JP2012073069A - Preparation method of specimen for observing defective part of semiconductor device substrate - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a preparation method of a specimen for observing a defective part of a semiconductor device substrate, which enables the fine defective part of the semiconductor device substrate to be easily specified in a short time, so that a thin specimen suitable for transmission electron microscope observation can be made.SOLUTION: A method comprises: a first step of marking an upper surface of a fine defective part in a semiconductor device formed on a silicon semiconductor substrate as an observation object with electron beams or laser beams for location specification using an optical microscope; a second step of fixing the specified location on a support base; a third step of making a specimen for plane observation by thinning the substrate through pruning both surfaces with FIB processing to such an extent that an electron penetrates the substrate; a fourth step of specifying the fine defective part with STEM observation; a fifth step of newly marking the defective part after specifying the fine defective part; and a sixth step of cutting the specimen with FIB to make a specimen for section observation.

Description

  The present invention relates to a method of evaluating a defective portion of a semiconductor device using a transmission electron microscope (hereinafter referred to as TEM), and more particularly, a semiconductor device fabricated on a silicon substrate (hereinafter referred to as a semiconductor device substrate). The present invention relates to a method for producing a defect portion observation sample of a semiconductor device substrate for accurately observing and evaluating a minute defect portion.

  As a method of observing extremely minute defects in the internal structure of a semiconductor device substrate, the observation surface of a cross-sectional sample is irradiated with an electron beam, and the difference in the crystal structure of the material is reflected from the transmitted or scanned electron beam. Examples thereof include a transmission electron microscope (TEM) or a scanning transmission electron microscope (hereinafter referred to as STEM) which forms an image of the internal structure to obtain an image of the internal structure. Usually, these devices are collectively classified as TEM. In order to observe a defect inside the semiconductor device substrate using these methods, it is necessary to process the thickness of the semiconductor device substrate on the observation surface so as to be thin enough to transmit an electron beam.

  FIG. 2 is a perspective view of a sample portion showing a conventional procedure for producing a cross-sectional sample for TEM and STEM observation using a FIB (Focused Ion Beam) processing method. (A) is process flow explanatory drawing, (b) is a perspective view of the preparation sample corresponding to the process flow of (a). First, a semiconductor device substrate (not shown) in a wafer state is prepared, a minute defect portion to be observed is observed with an optical microscope, and a specific place is marked on the upper surface of the defect portion with a laser marking 102 or the like. When the sample block 101 having a size of about 300 μm × 300 μm, which is much larger than the laser marking 102, is cut out with a dicer (not shown), FIB, or the like so that the defective portion is surely included Is removed by a cleavage method (procedure A1). In an FIB apparatus having a large sample chamber, a semiconductor device substrate having the size of the original wafer can be directly brought into the sample chamber, and the vicinity of a minute defect portion to be observed can be subjected to ion beam processing. Cutout processing can be omitted.

 Next, the sample block 101 including the defective portion cut out in the procedure A1 is introduced into the sample chamber in the FIB apparatus, and the laser marking 102 is used as a mark by the narrowed ion beam, and the defective portion to be observed is observed. The outer peripheral part of the specific location 103 (size: thickness 5 μm × length 20 μm × width 20 μm) is surely included in a box shape, and finally the bottom of the specific location 103 is 5 μm deep. To cut out the specific sample 103a. Since the specific sample 103a taken out from the sample block 101 cut out in the above-described procedure A1 includes a defect portion to be observed, the same FIB apparatus is used so that it is not altered or deformed by heat or the like during subsequent ion beam processing. Then, it is covered and protected with a tungsten film (not shown) of a metal thin film (called a deposition film) (procedure A2). This specific sample 103a is fixed to the support 104 prepared separately.

