. When the light is received by the photoelectric converter 7a, it becomes difficult to distinguish it from the scattered light from the scratch 23a. Therefore, the surface of the condenser lens 6 is not irradiated with the epi-illumination light 12.

【0036】

In the first embodiment, the detection by the epi-illumination light 12 and the detection by the oblique illumination light 11 are performed at the same time, but the detection by the epi-illumination light 12 is performed first, and the detection by the oblique illumination light 11 is performed later. However, the detection by the oblique illumination light 11 may be performed first, and the detection by the epi-illumination light 12 may be performed later. Further, in the first embodiment, the detected luminance value T (i) by the oblique illumination 11 which is the second detection is once written in the storage unit 17 after A / D conversion, and the second detected luminance value T ( The present invention is realized even if the comparison calculation unit 18 refers to the detection brightness value S (i) by the first epi-illumination 12 already stored at the same time as the detection without storing i) and performs the brightness comparison calculation. It is possible to do.

【0037】

Next, the discrimination principle for realizing the above-described embodiment according to the present invention will be described in detail with reference to FIGS. 3 and 4. In the present invention, one defect is discriminated by irradiating it with a luminous flux d from two different angles (for example, epi-illumination 12 and oblique illumination 11). First, as the epi-illumination light 12, the surface of the condenser lens 6 is not directly irradiated, but is irradiated with a luminous flux d from the normal direction of the wafer surface or its vicinity. Next, the oblique illumination light 11 is irradiated with the luminous flux d from an angle close to the horizontal direction with respect to the wafer surface. It does not matter which of the epi-illumination 12 and the oblique illumination 11 is performed first. Discrimination is performed by comparing the intensities of scattered light emitted from defects 23a and 24 obtained in the two-way illumination of the luminous flux d, respectively. The scattered light intensity from the defects 23a and 24 is emitted according to the amount of light source light received by the defects 23a and 24. As shown in FIG. 3, it can be considered that the amount of light source light received by the defects 23a and 24 is substantially proportional to the projected area of the defect dimension in the light source incident direction.

【0038】

In the case of scratch 23a, which is a concave defect, this projected area is approximately proportional to the width W during epi-illumination, and is approximately proportional to D'when illuminated at a shallow angle of about 30 ° or less. However, since the depth D of the scratch 23a is very shallow compared to the width W, this oblique illumination projection length D'is much shorter than the epi-illumination projection length W'. Therefore, the amount of light source light received by the scratch 23a is weaker in the oblique illumination 11 than in the epi-illumination 12, and as a result, the amount of scattered light emitted from the scratch 23a is weaker in the oblique illumination 11. .. On the other hand, in the case of the foreign matter (particulate foreign matter) 24 which is a convex defect, the projected lengths Φ of the oblique illumination 11 and the epi-illumination 12 are almost the same, so that the amount of scattered light emitted from the foreign matter 24 is , Even if you compare epi-illumination and oblique illumination, there is no big difference. Therefore, as shown in FIG. 4, the oblique illumination 11 is the epi-illumination by comparing the detected luminance values S (i) and T (i) of the scattered light by the epi-illumination 12 and the oblique illumination 11, respectively. If it is smaller than 12, it is possible to discriminate a scratch 23a, and an object having the same or larger oblique illumination as a foreign substance (particulate foreign substance) 24.

【0039】

Further, since the thin film foreign matter 23b is also very thin, the detected brightness value T (i) of the scattered light by the oblique illumination 11 is the detected brightness value S (i) of the scattered light by the epi-illumination 12 as in the scratch 23a. ) Is detected and can be regarded as a concave defect.

【0040】

A graph of an example of this discrimination result is shown in FIG. This is a graph in which the horizontal axis represents the detected luminance value S (i) during epi-illumination and the vertical axis represents the detected luminance value T (i) during oblique illumination. In this case, the region below the discrimination line 20 in the figure is the region of the scratch 23a, and the region above is the region of the foreign matter 24. However, as is clear from FIG. 5, in practice, in this way, the detected luminance value S (i) during epi-illumination is simply compared with the detected luminance value T (i) during oblique illumination. Even if the ratio is taken, the discrimination line (judgment threshold value) 20 cannot be drawn (set), and it is difficult to discriminate between the foreign matter 24 and the scratch 23a. Therefore, using the detected luminance value S (i) at the time of epi-illumination and the detected luminance value T (i) at the time of oblique illumination according to the present invention, specifically, the foreign matter (particulate foreign matter) 24 and the scratch 23a An example of the method of discriminating the above will be described later.

