JP6955675B2 - Sample preparation method, defect observation method - Google Patents

Description

The present invention relates to a method for preparing a sample (sample) for observing defects existing in a silicon crystal and a method for observing defects in the sample.

In silicon crystals, crystal defects introduced in the crystal manufacturing process, defects introduced due to processing strain and stress introduced in the wafer manufacturing process, heat and wafer support received during epitaxial layer film formation and device manufacturing process. There are various types and forms of crystal defects such as defects introduced by contact of parts.

The above-mentioned defect generation varies depending on various factors such as the wafer manufacturing process, the oxygen concentration of the wafer, and the dopant concentration. The size and density of the defects are not uniform, and it is extremely difficult to observe the actual condition of the defects. There was also.

Therefore, the position of the identified defect is set to 1 μm by using an inspection device that accurately shows the position information of the defect and a high-precision sample preparation device represented by a FIB device (FIB: Focused Ion Beam). At the same time that it has been required to process a sample for TEM (Transmission Electron Microscope) with the following accuracy, the inspection and sample processing equipment have been evolving day by day.

In the prior art, first, defects required by the position coordinates and light scattering of foreign matter and crystal defects that become defects by an optical surface defect inspection device (specifically, SP series manufactured by KLA-Tencor, MAGIC manufactured by Lasertec, etc.) Measure the size of the Then, depending on the measurement result, a Review SEM (SEM: Scanning Electron Microscope, scanning electron microscope) (specifically, RS series manufactured by Hitachi High Technology, SEM-Vision series manufactured by Applied Materials, etc.) is used. It distinguishes whether the object detected by the inspection device is a particle or a structural defect (scratch or protrusion), and determines whether it is a defect that is not manifested on the surface or the outermost layer and exists directly under it. Select a target for further detailed evaluation, such as by doing so. Then, after selecting the target object, marking is performed, and the FIB apparatus relies on the marking to perform sample processing that includes defects. After that, the type of defect and its composition were clarified by TEM observation of this sample.

On the other hand, for defects that do not appear on the wafer surface, the density of the defects inside is investigated by a selective etching method or an optical light scattering method, and the defects are the volume at the time of TEM sample preparation. Find the probability of being inside. Based on that probability, the number of samples existing at least one or more was prepared by the FIB method or the dimple grinding method, and the TEM observation of each prepared sample was performed.

Here, regarding the preparation of a sample for TEM, Patent Document 1 describes that a sample for TEM is prepared by a lithography process and reactive ion etching (RIE) using plasma. Further, Patent Document 2 describes that a sample for TEM is prepared by using FIB processing.

Japanese Unexamined Patent Publication No. 2004-286486 Japanese Unexamined Patent Publication No. 2011-184626

Conventionally, it may take a large amount of time to prepare a sample for the above-mentioned probability. Depending on the probability that a defect is present, there is even a probability that one defect can be observed by preparing tens to hundreds of samples.

Further, for example, when the sample is prepared by the dimple grinding method, the preparation time per sample is usually about 4 hours. Therefore, assuming that 6 samples can be produced at full operation per day, it will take 17 days to finish 100 samples. It is very inefficient to prepare a sample by this method, and even if a calculated number is prepared, it is often necessary to prepare the sample again in the absence of defects.

The present invention has been made in view of the above problems, and a sample preparation method capable of efficiently obtaining a plurality of samples for observing defects existing in a silicon crystal and the defects can be efficiently observed. An object of the present invention is to provide a defect observation method.

In order to solve the above problems, the sample preparation method of the present invention is a sample for defect observation by a transmission electron microscope in which a pedestal and a flaky portion forming a defect observation region are integrally formed on the pedestal. Is a method of making
From one silicon crystal plate , a plurality of silicon pieces having a shape corresponding to the sample are arranged in the in-plane direction inside unnecessary parts that are not required when observing defects, and a connection allowance that allows each silicon piece to be cut. The silicon piece array shape supported by the unnecessary part is formed by photolithography and etching based on the resist pattern formed by the photolithography .
The processing process is
A surface processing step of forming a top view shape of the silicon piece array shape by processing the surface of the silicon crystal plate by photolithography and etching.
A back surface processing step of forming a bottom view shape of the silicon piece array shape by processing the back surface of the silicon crystal plate by photolithography and etching.
To be equipped .

According to this, since a plurality of silicon pieces can be obtained from one silicon crystal plate at a time, a plurality of samples can be efficiently prepared as compared with the conventional method in which one silicon piece is produced at a time. Obtainable.