 Alternatively, when a well-known function called a microsampling method (not shown) is attached to the FIB apparatus, a thin wire probe is attached to the specific sample 103a and the specific sample 103a is lifted, and the support base 104 is Fix to a kind of copper mesh. Next, the specific sample 103a fixed on the support base 104 is further processed by FIB into a thin enough thickness (size (thinned portion); about 5 μm × 0.1 μm × 5 μm) to transmit a TEM electron beam. Then, the production of the observation cross-section sample 105 that has been thinned for use in TEM observation is completed (procedure A3). As the method for processing the specific sample 103a into the sliced cross-section sample 105 for observation, the specific sample 103a is processed by FIB or further thinned by argon ion etching using an Auger electron spectrometer. If the FIB apparatus does not have a micro-sampling function, after the procedure A1, the FIB process from the box-like process of the above-described procedure A2 is performed without removing the bottom surface and further to the cross-section slice processing of the procedure A3. And then suspending the tip of a tapered glass rod with an electrostatic force and attaching a flaky cross-section sample 105 including a specific location 103 to a support base 104 such as a copper mesh. Production of the cross-sectional sample 105 for use is completed (Patent Documents 2, 3, 4, 5, Non-Patent Documents 1 and 2).

  FIG. 3 is an outline of an observation apparatus using a TEM for observing the internal structure of the flaky observation cross-section sample 105 manufactured in the procedure A3 of FIG. 2 and fixed to a support base 104 such as a copper mesh. A block diagram is shown. The observation apparatus includes an electron beam 3 of a TEM (not shown) that irradiates the observation cross-section sample 105, an electron beam 4 that has passed through the observation cross-section sample 105, an electron beam detector 5 that detects the transmitted electron beam 4, A CRT (Cathode Ray Tube) 6 for image observation is used. However, the conventional method for producing the cross-sectional sample for observation 105 described in the procedure A1 to the procedure A3 can be applied to a large defect on the surface of the semiconductor device substrate of the order of microns or more. However, as shown in FIG. 4, for the nanometer-order minute defect portion 10 on the interface with the silicon substrate 7 below the doped polysilicon film 9 inside the semiconductor device substrate or on the gate insulating film 11. Since the interlayer insulating film 8 formed on the semiconductor device substrate and the electrodes and wirings (not shown) made of metal are in the way, the minute defect portion 10 to be observed cannot be observed with an optical microscope. Furthermore, there is a problem that the laser marking 102 to the specific location 103 on the corresponding substrate surface described in FIG. 2 cannot be performed.

  In particular, when several layers of the interlayer insulating film 8 are stacked above the minute defect portion 10, the thickness of the interlayer insulating film 8 is usually 1 to 2 μm, and the size that can be observed with an optical microscope is The size is about several μm. Electrodes such as metal and wiring material (not shown) on the interlayer insulating film 8 can be easily removed by etching or the like, but minute foreign matters that are under the interlayer insulating film 8 and between the silicon substrate 7 And defects cannot be observed from the surface with an optical microscope. In addition, when these minute defect portions 10 are present on the surface of the semiconductor device substrate, it is relatively easy to identify them with an optical microscope. However, in particular, when the defect is an internal defect related to the material of the silicon substrate 7, for example, a crystal defect. Has a problem that it cannot be identified by an optical microscope unless it is exposed by a method that does not affect the defects. As countermeasures, a method for observing crystal defects inside using a scanning electron microscope (hereinafter referred to as SEM) or a device for observing defects to be observed reliably by marking in the FIB have been devised. Yes. In addition, a method has been proposed in which a specific place including a defective portion is sampled, attached to a holder in the FIB apparatus, and processed from two or more directions while being observed in the FIB to create a thin film sample. In addition, there is an emission microscope method as a method for electrically detecting a defective portion. The defective portion emits light by laser irradiation, and the vicinity thereof is marked with a laser marker, and the vicinity of the marking portion by an apparatus combining FIB and SEM. In recent years, a method of observing a defective portion while processing a thin piece is also performed, but there is a problem that the defective portion may not be identified and observed even by this method.