【0041】

By the way, an insulating film (for example, SiO) in which scratch 23a is generated by CMP.₂Since the film) 22 is transparent to light, positively reflected light from the lower layer is generated including light interference. In particular, in the case of epi-illumination 12, for example, as shown in FIG. 1, a reflective mirror 4c is used. By installing it outside the field of view of the condenser lens 6, the positively reflected light (including light interference light) from the surface of the insulating film 22 and its lower layer goes out of the field of view of the condenser lens (objective lens) 6. For example, it is necessary to devise a way to prevent detection by the photoelectric converter 8a.

【0042】

Of course, in the case of oblique illumination 11, as shown in FIG. 1, since the reflection mirror 4b irradiates the light at a very shallow angle, the specularly reflected light (including light interference light) from the surface of the insulating film 22 and its lower layer is included. ) Goes out of the field of view of the condenser lens 6, and is not detected by the photoelectric converter 8b.

【0043】

Further, if a light source 2 that emits broadband light or white light is used, the problem of optical interference between the specularly reflected light from the surface of the insulating film 22 and the specularly reflected light from the lower layer does not occur. However, in order to obtain strong scattered light from the fine (particularly shallow depth D) scratch 23a and the foreign matter 24 on the insulating film 22, it is preferable to use UV light or DUV light as the illumination light.

【0044】

Next, an example of the installation method of the reflection mirror 4c will be described with reference to FIG. This is a method for preventing stray light in the dark field detection system and detecting defects with high sensitivity. The inspection of the scratch 23a requires illumination from a direction close to the normal with respect to the surface of the wafer 10, as can be seen from the principle described above.

【0045】

However, when UV light or DUV light is epi-illuminated, if the epi-illuminated light is transmitted through the condenser lens 6 to illuminate the wafer 10, so-called stray light is generated, and as a result, noise is generated in the detected image. It will end up. Specifically, this is because scattered light generated from fine polishing marks on the surface of the condenser lens 6 and dust adhering on the condenser lens 6 becomes stray light. Therefore, this stray light becomes fatal when the minute scattered light from the defects 23a and 24 is received by the photoelectric converter 8a and observed. That is, the scattered light from the extremely small scratch 23a is buried in the noise caused by the stray light and cannot be detected.

【0046】

Therefore, in the present invention, as shown in FIG. 6, the surface of the condenser lens 6 is not irradiated with strong incident light, and the wafer 10 (the surface (CMP surface) of the interlayer insulating film 22 and its lower layer) is not irradiated. Prevent the 0th-order diffracted light, which is a specularly reflected light component (including the interference light component) from the surface of the wiring layer and the surface of the scratch 23a and the surface of the foreign matter 24, from entering the pupil of the condenser lens 6, that is, the NA. It is necessary to provide a reflection mirror 4c.

【0047】

In FIG. 6A, a small reflection mirror 4c1 is placed between the wafer 10 and the condenser lens 6 approximately on the normal line of the wafer 10, so that the epi-illumination light 12a is not irradiated on the surface of the condenser lens 6. It is incident on a small reflection mirror 4c1 from the lateral direction and reflected, and the positive reflected light component (including the interference light component) from the wafer 10 is reflected by the reflection mirror 4c1 into the pupil of the condenser lens 6. Of the scattered light (first-order or higher-order diffracted light component) from scratch 23a or foreign matter 24 without incident light, the scattered light (lower-order diffracted light component) in the area indicated by the diagonal line (planar ring-shaped) is collected. The method of making the light incident into the pupil of the optical lens 6 is shown. The outer shape of this small reflection mirror 4c1 is almost elliptical. This is referred to as scattered light detection by vertical illumination. However, this method is not very preferable because the central portion of the condenser lens 6 loses its role as a lens.