Wherein in the processing step of forming the silicon piece 10 or more (e.g., 50 to 100) inclusive the silicon piece arrangement shape. By producing a large number of silicon pieces at one time in this way, a large number of samples can be efficiently obtained.

Further, in the processing step, a surface processing step of processing the front surface of the silicon crystal plate by photolithography and etching, and a back surface processing step of processing the back surface of the silicon crystal plate by photolithography and etching. Is carried out to form the silicon piece having a shard portion constituting the observation region of the defect and a pedestal supporting the shard portion. In this way, by performing the processing in two steps, the front surface processing step and the back surface processing step, a silicon piece having a thin piece portion and a pedestal can be produced.

Further, in the processing step, a shape in which each of the silicon pieces is connected to the inside of the frame portion, which is unnecessary for defect observation, with a connection allowance capable of cutting is formed by photolithography and etching. According to this, a plurality of silicon pieces separated from each other can be obtained by cutting the connection allowance.

In addition, the sample preparation method of the present invention includes a removing step of removing residual strain on the etching surface of the silicon piece obtained in the processing step. As a result, the residual strain on the etched surface of the silicon piece can be removed, and a high-quality sample can be obtained.

In the removing step, it is preferable to collectively process the plurality of the silicon pieces. According to this, a plurality of samples can be obtained more efficiently.

Further, the removing step can be a step of sacrificing oxidation treatment of the etched surface. The removing step may be a step of performing an ion etching treatment or an alkali etching treatment on the etched surface.

In addition, the sample can be a sample for observation with a transmission electron microscope. In the present invention, the transmission electron microscope also includes a scanning transmission electron microscope (STEM).

Further, the processing step and the removing step are carried out so that the silicon piece after the removing step has a thin section having a thickness of 2 to 6 μm, and the silicon piece after the removing step has an acceleration voltage of 1000 kV. This sample is used for observation with the above transmission electron microscope (hereinafter, may be referred to as ultra-high pressure TEM).

In a transmission electron microscope with an acceleration voltage of less than 1000 kV, the thickness of the sample must be 1 μm or less, but if an ultra-high voltage TEM is used, even a thick sample (a silicon piece having a thin section of 2 to 6 μm) can be observed. Can be done. This makes it possible to obtain a sample in which defects can be observed more efficiently.

In addition, the flaky section has a rectangular observation area of 100 to 200 μm × 1000 to 2000 μm. By using this silicon piece (sample), defect observation can be performed in the range of 100 to 200 μm × 1000 to 2000 μm, which is compared with a general TEM sample (for example, a sample having an observation region of about 20 μm × 100 μm). It is possible to suppress the fact that defects cannot be observed.

In addition, the sample can be a sample for observing defects existing at a density lower than ^{1 × 10 8} cm ^-3. By using this sample, low-density defects can be observed with high efficiency.

In the defect observation method of the present invention, defect observation is performed on a plurality of samples obtained by the sample preparation method of the present invention. According to this, since defect observation is performed on a plurality of samples produced with high efficiency by the sample production method of the present invention, defects existing in the silicon crystal can be efficiently observed.

Further, the defect observation method of the present invention includes an ultra-high voltage TEM observation step of observing a plurality of the samples with a transmission electron microscope (ultra-high voltage TEM) having an acceleration voltage of 1000 kV or more. By using the ultra-high pressure TEM in this way, it is possible to observe a thicker sample, so that it is possible to suppress the inability to observe defects, and it is possible to observe defects more efficiently.

Further, the defect observation method of the present invention has higher resolution and analysis sensitivity than the transmission electron microscope used in the ultra-high pressure TEM observation step for the sample whose presence of defects is shown in the ultra-high pressure TEM observation step. It is provided with a detailed evaluation step of performing defect evaluation with a transmission electron microscope having. In this way, by performing the detailed evaluation process on the samples showing the presence of defects in the ultra-high pressure TEM observation process, the detailed evaluation process is performed on all the samples without going through the ultra-high pressure TEM observation process. It is possible to evaluate the defects in detail more efficiently than in the case of doing so.

Further, the detailed evaluation step is a step of observing defects with a transmission electron microscope having an acceleration voltage of less than 1000 kV, and the sample showing the presence of defects in the ultra-high pressure TEM observation step is subjected to the detailed evaluation step. It is provided with an additional processing step of processing so as to have a shape suitable for the transmission electron microscope to be used, and in the detailed evaluation step, defect evaluation is performed on the sample obtained in the additional processing step. This makes it possible to perform a detailed evaluation of defects.

Further, a recording step of recording the defect position in the sample in which the presence of a defect is shown in the ultra-high pressure TEM observation step is provided, and in the additional processing step, processing is performed based on the defect position recorded in the recording step. .. As a result, a sample for the detailed evaluation process containing defects can be efficiently produced, and it is possible to prevent defects from being observed in the detailed evaluation process.