JP 2008-14899 A Japanese Patent No. 3965761 JP-A-8-327514 JP 2004-245660 A Japanese Patent No. 3843637 Japan Society for Invention and Innovation Open Technical Bulletin No. 2006-504555 Japan Society for Invention and Innovation Open Technical Bulletin No. 2006-503669

  However, all of the above-described methods for producing the cross-sectional sample for TEM observation are those in which the size of the defect portion is on the order of microns and the defect location to be observed by the optical microscope can be specified. It is not suitable as a sample preparation method for defective portions. Even if the vicinity of the defect portion to be observed can be collected as a sampling sample, it is complicated to identify and determine the existing defect portion by observing the sampling sample. For example, in order to make it possible to determine a defective portion, there is a problem that a considerable processing time is required and work efficiency is poor. In some cases, the defect part is often overlooked and disappears. Further, paragraph 0022 of Patent Document 1 only describes that a defect position is observed and identified two-dimensionally with a TEM / STEM.

  The present invention has been made in view of the above-described points, and an object of the present invention is to easily identify a minute defect portion of a semiconductor device substrate in a short time and to be a thin piece suitable for observation with a transmission electron microscope. It is an object of the present invention to provide a method for producing a defect observation sample of a semiconductor device substrate that can be used as a sample.

  In order to achieve the object of the present invention, a portion near a minute defect portion included in a semiconductor device substrate is cut out from the semiconductor device substrate as a sample block, and the minute defect portion in the sample block is optically formed. First step of laser marking on the surface of the substrate above the microdefect portion specified by a microscope, the substrate block including the microdefect portion specified by the laser marking is cut out from the sample block by a focused ion beam, A second step of fixing the substrate portion including the minute defect portion identified by the cut laser marking to a support base; the electron beam is transmitted by the focused ion beam through the substrate portion including the micro defect portion fixed to the support base; A third step of cutting into a thin piece having a thickness sufficient to obtain a flat observation sample, The micro defect portion is specified by magnifying and observing with a scanning secondary electron image of a microscope, and a contamination film generated by scanning the secondary electron is deposited on the surface of the planar observation sample above the micro defect portion. A semiconductor having a fourth step of marking, and a fifth step of cutting the planar observation sample with a focused ion beam on the surface of the planar observation sample above the minute defect portion specified by the marking to obtain a sectional observation sample A method for producing a defect observation sample of a device substrate is used. The sample for plane observation produced in the third step is made thick enough to allow the electron beam to pass through the substrate portion including the minute defect portion fixed to the support base by scraping from both sides with a focused ion beam. In addition, it is preferable that the minute defect portion is exposed on the surface. In addition, the third step is such that a focused ion beam and low-acceleration argon etching are performed in this order on the substrate portion including the minute defect portion fixed to the support base in order to transmit the electron beam from both sides. It can also be a step of making a flat observation sample in which the minute defect portion is exposed on the substrate surface. The low acceleration argon etching is preferably etching using an Auger electron spectrometer.

  The object of the present invention is to cut a portion near a micro-defect portion in a semiconductor device formed on a silicon substrate from the semiconductor device substrate using a focused ion beam and fix it to a support table, and then fix to the support table. First step of wet etching the surface of the vicinity of the micro defect portion to expose the micro defect portion on the surface, the exposed micro defect portion is enlarged and observed with a scanning secondary electron image of a scanning electron microscope, and exposed. A second step of depositing a contamination film generated at the time of observation of the scanning secondary electron image on the minute defect portion to form a marking; cutting the minute defect portion specified by the marking using a focused ion beam; It is also achieved by making a method for producing a defect portion observation sample of a semiconductor device substrate having a third step of making a thin piece to obtain a flat observation sample. As the wet etching, it is preferable to use a hydrochloric acid solution for etching the aluminum electrode in the vicinity of the minute defect portion, a hydrofluoric acid solution for the silicon oxide film, and a TMAH (tetramethylammonium hydroxide) solution for the polysilicon film.