【0048】

Further, in FIG. 6B, the reflection mirror 4c2 is arranged between the wafer 10 and the condenser lens 6 and outside from the NA of the condenser lens 6, and the epi-illumination light 12b is placed on the surface of the condenser lens 6. It is incident on the reflection mirror 4c2 from the lateral direction so as not to be irradiated to the light, and is reflected, and the positively reflected light component from the wafer 10 is outside the pupil of the condenser lens 6 to be included in the scattered light from the scratch 23a and the foreign matter 24. A method of incident the scattered light in the area indicated by the diagonal line into the pupil of the condenser lens 6 is shown. When the reflection mirror 4c2 is expanded in the circumferential direction, the illumination light illuminated by the reflection mirror 4c2 becomes ring-shaped illumination. In this case, for example, if three reflection mirrors 4c2 are provided at intervals of 120 degrees in the circumferential direction and each of the three illumination lights 12b is incident from between each of the three reflection mirrors 2c2, the ring is formed from the three directions. It is also possible to apply band lighting. However, as shown in FIG. 6 (b), when the reflection mirror 4c2 is made a part, it becomes a part of the illumination in the annular illumination. This is referred to as scattered light detection by pseudo vertical illumination. This method is very effective because the entire field of view (pupil) of the condenser lens 6 is used. However, it is necessary to match the luminous flux of the oblique illumination light 11 with the luminous flux of the epi-illumination light 12b, including the number of points.

【0049】

Further, in FIG. 6 (c), a small reflective mirror or a half mirror 4c3 is arranged near the optical axis above the condenser lens 6, and a condenser lens 6 having an opening 50 bored in the center is arranged. The vertical illumination light 12a reflected by the reflection mirror or the half mirror 4c3 of the above is not irradiated to the surface of the condenser lens 6, but is passed through the opening 50 to irradiate the insulating film CMP surface on the wafer 10 and also from the wafer 10. The positively reflected light component of the above is shielded by a spatial filter (light-shielding element) 51 provided on the Fourier conversion surface, and the scattered light obtained through the internal condensing lens 6 of the scattered light from the scratch 23a and the foreign matter 24 is transmitted by the photoelectric converter 8a. The method of receiving light is shown.

【0050】

Further, in FIG. 6 (d), as in FIG. 6 (c), the epi-illumination light 12a is transmitted through the central portion of the half mirror 52 and is perpendicular to the CMP surface of the wafer 10 through the opening 50 of the condenser lens 6. After illuminating, the positively reflected light from the wafer 10 is shielded by a spatial filter (light-shielding element) 53 provided on the Fourier conversion surface, and the scattered light obtained through the internal condensing lens 6 of the scattered light from the scratch 23a and the foreign matter 24 is emitted. A method of reflecting light at the peripheral portion of the half mirror 52 and receiving light with the photoelectric converter 8a is shown. The peripheral portion of the half mirror 52 may be composed of a reflective mirror.

【0051】

As described above, in FIGS. 6 (c) and 6 (d), stray light is generated from the surface of the condenser lens 6 by forming an opening 50 in the center of the condenser lens 6 as in FIG. 6 (a). It enables vertical illumination and detection of scattered light from the vertical direction without the need for this. Therefore, no matter how the orientation of the scratch 23a is formed in the horizontal plane, the scattered light generated from the very shallow edge of the scratch 23a can be received by the photoelectric converter 8a relatively uniformly, and is uniform. The detected luminance value S (i) can be obtained. Further, vertical illumination is preferable to pseudo-vertical illumination in order to obtain diffracted light having strong directivity in the direction perpendicular to a large scratch (not shown) which is a linear pattern. However, the embodiment of FIGS. 6 (c) and 6 (d) is not so preferable because the function of the condenser lens 6 is deteriorated due to the formation of the opening 50 in the central portion of the condenser lens 6.

【0052】

By the way, in the case of the scattered light detection by the vertical illumination of FIG. 6A, the incident light passes under the lens 6, and the surface of the condenser lens 6 is clearly not irradiated, and no stray light is generated. Further, since the specularly reflected light from the wafer 10 is reflected by the reflection mirror 4c1, it does not enter the pupil of the condenser lens 6. Further, the same applies to the vertical illumination shown in FIGS. 6 (c) and 6 (d). Also, in the case of the scattered light detection by the pseudo vertical illumination shown in FIG. 6B, the incident light clearly does not pass through the condenser lens 6. Further, since the reflection mirror 4c2 is arranged outside the NA of the condenser lens 6, the specularly reflected light component from the wafer 10 does not enter the pupil of the condenser lens 6. In other words, epi-illumination is realized so that the incident light, which has strong light intensity and easily causes stray light, is not applied to the surface of the condenser lens 6 and the specular reflected light from the wafer is not incident on the condenser lens 6. doing. Therefore, stray light is less likely to occur, and it is possible to obtain a detection image having a high S / N ratio from the scratch 23a and the foreign matter 24 generated on the CMP surface where the interlayer insulating film 22 is CMP-applied. Since the interlayer insulating film 22 is transparent to light, the light reflected specularly from the lower layer returns when the epi-illumination is performed. However, as will be described next, the condenser lens (objective lens) 6 Since it is not incident in the NA of the lens, the scratch 23a and the foreign matter 24 can be detected by the signal obtained from the photoelectric converter 8a without affecting the detection of scattered light from the scratch 23a and the foreign matter 24.