Further, in the recording step, the position corresponding to the defect position in the sample in which the presence of the defect was shown in the ultrahigh pressure TEM observation step is marked by the electron beam of the transmission electron microscope used in the ultrahigh pressure TEM observation step. I do. With this marking, it is possible to easily identify (record) the position of the defect in the sample before the additional processing.

It is a front view of the sample of an Example. It is a side view of the sample of an Example. It is a flowchart which showed the manufacturing method of the sample in an Example, and the defect observation method using this sample. It is a figure which showed the state of the silicon crystal plate after the process of S3 of FIG. It is a figure which showed the state of the silicon crystal plate after the process of S4 of FIG. It is a figure which showed the state of the silicon crystal plate after the process of S6 of FIG. It is a figure which showed the state of the silicon crystal plate after the process of S7 of FIG. It is a figure which showed the silicon piece finally obtained by the process of S2 S7 of FIG. It is the figure which illustrated the silicon crystal plate after the process of S7 of FIG. It is an observation photograph of SEM which showed the state which attached the sample to the grit for TEM observation.

Hereinafter, embodiments of the present invention will be described. Until now, in order to observe defects in silicon crystals by TEM (including STEM), sample preparation methods such as the FIB method and dimple grinding method have been appropriate, and the technology has developed related academic fields and their equipment. It has been progressing day by day by the equipment manufacturers. However, in both sample preparation methods, the maximum size of the observation area was about 100 × 100 μm at the largest. In this case, the defect density that can be observed with a general (acceleration voltage is 300 kV class) TEM ^{is required to be at least 5 × 10 7} cm- ^3, and an ultra-high voltage TEM (1.3 MV class) is used. Also requires a defect density of about 1 x 10 ⁷ cm- ^{3 pieces.}

In this embodiment, the thickness of one sample is tens of times (for example, about 50 times) that of the conventional one by using an ultra-high pressure TEM, and a large number of samples (the observation scan range of one sample is up to 2000 ×) by the MEMS processing technique. A production method of about 200 μm) is provided. Specifically, a large number of silicon pieces having the observation region are formed at one time by a photolithography and dry etching method, which is one of the MEMS (Micro Electro Mechanical System) processing technologies, and each silicon piece is used as a sample for TEM observation. Attach it to the grit for observation. When the thickness of one sample is about 50 times that of the conventional one and the number of samples to be produced at one time is about 50 to 100 times that of the conventional one, the efficiency is 50 × 50 to 50 × 100, that is, 2500 to 5000 times that of the conventional one. Crystal defects that exist only below the defect density can also be observed. The details will be described below.

1 and 2 illustrate the TEM sample of this embodiment. Sample 1 of FIGS. 1 and 2 is formed of a silicon single crystal. The sample 1 (silicon piece) has a thin piece portion 2 constituting a defect observation region, a pedestal 3 for supporting the thin piece portion 2, and support columns 4 provided on both left and right sides of the thin piece portion 2. .. The thin section 2, the pedestal 3, and the support column 4 are integrally formed.

The flaky portion 2 is formed in the form of a rectangular thin film when viewed from the front in FIG. The width d1 of the long side of the thin section 2 is, for example, 1000 to 2000 μm. The width d2 of the short side of the thin section 2 is, for example, 100 to 200 μm. The thickness d3 (see FIG. 2) of the thin section 2 is, for example, 2 to 6 μm, preferably about 5 μm (4 to 6 μm). The thin section 2 is formed on the pedestal 3 so that the long side is parallel to the upper surface of the pedestal 3 and the short side is perpendicular to the upper surface thereof. That is, all of one long side of the thin section 2 is connected to the upper surface of the pedestal 3. The rectangular surface (that is, the defect observation region) of the slice portion 2 is, for example, a surface perpendicular to the surface of the silicon crystal plate (silicon wafer) before sample processing.

The pedestal 3 is formed so as to support the thin section 2 from the long side of one of the flakes. The pedestal 3 is formed in a rectangular parallelepiped shape which is a rectangle in the front view of FIG. 1 and a right-angled quadrangle in the side view of FIG. When viewed from the front in FIG. 1, the long side of the pedestal 3 is parallel to the long side of the thin section 2, and the short side of the pedestal 3 is parallel to the short side of the thin section 2. Further, the width d4 of the long side of the pedestal 3 is slightly larger (for example, about 100 μm) than the combined width (= d1 + 2 × d7) of the width d1 of the long side of the thin section 2 and the width d7 of the two strut portions 4. It is set to a value. The pedestal 3 is formed so as to slightly protrude in the left-right direction from the support columns 4 on both the left and right sides on the paper surface of FIG. Further, the width d5 of the short side of the pedestal 3 is set to be about the same as the width d2 of the short side of the thin piece portion 2, for example. Further, the thickness d6 of the pedestal 3 (see FIG. 2) is set to a value larger than the thickness d3 of the thin section portion 2 and the thickness d8 of the strut portion 4, and specifically, it is set to about 50 μm, for example.