  According to the present invention, a minute defect portion of a semiconductor device substrate can be easily identified in a short time, and a defect observation sample of a semiconductor device substrate that can be made into a thin piece sample suitable for transmission electron microscope observation A manufacturing method can be provided.

It is a perspective view of the sample part which shows the preparation processes of the cross-sectional sample for TEM concerning SEM and the STEM observation concerning Example 1 of this invention. It is a perspective view of the sample part which shows the preparation process of the cross-sectional sample for the conventional TEM and STEM observation. The schematic block diagram of the observation apparatus using TEM for observing the inside of the cross-sectional sample for observation produced by the procedure A3 of the said FIG. 2, and being fixed to the copper mesh is shown. It is typical sectional drawing of the defect part vicinity of the cross-sectional sample for observation produced in the procedure of the conventional FIG. It is typical sectional drawing of the sample for cross-sectional observation produced by the production process shown in the said FIG. 1 concerning Example 1 of this invention. It is the schematic which shows the preparation process of the sample concerning Example 1 of this invention. It is a perspective view of the sample part which shows the preparation processes of the sample for TEM and STEM cross section observation concerning Example 2 of this invention. It is typical sectional drawing of the sample for cross-sectional observation produced by the production process shown in the said FIG. 1 concerning Example 2 of this invention.

  Embodiments of a method for producing a defect observation sample of a semiconductor device substrate according to the present invention will be described below in detail with reference to the drawings. The present invention is not limited to the description of the examples described below unless it exceeds the gist.

  Hereinafter, a preferred embodiment of a method for producing a defect observation sample of a semiconductor device substrate according to Example 1 of the present invention will be described in detail with reference to the drawings. FIG. 1 shows a procedure for producing a defect observation sample according to Example 1 for TEM and STEM observation. (A) is explanatory drawing of the flow of a 1st process-a 5th process, (b) is a perspective view of the sample of the production process step corresponding to (a). Note that the minute defect to be analyzed is the minute defect portion 10 in the nanometer order at the interface between the doped polysilicon film 9 and the gate insulating film 11 inside the semiconductor device substrate shown in FIG.

(First step)
First, as a first step, an emission microscope (not shown) or the like is used to find a place where the semiconductor device substrate has become electrically defective, and the attached laser is applied to the surface corresponding to the defective place (small defect portion). A wafer (not shown) on which a first laser marking is attached with a marker is prepared. As with the conventional sample preparation procedure shown in FIG. 2, the first laser marking (not shown) corresponding to the minute defect portion (the defective place) to be observed is enlarged using a dicer or FIB as a mark. If the sample block 20 is cut out with a size of 1 mm, or if the specific location including the minute defect portion to be observed is cracked without affecting the crystal plane, the sample block 20 is removed by cleaving. In the FIB apparatus having a large sample chamber, this operation can be omitted because the specific sample 23 including the minute defect portion can be cut out directly from the original wafer by FIB processing. Next, the cut-out sample block 20 is enlarged and observed with an optical microscope, the approximate position of the minute defect portion is confirmed, and the second laser marking 21 is formed on the surface corresponding to the minute defect portion with a laser marker. After that, electrodes such as metal and wiring provided on the substrate surface of the sample block 20 are removed by etching or the like. Since the optical microscope has a low magnification, the position of the minute defect portion can be confirmed, and the second laser marking 21 can be attached, but the details of the minute defect portion are unknown. If necessary after the metal etching, the second laser marking 21 is attached again.