【0053】

Further, the epi-illuminations 12a and 12b shown in FIG. 6 are more likely to receive the component having a strong scattered light intensity distribution from the scratch 23a, not only because of the solution of the stray light, so that the epi-illumination 12a and 12b are more easily received than the oblique illumination 11 alone. High detection sensitivity can be obtained. This is because the low-order diffracted light component is relatively strong among the scattered light intensities from the scratch 23a. That is, if the illumination is performed from the vicinity of the normal line of the wafer surface, the low-order diffracted light component is reflected from the wafer 10 and easily collected by the condenser lens 6. However, for example, it is necessary to completely block the specularly reflected light (0th-order diffracted light) from the base of the insulating layer or the surface of the insulating layer so that it is not incident on the pupil of the condenser lens 6.

【0054】

As a result, the scratch 23a can be detected with higher sensitivity than the case where only the oblique illumination 11 is used. In this way, by using only the vertical illumination 12a or the pseudo vertical illumination 12b, it is possible to realize a highly sensitive scratch 23a inspection.

【0055】

By the way, if the shape of the reflection mirror 4c1 is formed to be substantially elliptical so that the imaging characteristics of the lens 6 and the like are not affected even if the reflection mirror 4c1 is arranged in the NA of the condenser lens 6, FIG. The scattered light in the region indicated by the diagonal line in (a) (the annular region in the plane) can be condensed by the condenser lens 6 to form an image. However, if the presence of the reflection mirror 4c1 in the NA of the condenser lens 6 adversely affects the imaging characteristics, a mechanism for retracting the reflection mirror 4c1 out of the NA is required during vertical illumination. In the case of semiconductor inspection, it is necessary to eliminate dust generated from the defect inspection device as much as possible. From this point of view, it is not preferable to provide the movable mechanism above the wafer. However, even in such a case, the pseudo vertical illumination 12b may be used. In the case of the pseudo vertical illumination 12b, since the reflection mirror 4c2 is outside the NA, it never adversely affects the imaging characteristics, and there is no need to provide a separate evacuation mechanism.

【0056】

Further, when the surface inspection device such as a scratch according to the present invention is used as a foreign matter inspection device using only oblique illumination, vertical illumination becomes unnecessary, so that the reflection mirror 4c1 shown in FIG. 6A is retracted. It is also possible to effectively collect the scattered light generated from the foreign matter by using all of the NA of the condenser lens 6 and receive it with the photoelectric converter 8b. However, in order to prevent the reflection mirror 4c1 from retracting and to eliminate the generation of dust, it is sufficient to use the pseudo vertical illumination 12b as the vertical illumination of the surface inspection device, which slightly lowers the scratch detection accuracy. In addition, when the methods shown in FIGS. 6 (c) and 6 (d) are used as vertical illumination, it can be applied by stopping the vertical illumination even when used as a foreign matter inspection device using only oblique illumination. It becomes. Further, when used as a foreign matter inspection device using only oblique illumination, when trying to detect a foreign matter on a memory cell in which a periodic wiring pattern is formed, the diffraction pattern based on the diffracted light from the periodic wiring pattern is shielded. Therefore, the space filters 51 and 53 may be replaced with linear space filters.

【0057】

Next, in the comparison calculation unit 18 and the like, the luminance signal S (i) for each defect i due to the epi-illumination 12 and the luminance signal T (i) for each defect i due to the oblique illumination 11 stored in the storage units 17a and 17b are stored. Based on this, a method for estimating the size of defects will be described with reference to FIG. 7 (a) and 7 (b) show the waveforms 301 and 302 of the luminance signals S (i) and T (i) detected from each of the foreign matter 24, the scratch 23a and the thin film foreign matter 23b. The luminance signal waveform 302 shown in FIG. 7 (b) is squeezed at the level of 303 as shown in FIG. 7 (c) due to the dynamic range of the detectors (photoelectric converters) 8a and 8b. Therefore, the volume is obtained by interpolating based on the plot point 304 and integrating the interpolated signal waveform in two dimensions. Since the luminance signal shown in FIG. 7A is not specified, the volume is obtained by integrating the luminance signal waveform in two dimensions as it is. Since there is a correlation between these obtained volume values (two-dimensional integral values) and the defect size, it is possible to obtain estimated data according to the defect size by multiplying this correction coefficient.