The center line that divides the left and right width d4 of the pedestal 3 into two equal parts on the paper surface of FIG. 1 and the center line that divides the left and right width d1 of the thin section portion 2 into two equal parts are located on the same straight line L1. Further, on the paper of FIG. 2, the center line that divides the width d6 of the pedestal 3 into two equal parts, the center line that divides the width d3 of the slice portion 2 into two equal parts, and the center line that divides the width d8 of the support portion 4 into two equal parts. Is located on the same straight line L2. The pedestal 3 is not limited to the above shape, and may have any shape as long as it can support the thin section 2 and can be attached to the TEM observation grit.

The strut portions 4 on both the left and right sides are formed in the same shape as each other. Each strut portion 4 is formed in a columnar shape having the same length d2 as the short side along the short side of the thin section portion 2 on the paper surface of FIG. Further, when viewed on the paper of FIG. 2, the thickness d8 of the support column portion 4 is larger than the thickness d3 of the thin section portion 2 and smaller than the thickness d6 of the pedestal portion 3. These two support columns 4 have a role of supporting the thin section 2 from the left and right (short side). In other words, the two strut portions 4 have a role of reinforcing the flaky portion 2 so as not to be deformed.

Next, with reference to the flowchart of FIG. 3, the details of the method for producing the sample 1 of FIGS. 1 and 2 and the defect observation method using the sample 1 will be described. First, the silicon crystal plate for defect observation is cleaned with a cleaning liquid to remove foreign substances such as particles and oxide films adhering to the front surface and the back surface of the silicon crystal plate (S1). The silicon crystal plate may be a silicon single crystal wafer before epitaxial growth, or may be a wafer after epitaxial growth (that is, a silicon epitaxial wafer). Further, the characteristics such as resistivity, oxygen concentration, dopant concentration, and conductive type of the silicon crystal plate are not particularly limited. Further, in order to facilitate the formation of a sample in each of the subsequent steps, a circular silicon wafer may be processed into a predetermined shape (for example, a quadrangular plate shape) with a dicing saw or the like, and then the steps of S1 or less may be carried out.

Next, a resist is applied to the surface of the silicon crystal plate after cleaning (S2). Next, the resist is patterned by a photolithography technique (exposure technique) to form a pattern having a predetermined shape (S3). Here, FIG. 4 shows a state in which the resist pattern 12 is formed on the surface 11 of the silicon crystal plate 10. In addition, although one circular pattern 12 is shown in FIG. 4 for the sake of clarity, in reality, in the process of S3, the top view shape (thin section portion 2 and strut portion) of the sample 1 of FIGS. 1 and 2 is shown. A pattern (hereinafter, referred to as a sample top view pattern) corresponding to the top view shape of 4) is arranged in a predetermined number in the in-plane direction of a predetermined area (area on which defect observation is to be performed) on the surface 11 (hereinafter, referred to as a sample top view pattern). Form a surface-side pattern). In the example of FIG. 4, the pattern 12 corresponds to one sample top view pattern. Further, the surface side patterns include a pattern (frame portion pattern) corresponding to the frame portion 19 of FIG. 9, which will be described later, a predetermined number of sample top view patterns arranged inside the frame portion pattern, and FIG. It is configured to include a pattern (connection allowance pattern) corresponding to the connection allowance 20.

Further, the number of sample top view patterns (pattern 12 in the example of FIG. 4) included in the surface side pattern can be, for example, 10 or more (for example, 50 to 100). The number of sample top view patterns is not limited to the above number, and can be appropriately set according to the size of the area where defect observation is to be performed.

Before carrying out the steps after S2, a preliminary inspection is performed for defects in the silicon crystal plate using an optical surface defect inspection device, SEM, or the like, and an area (TEM observation area) in which sample 1 is formed based on the inspection results. ) May be determined. In this case, the surface of the silicon crystal plate may be marked with a laser or the like at the time of the preliminary inspection so that the position of the processing area (sample forming area) in the steps S2 to S7 can be easily specified. This marking may be formed, for example, around the processing area or at a predetermined distance from the processing area.