(Second step)
Next, as a second step, the sample block 20 is placed in a sample chamber in an FIB apparatus (not shown) in a direction in which a plane parallel to the main surface of the substrate is perpendicular to the ion beam (FIB) emission direction. Using a finely squeezed FIB, the outer periphery of the specific location 22 (size: about 5 μm × 20 μm × 20 μm) where the minute defect portion is included is boxed with the second laser marking 21 as a mark so as to dig up the root of the tree Finally, the sample block 20 is inclined and the bottom surface of the specific location 22 is cut and separated to take out the specific sample 23. A tungsten film (not shown) of a metal thin film (called a deposition film) is formed on the surface corresponding to the upper portion of the minute defect portion in the specific location 22 in the same FIB apparatus so as not to be processed by the FIB in a later process. As a protective film, it is fixed to a support base 104 such as a semi-circular mesh made of copper or molybdenum prepared separately. Or, when a function called microsampling method is attached to the FIB apparatus, a thin wire-like probe is attached to the specific location 22 and then the bottom surface is cut off, and the specific sample 23 is lifted and fixed to the mesh 104 as described above. To do.

(Third step)
Next, the specific sample 23 fixed to the support base 104 is thin enough to allow transmission of an electron beam at the time of STEM observation centering on the deposition film portion, and is processed by the FIB. Production is complete. Processing by FIB is performed by thinning from both sides. The size of the portion to be thinned is about 5 μm × 0.1 μm × 20 μm.

(4th process)
After this third step, the planar observation sample 24 is rotated by 90 ° and introduced into the STEM chamber, an electron beam is transmitted through the sample at an acceleration voltage of 200 kV, and a bright field image and a dark field image by transmitted electrons are observed. The defect is observed with a contrast different from that of the original constituent material of the semiconductor device. Therefore, the minute defect portion can be distinguished from the original constituent material of the semiconductor device. For example, in the semiconductor device as shown in FIG. 4, when there is a minute defect portion 10 at the interface between the gate insulating film 11 and the doped polysilicon film 9, an electrical characteristic test on a wafer in a previous process or an emission microscope Many micro-defects are in a melted state due to the light emission operation. This minute defect portion is observed as an amorphous contrast different from the contrast due to the thinly processed crystal of the doped polysilicon film 9 in the upper layer. Even when the minute defect portion 10 is at the interface between the silicon substrate 7 and the gate insulating film 11, it is observed as an amorphous contrast different from that of the silicon substrate 7.

  In the fourth step, when the minute defect portion cannot be specified by observation of the bright field image and the dark field image by transmission electrons, the third step is performed again. By repeating the third step and the fourth step as necessary, the position of the minute defect portion in the plane direction is specified.

  Compared to conventional STEM observation, processing performed over a long period of time only with FIB in the cross-sectional direction as in the past, or a method of identifying a defective portion by observing with SEM while processing with FIB (dual beam processing method) Therefore, the oversight of the minute defect portion is reduced, and the entire minute defect portion can be observed in detail. In the case where a specific substance that causes a problem exists in the minute defect portion, it can be determined that the portion shows a different contrast than the peripheral portion.

Further, when observing a minute defect portion by a secondary electron image, depending on the type of the minute defect portion, the contrast of the minute defect portion may be relatively small due to the contrast of the doped polysilicon film 9 and may not be clearly visible. . In that case, it is effective to remove the doped polysilicon film 9 covering the minute defect portion with an etching apparatus that irradiates argon ions at a low acceleration voltage so that the minute defect portion is clearly exposed. A specific method will be described later.
(5th process)
On the other hand, in the defect analysis, in order to identify the specific substance that causes the defect and know the position of the constituent material layer, etc., not only the observation of the sample for plane observation described above, but also the observation in the cross-sectional direction of the sample. There is a need to.

  In order to create the cross-sectional observation sample 26, first, the cross-sectional observation position of the specific substance is marked. The marking is performed by irradiating a minute defect identified in the fourth step with an electron beam at an acceleration voltage of 200 kV in the same STEM apparatus, and observing a bright-field image and a dark-field image by transmitted electrons, Switch to observation. Normal scanning secondary electron image observation is up to about 100,000 times, but in order to accumulate the contamination for marking that serves as a mark, the electron beam irradiation amount per unit area is increased, and the magnification is 1 million times or more. Observe for a few minutes. If the electron beam dose cannot be increased by the magnification, increase the primary electron current amount, further increase the condenser lens current amount, increase the electron beam irradiation amount by narrowing the electron beam to enlarge the objective movable aperture hole, etc. Then, the marking 25 by contamination is attached to the surface of the cross-sectional observation position of the target location. The marking 25 is convenient to have a size of several μm that can be sufficiently seen during the FIB processing to be performed later.