【0058】

FIG. 8 shows the foreign matter size (μm) estimated as the estimated data for the foreign matter 2901 by, for example, the comparison calculation unit 18 of the inspection device according to the present invention, and the size (μm) actually measured by SEM. Show the relationship. As shown in FIG. 8, a plurality of process processing (for example, CMP) steps on the wafer 10 have different correlations as shown in 2902 and 2903. Therefore, the correction coefficient changes depending on the surface condition of the wafer. Therefore, it is necessary to obtain the correction coefficient according to the process process (surface condition of the wafer) in advance based on the SEM length measurement.

【0059】

In addition, Fig. 9 (a) shows the foreign matter size (μm) estimated from the brightness signal waveform and the SEM length-measured size for the defect 3101 in the front-end process wafer (wafer in the transistor formation process, which is the initial process). Correlation coefficient R at 3102 with (μm)²= 0.7945 shows that there is a correlation. As described above, in the transistor forming step, the minute defects of 0.1 μm to 0.4 μm affect the performance of the transistor, and thus it can be seen that there is a correlation even in such a minute defect. The correlation coefficient R is expressed by the following equation (Equation 1).

【0060】

R = (NΣx_iy_i-(Σx_i) (Σy_i)) / ( (NΣx)_i²-(Σx_i)²) (NΣy_i²-(Σy_i)²)) (Number 1) However, x and y indicate variables.

【0061】

In addition, Fig. 9 (b) shows the foreign matter size (μm) estimated from the brightness signal waveform and the SEM length-measured size for the defect 3101 in the back-end process wafer (wafer in the wiring formation process, which is the latter process). Correlation coefficient R at 3102 with (μm)²= 0.7147 shows that there is a correlation. As described above, in the wiring forming step, minute foreign substances of 0.3 μm or more and further up to about 5 μm affect the wiring, and thus it can be seen that there is a correlation even with such minute foreign substances. In the wiring process, minute foreign matters of 0.3 μm or less are eliminated because their importance decreases.

【0062】

Next, the defects to be inspected by the inspection apparatus according to the present invention will be described with reference to FIG. CMPed surface defects include convex defects 24 based on ordinary foreign matter (about 0.1 μm to 5 μm) and scratches (width W is about 0.2 μm to 0.4 μm, depth D is about several nm to several + nm. ), And a flat defect 23b to which a thin film-like foreign substance (diameter: about 0.5 μm to 2 μm, thickness: about several nm to several + nm) is attached.

【0063】

Furthermore, the correlation diagram (correlation coefficient R) between the size (μm) from the luminance signal and the size (μm) based on the SEM length measurement.²From = 0.3847), it was found that these convex defects 24 and the concave defects 23a and the flat defects 23b have different correlations.

【0064】

Furthermore, the size of the defect estimated from the luminance signals S (i) and T (i) detected by epi-illumination and / or oblique illumination of the concave defect 23a such as scratch and the flat defect 23b such as thin film foreign matter. It was found that it is possible to discriminate based on.

【0065】

Further, by discriminating whether the region where the convex defect 24 such as the foreign matter and the concave defect 23a such as the scratch are generated is within the circuit pattern region or outside the circuit pattern region, the convex shape of the foreign matter or the like is formed. It is possible to discriminate the fatality of the defect 24 and the concave defect 23a such as scratch to the circuit pattern.

【0066】

Therefore, these discrimination methods, which are calculated by the comparison calculation unit 18 and the like, will be described with reference to FIG. First, in step S111, the luminance signal S (i) for each defect i due to the epi-illumination 12 detected from the photoelectric converter 8a is stored in the storage unit 17a after A / D conversion by the A / D converter 16a. At the same time or thereafter, in step S112, the luminance signal T (i) for each defect i due to the oblique illumination 11 detected from the photoelectric converter 8b is A / D converted by the A / D converter 16b, and then stored in the storage unit 17b. Remember in. Then, in step S113, the comparison calculation unit 18 displays the luminance signal S (i) for each defect i detected by the epi-illumination stored in each of the storage units 17a and 17b and the luminance signal S (i) for each defect i detected by the oblique illumination. The degree of unevenness (b / a) shown in FIG. 12 is obtained by the following equation (Equation 2) with the ratio R (i) to the signal T (i).