Next, using the surface side pattern formed in the step of S3 as a mask, the surface of the silicon crystal plate is etched by reactive ion etching (dry etching) using plasma (S4). FIG. 5 shows a state in which a shape 14 based on the resist pattern 12 is formed on the pedestal 13 as a state of the silicon crystal plate after the step of S4. FIG. 5 shows an example in which one columnar shape 14 is formed on the pedestal 3 for the sake of clarity, but in reality, a thin section of sample 1 of FIGS. 1 and 2 is shown on the upper surface of the pedestal 13. A predetermined number of shapes corresponding to the portions 2 and the support columns 4 are arranged. Although not shown in FIG. 5, the resist 12 of FIG. 4 remains on the upper surface of the columnar shape 14.

Next, a resist is applied to the back surface of the silicon crystal plate (pedestal 13 in the example of FIG. 5) (S5). Next, the resist is patterned by a photolithography technique to form a pattern having a predetermined shape (S6). Here, FIG. 6 shows a state in which the resist pattern 16 is formed on the back surface 15 of the pedestal 13 as the back surface of the silicon crystal plate. Note that FIG. 6 shows one triangular pattern 16 for the sake of clarity, but in reality, in the process of S6, the bottom view shape of the sample 1 of FIGS. 1 and 2 (bottom view shape of the pedestal 3). ) (Hereinafter referred to as a sample bottom view pattern) forms a pattern (hereinafter referred to as a back surface side pattern) arranged in the same number and coordinates as the sample top view pattern. Further, the back surface side pattern is configured to include the frame portion pattern and the connection allowance pattern in addition to a predetermined number of sample bottom surface viewing patterns.

Next, using the back surface side pattern formed in the step of S6 as a mask, the back surface of the silicon crystal plate is etched by reactive ion etching (dry etching) using plasma (S7). FIG. 7 shows a state in which the quadrangular pedestal 13 (see FIG. 6) is processed into a shape 17 based on the resist pattern 16 (see FIG. 6) as the state of the silicon crystal plate after the step of S7. FIG. 7 shows an example in which a triangular pedestal is formed as the shape 17 for the sake of clarity, but in reality, a predetermined number of shapes corresponding to the pedestal 3 of the sample 1 of FIGS. 1 and 2 are arranged. It will be in a state of being. Although not shown in FIG. 7, the resist 16 of FIG. 6 remains on the upper surface of the triangular pedestal 17.

In this way, the shape shown in FIG. 8 is formed by the steps S2 to S7. Note that FIG. 8 shows a silicon piece 18 having a triangular pedestal 17 and a cylindrical portion 14 on the triangular pedestal 17 for the sake of clarity, but in reality, the triangular pedestal 17 is the pedestal of sample 1 of FIGS. 1 and 2. The cylindrical portion 14 is a portion corresponding to the thin section portion 2 and the strut portion 4, and the silicon piece 18 is a portion corresponding to the sample 1. Further, although one silicon piece 18 is shown in FIG. 8, in reality, the silicon pieces 18 are processed into a shape in which a predetermined number of silicon pieces 18 are arranged in a predetermined area of the silicon crystal plate.

Here, FIG. 9 illustrates the state of the silicon crystal plate after the step of S7. The silicon crystal plate 30 of FIG. 9 is formed as a right-angled quadrangular plate, and is arranged in a quadrangular frame portion 19 (outer peripheral portion) provided along the outer periphery of the quadrangle and a predetermined area 21 inside the frame portion 19. It is configured to include a predetermined number of silicon pieces 18 and a connection allowance 20 for holding each silicon piece 18 so as not to be separated from the silicon crystal plate 30. The frame portion 19 is a portion that becomes unnecessary after the step of S12 described later. The width d9 of the frame portion 19 is the same value at any position along the outer periphery of the silicon crystal plate 30, and is set to, for example, about 8 mm.

Further, in the example of FIG. 9, an elongated quadrangular area 21 extending parallel to one side of the silicon crystal plate 30 is shown, but the area 21 has any position and shape as long as it is inside the frame portion 19. You may.

The connection allowance 20 is set to a thickness that can be easily cut when a load is applied. Further, in the area 21, a plurality of silicon pieces 18 are arranged at regular intervals along the longitudinal direction of the area 21. Two or three pieces of silicon 18 are arranged within a range of, for example, 1 mm × 1.5 mm. The silicon piece 18 is formed into a shape larger than that of the sample 1 by the amount of etching in the damage removing steps (steps S9 to S11) described later.