In the fifth step, when a scanning secondary electron image is observed, it is desirable to increase the magnification stepwise in order to efficiently identify the minute defect portion. For example, first, the magnification is several tens of thousands of times, the position of a more accurate minute defect portion is specified, then the magnification is several hundred thousand times with respect to the specified portion, and the more accurate position of the minute defect portion is determined. Specify 1 million times more. In this way, there are three stages. Of course, it is possible to have four or more stages.
(Sixth step)
Next, as a sixth step, the planar observation sample 24 with the marking 25 is introduced into the sample chamber in the apparatus so that the position of the minute defect (marking 25) is not processed by the ion beam. Then, a tungsten film (not shown) of a metal thin film (called a deposition film) is attached in the same FIB. When a function called microsampling method is attached to the FIB apparatus, a specific sample 24 formed by attaching a thin wire probe to the same surface as the minute defect and then cutting the bottom is prepared separately. It is fixed in a cross-sectional direction to a support base 104 such as a semicircular mesh made of copper or molybdenum, and a tungsten film is deposited on the processed surface. Next, the cross-sectional observation sample used for the TEM observation is obtained by cutting the flat observation sample 24 with the marking 25 on the specific sample 24 fixed to the support base 104 such as a mesh with a FIB around the deposition film portion. 26 is completed. The size of the thinned portion is about 5 μm × 0.1 μm × 0.1 μm.

Thereafter, the sample holder on which the thinned cross-sectional observation sample 26 is mounted is removed from the FIB apparatus, and the sample holder is replaced in the STEM apparatus, so that the cross-sectional observation position of the minute defect portion of the cross-sectional observation sample 26 is obtained. Observe.
(Regarding the formation of the contamination film 12)
FIG. 5 is a schematic cross-sectional view of the cross-sectional observation sample 26 manufactured by the manufacturing process shown in FIG. Marking by the contamination film 12 formed in the fourth step can be seen.

  As described above, after the plane observation sample 24 manufactured in the third step is specified in the plane direction of the minute defect portion 10 by the transmission electron image (bright field image or dark field image) in the STEM apparatus, Only by observing the scanning secondary electron image at the specified position, the formation of the marking by the contamination film 12 by the electron beam proceeds in the observation portion. Therefore, the minute defect portion 10 of the nanometer order can be identified in a short time. Can be easily. Usually, STEM images and observes scanning transmission electrons (bright-field image signals) or annular scanning transmission electrons (dark-field image signals) from the inside of a material obtained by irradiating the surface of a thinned sample with an electron beam. Although it is the main purpose, the STEM generally includes an electron detector for scanning secondary electron images in order to observe the surface state of the thin film sample. The former bright-field image or dark-field image is for observing crystal information and the like regarding the inside of the substance, and the latter scanning secondary electron image is for showing information such as surface irregularities. Accordingly, both the information on the inside and the surface of the minute defect portion 10 can be obtained, and the marking by the contamination film 12 serving as a mark is simultaneously performed by increasing the observation magnification in order to increase the amount of irradiation electron current. It is possible to apply. As for the amount of contamination, the product of the irradiation electron current amount and the time on the surface of the cross-sectional observation sample 26 increases in proportion to the observation region (magnification), but the scanning secondary electron image is larger than the scanning transmission electron image or the annular scanning transmission electron image. In the case of observing an electronic image, more contamination film 12 adheres to the surface of the cross-sectional observation sample 26. Since the component of the contamination film 12 is generally amorphous carbon, the composition contrast is clearly different from that of the deposition film 13 such as a tungsten film which is a peripheral constituent material. Convenient.