【0067】

Note that FIG. 12 is a logarithmic table because both the horizontal axis and the vertical axis are shown in logarithms. In FIG. 12, the direction of the arrow 121 from the lower left to the upper right corresponds to the size of the defect, and the arrow 122 perpendicular to the arrow 121 is indicated by the degree of unevenness (b / a) of the defect. The degree of unevenness (b / a) of the defect is indicated by the ratio of the vertical dimension b to the horizontal dimension a shown in FIG. However, the degree of unevenness and the process are based on the discrimination of the degree of unevenness of defects based on the ratio of luminance signals (T (i) / S (i)) and the integrated value of the luminance signals (S (i), T (i)). Determining the size of a defect based on multiplying it by a corresponding correction factor does not necessarily require taking the logarithm of each luminance signal.

【0068】

R (i) = T (i) / S (i) = b / a (number 2) Here, i is a recognition number assigned to each defect in order to evaluate a plurality of defects. Since one defect may be detected as a plurality of defects depending on the size of the luminous flux d and the pixel size of the photoelectric converter 7, expansion processing (linkage processing) is performed on signals indicating defects detected in close proximity. ) Needs to be converted into a signal indicating one defect. Therefore, the identification number i assigned to each defect is assigned to the signal indicating one defect that has been concatenated.

【0069】

Further, in step S114, if the obtained luminance ratio R (i) is larger than the preset threshold value (judgment reference value: discrimination line 20 shown in FIG. 5), the comparison calculation unit 18 is convex such as particulate foreign matter. If it is small, it is determined to be a concave defect 24, and if it is small, it is determined to be a concave defect 23 such as a scratch or a thin film foreign substance. In this embodiment, the detected brightness T (i) during oblique illumination is divided by the detected brightness value S (i) during epi-illumination, but conversely, the detected brightness value S (i) during epi-illumination is used. ) May be divided by the detected brightness value T (i) during oblique illumination. In this case, if the ratio R (i) is larger than the preset threshold value (judgment reference value: discrimination line 20 shown in FIG. 5), it is a concave defect 23 such as a scratch or a thin film foreign substance, and if it is small, it is a convex defect such as a foreign substance. It can be discriminated from the defect 24.

【0070】

Next, in step S115, the overall control unit 30 data according to the size of the defect based on the unevenness (b / a) obtained in step S113 and the process information of the wafer 10 obtained from the process control computer 101. Calculate the correction factor for estimating.

【0071】

Next, in step S116, the luminance signal S (i) for each defect i detected by the epi-illumination stored in each of the storage units 17a and 17b by the comparison calculation unit 18 and the overall control unit 30, and the oblique illumination. Based on the luminance signal T (i) for each defect i detected by the above, each luminance signal waveform is two-dimensionally integrated to obtain a volume value, which is calculated by the overall control unit 30 of the wafer. By multiplying the surface condition (which can be obtained as process information from the process control computer 101) and the correction coefficient that matches the unevenness (b / a), the estimated size of the defect (estimated according to the size). Data) (μm) is calculated.

【0072】

Next, in step S117, the concave defect 23 is referred to as scratch 23a as shown in FIG. 12 based on the data corresponding to the defect size estimated in step S115 by the comparison calculation unit 18 and the overall control unit 30. It can be discriminated from the thin film foreign matter 23b. As shown in FIG. 12, the waveform of the luminance signal S (i) due to epi-illumination is integrated two-dimensionally to obtain the volume value, and the obtained volume value is multiplied by the above correction coefficient (depending on the size). It is also possible to discriminate between scratch 23a and thin-film foreign matter 23b based on the estimated data).

【0073】

As described above, it is possible to discriminate between the scratch 23a, which is a concave defect, and the thin film foreign matter 23b.