Further, inside the frame portion 19 of the silicon crystal plate 30, an inner unnecessary portion that is connected to each silicon piece 18 via a connection allowance 20 and becomes unnecessary when observing defects is provided. The inner unnecessary portion is connected to the frame portion 19. In this way, each silicon piece 18 is supported by the frame portion 19 via the connection allowance 20 and the inner unnecessary portion.

In FIG. 9, a minute marking 22 is added to the surface of the silicon crystal plate 30. The marking 22 is added by, for example, a laser at the time of a preliminary inspection by an optical surface inspection device or the like, and is so small that it can be finally confirmed with an optical microscope or an actual state microscope. The marking 22 is added in the vicinity of the area 21 in the frame portion 19. The marking 22 makes it possible to easily identify that the silicon piece 18 is formed in the area 21 in the steps S2 to S7.

Returning to the description of FIG. 3, next, the resist remaining on the front surface and the back surface of the silicon crystal plate is peeled off (S8). Next, a damage removing step is carried out to remove the distortion and damage generated on the side surface (hereinafter referred to as the dry etching surface) of each silicon piece that has been dry-etched by the steps S4 and S7. In this damage removal process, the process changes depending on whether or not heat may be applied to the silicon pieces. Specifically, if there is no effect even if heat treatment is applied to the silicon pieces, sacrificial oxidation treatment is performed on each silicon piece. That is, each silicon piece is heat-treated to form a thermal oxide film on the surface including the dry etching surface of the silicon piece. Next, the silicon piece is washed with a chemical solution such as HF (hydrofluoric acid) to remove the thermal oxide film. As a result, the dry-etched surface of the silicon piece is etched, and residual strain and damage on the dry-etched surface can be removed.

Further, the thickness d3 (see FIG. 2) of the thin section of the silicon piece after the sacrificial etching treatment becomes a thickness (for example, about 5 μm) that can be observed by an ultra-high pressure TEM in the step of S12 described later. The amount of processing in the process of S7 and the amount of etching in the process of S8 are determined. If the thickness of the silicon piece is not observable by the ultra-high pressure TEM in one sacrificial oxidation treatment, the sacrificial oxidation treatment (formation of a thermal oxide film and heat by HF, etc.) is performed until the desired thickness is reached. Removal of the oxide film) may be repeated a plurality of times.

Further, from the viewpoint of efficiency improvement, in the step S9, it is preferable that a plurality of silicon pieces produced in the steps S2 to S8 are subjected to sacrificial oxidation treatment at once. In this case, to explain with reference to FIG. 9, the sacrificial oxidation treatment is performed in a state where each silicon piece 18 is connected to the frame portion 19 by the connection allowance 20 (that is, the state of the silicon crystal plate 30).

On the other hand, when it is desired to eliminate the influence of thermal oxidation on the silicon pieces, each silicon is etched by etching (alkaline etching treatment) using an alkaline solution (specifically, KOH or TMAH, etc.) instead of the sacrificial oxidation treatment. Residual strain and damage on the dry-etched surface of one piece are removed and the thickness is adjusted (S10). If necessary, it is better to carry out an ion etching process (for example, Ar ion etching) in which ionized atoms are made to collide with the object to etch the object after the alkali etching.

Further, when the thickness of the silicon piece is sufficiently thin (close to the desired value) even if the alkali etching process is not required, ion etching is performed on each silicon piece after the step of S8 only in the sense of removing damage. Only the processing may be carried out (S11).

From the viewpoint of efficiency, it is preferable that the alkali etching treatment of S10 and the ion etching treatment of S11 are also collectively treated on a plurality of silicon pieces produced in the steps S2 to S8. That is, in the example of FIG. 9, it is preferable to perform the alkali etching treatment of S10 and the ion etching treatment of S11 in the state of the silicon crystal plate 30.

As a result, a large number of samples 1 of FIGS. 1 and 2 are completed at the same time. Then, each sample 1 is observed by an ultra-high pressure TEM (S12). Specifically, to explain with reference to FIG. 9, each silicon piece 18 after the steps of S9 to S11 is used as sample 1 of FIGS. 1 and 2, and each sample 1 (silicon piece 18) is made into a silicon crystal by tweezers or the like. Separate from the plate 30. As shown in FIG. 10, the separated sample 1 is attached to the TEM observation grit 23. Specifically, the lower surface of the pedestal 3 of the sample 1 is adhered to the grit 23 with the silver paste 24. One sample 1 is attached to one grit 23.