  A method for exposing the minute defect portion by removing the doped polysilicon film 9 covering the minute defect portion with the above-described etching apparatus that irradiates argon ions at a low acceleration voltage will be described below.

  A step of thinly processing the thin film in the planar direction by FIB to leave the doped polysilicon film 9 slightly, or a step of etching the doped polysilicon film 9 by wet etching to leave the doped polysilicon film 9 slightly. Then, the etching is preferably performed with a low acceleration voltage device such as an argon polisher.

  However, a planar observation sample 24 is placed in an Auger Electron Spectroscope (abbreviated as AES) that can be etched while observing minute defects, and the doped polysilicon is removed by irradiation with argon ions. Since it is a reliable method to expose a minute defect part, it is preferable. A specific method is shown in FIG. The planar observation sample 24 is placed in the STEM apparatus, the transmission electron image is observed, the micro defect part 10 is specified by the difference in contrast between the micro defect part 10 and another region, and the transmission electron image is observed in the STEM apparatus. Switching to the observation of the next electron image Step of forming the contamination film 12 above the minute defect portion 10 (FIGS. 6A and 6C), and etching the doped polysilicon film 9 with an Auger electron spectrometer 6 (b) and 6 (d) are repeated to expose the minute defect portion 10. Thereafter, the fourth step and the fifth step are performed, and the contamination film 12 shown in FIG. Then, the sixth step is performed to produce the cross-sectional sample 26.

  In this AES, since it can be dug down in the depth direction by argon ion irradiation and the elemental analysis of the surface is possible, most of the defect portions can be analyzed. In FIB, since processing is performed with argon ions having a large energy (acceleration voltage 30 to 40 kV), the etching rate is fast and it is difficult to stop etching on the surface of the defect to be observed. Etching by AES using argon ions of 5 to 10 kV) is characterized in that the defect rate is less overlooked because the etching rate is slow.

FIG. 7 shows a method for producing a defect observation sample of a semiconductor device substrate according to Example 2 of the present invention. (A) is explanatory drawing of the flow of a 1st process-a 3rd process, (b) is a perspective view of the sample of the production process step corresponding to (a).
(First step)
The specific sample 23 is fixed to the support base 104 such as a mesh in the same manner as in the first step of the first embodiment (FIG. 1). Thereafter, the surface of the specific sample 23 is etched using TMAH (tetramethylammonium hydroxide) or the like to expose the minute defect portion 10.
(Second step)
Next, this specific sample 23 is put into the sample chamber of the SEM apparatus, and the micro defect portion 10 is observed in a plane. In the observation, an electron beam is irradiated at an acceleration voltage of 25 kV, and the minute defect portion 10 is specified by a scanning secondary electron image. The contamination film 12 is directly accumulated on the exposed portion of the microdefect 10 and becomes the marking 25 in a few minutes at a magnification of several tens of thousands of times with respect to the specified microdefect 10.

  In the defect analysis by SEM, a scanning secondary electron image is observed. The secondary electrons generated from the very surface of the sample are slightly different from each other in the minute defect portion and the base silicon substrate. The amount of secondary electrons generated is different, and as a result, different contrast images are observed. In addition, since the scanning secondary electron image is observed at a high magnification, the amount of electron current to be irradiated per unit area is increased, and contamination is locally attached to the peripheral part by electron beam irradiation. It is possible to protect the minute defect portion and to mark the mark.

However, as in the first embodiment, the FIB plane processing is not performed for plane observation, so that the transmission electron image cannot be observed using the STEM apparatus. Therefore, it is necessary to specify the minute defect portion only by the scanning secondary electron image, and it takes time until the smile defect portion is specified.
(Third step)
Thereafter, the specific sample 23 is placed on the FIB apparatus, the peripheral processing of the minute defect portion 10, the lifting by the probe, the fixing of the mesh or the like to the support base 104, etc. are performed, and the cross-section processing and thinning by the FIB are performed.