【0074】

Next, the overall control unit 30 can discriminate the ordinary particulate foreign matter 24 as the convex defect discriminated in step S114. Further, when it is necessary to discriminate the discriminated particulate foreign matter into large and small particles, in step S118, by using the size estimated value (estimated data according to the size) of the foreign matter obtained from step S116, FIG. 12 shows. It is possible to discriminate between large and small from the relationship shown.

【0075】

Further, in step S119, in the comparison calculation unit 18, the overall control unit 30, and the like, as shown in FIG. 14, for the small-sized particulate foreign matter 24 and the scratch 23a, the network 103 is connected from the CAD system (not shown). The particulate foreign matter 24 and the scratch 23a are generated based on the arrangement information of the circuit pattern on the wafer 10 obtained through the circuit pattern or the arrangement information of the circuit pattern obtained based on the image signal of the circuit pattern detected by the detectors 8a and 8b. By discriminating whether it occurred on the circuit pattern or outside the circuit pattern, it is possible to determine the fatality of the foreign matter 24 or the scratch 23a with respect to the circuit pattern. That is, the case where the foreign matter 24 having a small defect size occurs on the circuit pattern is classified as category 1, and the case where the particulate foreign matter 24 occurs outside the circuit pattern is classified as category 2, and the scratch 23a having a small defect size is classified. When is generated on the circuit pattern, it is classified as category 3, when scratch 23a occurs outside the circuit pattern, it is classified as category 4, and when a large defect size, for example, a thin film-like foreign substance is generated, it is classified as category 5. can do. In this way, the overall control unit 30 can classify categories by process, at least in lot units, so that it can be useful not only for evaluating the fatality of defects but also for investigating the causes of defects. It will be possible. When the circuit pattern arrangement information is obtained based on the circuit pattern image signals detected by the detectors 8a and 8b, the luminance signals S (i) and T (i) for defects are detected by the detector 8a. By, for example, repeating chip comparison or die comparison of the image signal detected by 8b, the image signal of the repeated circuit pattern can be erased and extracted.

【0076】

Next, the overall control unit 30 displays the detection luminance value S (i) at the time of epi-illumination and the time of oblique illumination for the particulate foreign matter (indicated by ) and scratches (indicated by Δ) displayed on the display device 33. The correlation diagram with the detected luminance value T (i) of is shown in FIGS. 15 and 16. As shown in FIG. 16, both brightness values S (i) and T (i) are logarithmic, and as compared with the normal scale shown in FIG. 15, foreign matter 24 and scratch 23a and the like are displayed. Can be easily identified, and the threshold value (discrimination line 20) for discriminating between the two can be easily set on the screen. As shown in FIG. 16, when both the horizontal axis and the vertical axis are logarithmic, the correlation line 161 indicating the particulate foreign matter 24 and the correlation line 162 indicating the scratch 23a and the like are parallel lines.

【0077】

Next, the defect map displayed on the screen of the display device 33 by the overall control unit 30 will be described with reference to FIG. FIG. 17A shows a state in which the foreign matter map on the wafer in the predetermined CMP process, which is the inspection result discriminated in step S114 shown in FIG. 11, is displayed on the screen of the display device 33. Similarly, FIG. 17B shows a state in which the scratch map on the wafer in the predetermined CMP process, which is the inspection result discriminated in step S117 shown in FIG. 11, is displayed on the screen of the display device 33. Similarly, FIG. 17C shows a state in which the thin film foreign matter map on the wafer in the predetermined CMP process, which is the inspection result discriminated in step S117 shown in FIG. 11, is displayed on the screen of the display device 33. .. From each of these foreign matter maps, scratch maps, and thin-film foreign matter maps, it is possible to know the generation distribution of each of the particulate foreign matter, scratches, and thin-film foreign matter on the wafer.

【0078】

[Effect of the invention]

According to the present invention, in semiconductor manufacturing or magnetic head manufacturing, when an object to be processed such as an insulating film is polished or ground by CMP or the like, it adheres to scratches or the like having various shapes generated on the surface thereof. It has the effect of being able to discriminate from particulate foreign matter and inspect it.

【0079】

Further, according to the present invention, since the shape of the scratch can be classified in detail, it is possible to quickly identify the cause of the defect.

【0080】

Further, according to the present invention, since it is possible to perform sampling inspection of all or frequently in the flattening polishing process, it is possible to quickly find a defect in the polishing apparatus, and as a result, take appropriate measures. This makes it possible to dramatically improve the yield in the polishing process.