After that, the sample 1 and the grit 23 to which the sample 1 is mounted are installed in an ultra-high voltage TEM device, and the sample 1 is observed. At this time, by applying an electron beam of an ultra-high voltage TEM to the rectangular surface of the thin section 2 of the sample 1, the presence or absence of defects and the morphology of the defects are observed in the entire observation region of the thin section 2. In addition, if necessary, elemental analysis of defects is carried out by EDX (Energy Dispersive X-ray spectroscopy, energy dispersive X-ray analysis) or the like.

When a defect is observed by an ultra-high voltage TEM, an enlarged image (detailed defect image) of the defect is acquired and stored, or an elemental analysis result of the defect is acquired and stored (S13). ..

The steps S12 and S13 are sequentially carried out one by one for each of the plurality of prepared samples 1. That is, after performing the steps S12 and S13 on one sample 1, the sample 1 is taken out from the ultra-high voltage TEM device, and the other sample 1 and the grit 23 to which the sample 1 is mounted are put into the ultra-high voltage TEM device. The procedure of installing and observing defects is repeated until the defect observation is completed for all the samples 1. In addition, in the steps of S12 and S13, it may be carried out for all the samples 1 prepared in the steps of S2 to S11, or may be carried out for some samples 1.

Further, the flow of FIG. 3 may be terminated in the step of S13, but a more detailed defect evaluation is carried out using a TEM having an acceleration voltage of less than 1000 kV (for example, 300 kV class) (hereinafter, may be referred to as a normal TEM). You may. In this case, in the steps of S12 and S13, it is preferable to record the defect position in the sample 1 in which the presence of the defect is shown. Specifically, for example, the beam system of the ultra-high voltage TEM used in the step of S12 is sufficiently finely focused and marked around the defect of the sample 1. At this time, it is more preferable to leave an observation image or the like so that the marking position and the defect position of the target can be understood. Further, as the defect position, the coordinates (distance) of the defect based on the mark such as the edge of the sample 1 or the coordinates (distance) of the additional processing area containing the defect may be recorded.

Next, the sample 1 to be evaluated in detail is placed in a FIB device, and the sample 1 is usually shaped to match the TEM by relying on the defect positions (coordinates and markings from the mark) acquired and recorded by the ultra-high voltage TEM. The thin film treatment (additional processing) is performed again so as to be (S14). At this time, the additional processing is performed so that the thickness of the observation region of the sample after the additional processing is, for example, 1 μm or less (for example, about 0.2 μm).

After that, the completed sample is usually placed in a TEM and observed (S15), and if necessary, elemental analysis of defects is performed by EDX or the like (S16). As the normal TEM used at this time, a TEM having a higher resolution and analysis sensitivity than the ultra-high voltage TEM used in the step of S12 is adopted while having a low acceleration voltage. As a result, detailed defect evaluation that cannot be obtained by ultra-high pressure TEM can be obtained.

As described above, in the present embodiment, a large number of samples are mechanically prepared at one time by photolithography and dry etching technology, and all or defects are observed up to the required number of minutes to observe the samples. It is freed from the inefficiency of fabrication and the low defect hit rate, and smooth observation results can be obtained. Specifically, 100 sample processing (processing of preparing 100 samples and observing the prepared 100 samples by TEM) was required in the past at full operation for 17 days. The process is completed in 5 days, which is one-third of the time.

In addition, by using photolithography and dry etching technology, it is possible to easily prepare a sample with a thicker and wider observation area than before, and by observing this sample with an ultra-high pressure TEM, from 1 × 10 ⁸ cm ^-3 . The defect hit rate can be improved even for defects existing at a low density. For example, it is assumed that the thickness of the sample is set to about 5 μm, which is 50 times that of the conventional one, and 50 to 100 sheets are cut out from one wafer (silicon crystal plate). In this case, the defect can be measured with an efficiency of 50 × 50 to 50 × 100 times, that is, 2500 to 5000 times that of the conventional method.

The present invention is not limited to the above embodiment. The above embodiment is an example, and any one having substantially the same configuration as the technical idea described in the claims of the present invention and exhibiting the same effect and effect may be used. It is included in the technical scope of the present invention.

For example, in the above embodiment, an example in which the prepared sample is observed using an ultra-high voltage TEM is shown. However, since ultra-high voltage TEM is expensive, it is owned only by some research institutes and universities, and its use is restricted. Therefore, it is possible to substitute this process with an optical defect inspection device instead of the ultra-high voltage TEM. That is, a plurality of samples for an optical defect inspection device are prepared at once from one silicon crystal plate using photolithography and dry etching technology, and an optical defect inspection device is used for each of the prepared samples. You may perform defect inspection.