  Thus, the production of the cross-sectional observation sample 26 is completed. Thereafter, the sample holder on which the thinned cross-sectional observation sample 26 is mounted is removed from the FIB apparatus, and the sample holder is replaced in the STEM apparatus, so that the cross-sectional observation position of the minute defect portion of the cross-sectional observation sample 26 is obtained. Observe.

  FIG. 8 shows a schematic cross-sectional view of a cross-sectional observation sample 26 produced in Example 2. FIG. The difference from the cross-sectional observation sample 26 described in Example 1 is that the thinning is not performed by FIB. In this embodiment, since the transmission electron image is not observed with the planar observation sample 24, it is difficult to find the minute defect portion 10 as compared with the first embodiment. However, since the contamination film 12 is directly accumulated on the surface of the minute defect portion by SEM observation after the minute defect portion 10 is exposed, the contamination film 12 is interposed via the polysilicon film 9 described with reference to FIG. There is an advantage that the minute defect portion 10 can be protected against heat or the like rather than accumulating. Reference numeral 13 denotes a deposition film.

3 ... TEM electron beam 4 ... Transmitted electron beam 5 ... Electron beam detector 6 ... CRT for image observation
DESCRIPTION OF SYMBOLS 7 ... Silicon substrate 8 ... Interlayer insulating film 9 ... Polysilicon film 10 ... Minute defect part 11 ... Gate insulating film 12 ... Contamination film 13 ... Deposition film 20 ... Sample block 21 ... Laser marking 22 ... Specific place 23 ... Specific sample 24 ... Planar observation sample 25 ... Marking 26 ... Cross-section observation sample 101 ... Sample block 102 ... Laser marking 103 ... Specific place 103a ... Specific sample 104 ... Support base 105 ... Cross-section sample for observation

Claims (4)

A portion near the minute defect portion included in the semiconductor device substrate is cut out from the semiconductor device substrate as a sample block, and the minute defect portion in the sample block is identified with an optical microscope, and an upper portion of the minute defect portion A first step of laser marking on the substrate surface of
A substrate portion including the minute defect portion specified by the laser marking is cut out from the sample block by a focused ion beam, and the substrate portion including the minute defect portion specified by the cut laser marking is fixed to a support base. 2 steps,
A third step in which a substrate portion including a minute defect portion fixed to the support base is cut into a thin piece having a thickness enough to transmit the electron beam by a focused ion beam to obtain a planar observation sample;
A fourth step of observing the planar observation sample with a transmission electron image of a transmission scanning electron microscope to identify a minute defect portion;
The planar observation sample is magnified and observed with a secondary electron image with respect to the minute defect portion specified in the fourth step, and the contamination film generated by scanning the electron beam is observed on the planar surface of the minute defect portion. A fifth step of depositing and marking on the sample surface
A semiconductor device comprising: a sixth step of cutting the planar observation sample with a focused ion beam on the surface of the planar observation sample above the minute defect portion specified by the marking to obtain a sectional observation sample A method for manufacturing a defect observation sample of a substrate. The sample for plane observation produced in the third step is made thick enough to allow the electron beam to pass through the substrate portion including the minute defect portion fixed to the support base by scraping from both sides with a focused ion beam. The method for manufacturing a sample for observing a defect portion of a semiconductor device substrate according to claim 1, wherein the minute defect portion is exposed on the surface. In the third step, a focused ion beam and a low-acceleration argon etching are performed in this order on the substrate portion including the micro defect portion fixed to the support base so as to transmit the electron beam from both sides. 3. The method for producing a defect portion observation sample of a semiconductor device substrate according to claim 2, wherein the method is a step of forming a flat observation sample in which the minute defect portion is exposed to the substrate surface while being sharpened. 4. The method for producing a defect observation sample of a semiconductor device substrate according to claim 3, wherein the low acceleration argon etching is etching using an Auger electron spectrometer.