[Simple explanation of drawings]

[Figure 1]

It is a schematic block diagram which shows the 1st Embodiment of the defect inspection apparatus which concerns on this invention.

[Figure 2]

It is a figure which shows the shape parameter of the scratch and the foreign matter generated on the insulating film by CMP or the like which concerns on this invention.

[Fig. 3]

It is a figure for demonstrating the incident light projection length at the time of irradiating a scratch and a foreign substance with a light flux d which concerns on this invention.

[Fig. 4]

It is a figure which shows the discriminating principle between a scratch and a particulate foreign matter which concerns on this invention.

[Fig. 5]

It is a figure which shows one Example of the discrimination result between the scratch and the particulate foreign matter which concerns on this invention.

[Fig. 6]

It is a figure which shows the example of the vertical irradiation and pseudo vertical illumination which concerns on this invention.

[Fig. 7]

It is a figure which shows the luminance signal waveform detected by the detection optical system which concerns on this invention.

[Fig. 8]

It is a correlation diagram which shows the foreign matter generated on the CMP surface of the wafer which concerns on this invention by the correlation of the foreign matter size (μm) estimated by the defect inspection apparatus, and the SEM length measurement size (μm) measured by SEM.

[Fig. 9]

The horizontal axis is the size (μm) from the luminance signal of the front-end process wafer (initial process wafer) and the back-end process wafer (late process wafer) according to the present invention, and the vertical axis is the SEM length measurement size (μm). It is a correlation diagram which plotted the state of foreign matter generation in the case of.

[Fig. 10]

Convex defects (particle-like foreign matter) and concave defects (scratch, thin-film foreign matter) generated on the CMP of the wafer according to the present invention are measured by the size (μm) from the brightness signal detected by the defect inspection device and SEM. It is a correlation diagram showing the correlation with the lengthened SEM length measurement size (μm).

[Fig. 11]

It is a figure which shows one Example of the discrimination processing flow of the scratch, the thin film foreign matter, and the particulate foreign matter (ordinary foreign matter) which concerns on this invention.

[Fig. 12]

Scratches and thin-film foreign matter based on the degree of unevenness (b / a) and size based on the relationship between the luminance signal S (i) by epi-illumination and the luminance signal T (i) by oblique illumination, which show the basic idea according to the present invention. , And a correlation diagram for explaining that particulate foreign matter (ordinary foreign matter) is discriminated.

[Fig. 13]

It is explanatory drawing of the degree of unevenness.

[Fig. 14]

It is explanatory drawing for obtaining the inspection result classified into categories 1 to 5 so that the fatality of the defect which concerns on this invention can be discriminated.

[Fig. 15]

It is a figure which shows the distribution of the particulate foreign matter and scratch based on the relationship between the luminance signal S (i) by epi-illumination and the luminance signal T (i) by oblique illumination which concerns on this invention.

[Fig. 16]

It is a figure which shows the distribution of the particulate foreign matter and scratch based on the relationship between the logarithmic value of the luminance signal S (i) by epi-illumination and the logarithmic value of the luminance signal T (i) by oblique illumination which concerns on this invention.

[Fig. 17]

It is a figure which shows the wafer map which shows the distribution of each defect discriminated by the defect inspection apparatus which concerns on this invention.

[Explanation of symbols]

1 ... Illumination optics, 2a, 2b ... Light source, 4b, 4c, 4c1 ~ 4c3 ... Reflective mirror, 5 ... Detection optics, 6 ... Condensing lens (also imaging optics) Includes), 7 ... beam splitter, 8a, 8b ... photoelectric converter (CCD, TDI sensor), 9 ... arithmetic processing unit, 10 ... inspected object (wafer), 11 ... oblique Directional optics, 12 ... epi-illumination, 12a ... vertical optics, 12b ... pseudo-vertical optics, 14 ... stage controller, 15 ... stage, 16a, 16b ... A / D conversion unit, 17a, 17b ... storage unit, 18 ... comparison calculation unit (comparison judgment unit), 21 ... substrate, 22 ... insulating film (SiO)₂Membrane), 23a ... Scratch, 23b ... Thin film foreign matter, 24 ... Foreign matter (particulate foreign matter), 20 ... Discrimination line (threshold value), 30 ... Overall control unit, 31 ... Storage device, 32 ... input means, 33 ... display device, 100 ... defect inspection device, 101 ... process control computer, 102 ... yield management system, 103 ... network.