In FIG. 3, the steps S1 to S11 correspond to the sample preparation method of the present invention. The steps S2 to S8 correspond to the processing steps of the present invention. The steps S2 to S4 correspond to the surface processing step of the present invention. The steps S5 to S7 correspond to the back surface processing step. The steps S9 to S11 correspond to the removal steps. The steps S12 to S16 correspond to the defect observation method of the present invention. The steps S12 and S13 correspond to the ultrahigh pressure TEM observation step and the recording step. The process of S14 corresponds to the additional processing process. The steps S15 and S16 correspond to the detailed evaluation step.

1 sample 2 flakes 3 pedestal

Claims (17)

This is a method for producing a sample for defect observation by a transmission electron microscope in which a pedestal and a flaky portion constituting a defect observation region are integrally formed on the pedestal.
From one silicon crystal plate , a plurality of silicon pieces having a shape corresponding to the sample are arranged in the in-plane direction inside unnecessary parts that are not required when observing defects, and a connection allowance that allows each silicon piece to be cut. The silicon piece array shape supported by the unnecessary part is formed by photolithography and etching based on the resist pattern formed by the photolithography .
The processing process is
A surface processing step of forming a top view shape of the silicon piece array shape by processing the surface of the silicon crystal plate by photolithography and etching.
A back surface processing step of forming a bottom view shape of the silicon piece array shape by processing the back surface of the silicon crystal plate by photolithography and etching.
Sample preparation method, characterized in that it comprises a. Sample Preparation The method of claim 1, wherein the forming the silicon piece arrangement shape containing the silicon pieces or 10 pieces in the above processing step. The sample preparation method according to claim 1 or 2, wherein in the processing step, the silicon piece array shape containing 50 to 100 pieces of the silicon piece is formed. The flakes are provided so that the surface forming the rectangular observation area is perpendicular to the upper surface of the pedestal.
The sample has strut portions provided on both the left and right sides of the flaky portion.
The sample preparation method according to any one of claims 1 to 3, wherein the pedestal, the thin section portion, and the strut portion are integrally formed. The sample preparation method according to any one of claims 1 to 4 , further comprising a removing step of removing residual strain on the etching surface of the silicon piece obtained in the processing step. The sample preparation method according to claim 5 , wherein in the removal step, a plurality of the silicon pieces are collectively processed. The sample preparation method according to claim 5 or 6 , wherein the removing step is a step of sacrificing oxidation treatment of the etched surface. The sample preparation method according to claim 5 or 6 , wherein the removing step is a step of performing an ion etching treatment or an alkali etching treatment on the etched surface. The silicon piece after said removing step, thickness and performing the processing step and the removing step so as to have the thin section of 2-6 [mu] m,
The sample preparation method according to any one of claims 5 to 8 , wherein the silicon piece after the removal step is used as a sample for observing with a transmission electron microscope having an acceleration voltage of 1000 kV or more. Sample Preparation The method of claim 9 wherein the louver part is characterized by having a rectangular said observation region of 100~200μm × 1000~2000μm. The sample preparation method according to any one of claims 1 to 10 , wherein the sample is a sample for observing defects existing at a density lower than ^{1 × 10 8} cm ^-3. A defect observation method, characterized in that defect observation is performed on a plurality of samples obtained by the sample preparation method according to any one of claims 1 to 11. The defect observation method according to claim 12 , further comprising an ultra-high voltage TEM observation step of observing defects in a plurality of the samples with a transmission electron microscope having an acceleration voltage of 1000 kV or more. For the sample showing the presence of defects in the ultra-high pressure TEM observation step, defect evaluation is performed with a transmission electron microscope having higher resolution and analysis sensitivity than the transmission electron microscope used in the ultra-high pressure TEM observation step. The defect observation method according to claim 13 , further comprising a detailed evaluation step to be performed. The detailed evaluation step is a step of observing defects with a transmission electron microscope having an acceleration voltage of less than 1000 kV.
The sample is provided with an additional processing step of processing the sample showing the presence of defects in the ultra-high voltage TEM observation step so as to have a shape suitable for the transmission electron microscope used in the detailed evaluation step.
The defect observation method according to claim 14 , wherein in the detailed evaluation step, defect evaluation is performed on a sample obtained in the additional processing step. A recording step of recording the defect position in the sample in which the presence of a defect is shown in the ultra-high pressure TEM observation step is provided.
The defect observation method according to claim 15 , wherein in the additional processing step, processing is performed based on the defect position recorded in the recording process. In the recording step, the position corresponding to the defect position in the sample in which the presence of a defect is shown in the ultra-high pressure TEM observation step is marked by the electron beam of the transmission electron microscope used in the ultra-high pressure TEM observation step. The defect observation method according to claim 16 , wherein the defect is observed.

Images