SiC substrate and SiC composite substrate
By analyzing a biaxially oriented SiC layer using photoluminescence to define specific conditions, the SiC substrate achieves a low TSD density, addressing macro defects and enhancing substrate stability for high-voltage applications.
Patent Information
- Application Number
- JP2023514848
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2021-10-22
- Filing Date
- 2022-10-19
- Publication Date
- 2024-11-20
- Estimated Expiration
- 2042-10-19
AI Technical Summary
Conventional SiC substrates face challenges in achieving low threading screw dislocation (TSD) density due to macro defects such as solvent inclusions and heterogeneous polymorphisms, and TSD density on the substrate surface is difficult to significantly reduce even with epitaxial layer formation.
A biaxially oriented SiC layer is analyzed using photoluminescence (PL) to define specific conditions, including maximum and minimum point ratios and distances, to achieve a low TSD density, and is manufactured without solution growth methods.
The SiC substrate achieves a very low TSD density on the surface, reducing the occurrence of macro defects and providing a stable substrate for high-voltage applications.
Smart Images

Figure 0007587688000002 
Figure 0007587688000003 
Figure 0007587688000004
Abstract
Description
[Technical field]
[0001] The present invention relates to a SiC substrate and a SiC composite substrate. [Background technology]
[0002] Silicon carbide (SiC) has been attracting attention as a wide band gap material that can control high voltage and high power with low loss. In particular, in recent years, power semiconductor devices using SiC materials (SiC power devices) are more compact, consume less power, and are more efficient than those using Si semiconductors, and are therefore expected to be used in a variety of applications. For example, the use of SiC power devices can reduce the size and increase the efficiency of converters, inverters, and on-board chargers for electric vehicles (EVs) and plug-in hybrid vehicles (PHEVs).
[0003] In order to use SiC power devices in high-voltage applications, it is necessary to minimize defects in SiC wafers and suppress deterioration of device characteristics in order to handle large currents. In this regard, a solution growth method is known as a method for reducing defects in wafers, particularly threading screw dislocations (TSDs) that are believed to cause breakdown voltage deterioration. For example, Patent Document 1 (JP Patent Publication 2014-043369) discloses a method for producing a SiC single crystal, which includes a step of forming macrosteps made of SiC single crystals and having a height of more than 70 nm on the (0001) plane of a SiC seed crystal to obtain a second seed crystal, and a step of growing a SiC single crystal on the (0001) plane of the second seed crystal in a reaction atmosphere containing Si and C to advance the macrosteps on the threading screw dislocations in the second seed crystal. It is described that this production method can produce a SiC single crystal with few threading screw dislocations. However, when a SiC substrate is produced by the solution growth method, there are problems such as the tendency for macrodefects such as solvent inclusions and heterogeneous polymorphs due to surface roughness to occur.
[0004] On the other hand, as a method other than the solution growth method, a method of forming a SiC epitaxial layer by chemical vapor deposition is known. For example, Non-Patent Document 1 (Lixia Zhao et al. "Surface defects in 4H-SiC homoepitaxial layers" Nanotechnology and Precision Engineering 3 (2020) 229-234) discloses that a SiC epitaxial layer was obtained by chemical vapor deposition and the morphology and structure of the surface defects were investigated. In general, the TSD density of a SiC substrate is 300 to 500 cm -2 It is described that: [Prior art documents] [Patent documents]
[0005] [Patent Document 1] JP 2014-043369 A [Non-patent literature]
[0006] [Non-Patent Document 1] Lixia Zhao et al. "Surface defects in 4H-SiC homoepitaxial layers" Nanotechnology and Precision Engineering 3 (2020) 229-234 [Non-Patent Document 2] Toru Ujihara et al. “Conversion Mechanism of Threading Screw Dislocation during SiC Solution Growth” Materials Science Forum Vols. 717-720, pp. 351-354 (2012) Summary of the Invention
[0007] However, according to the disclosure of Non-Patent Document 1, most of the TSDs on the substrate are propagated to the epitaxial layer, so the TSD density on the substrate surface is set to 300 cm -2 Therefore, it is desirable to further reduce the TSD density on the SiC substrate surface.
[0008] The present inventors have now discovered that a biaxially textured SiC layer that satisfies certain conditions when analyzed by photoluminescence (PL) provides a SiC substrate with an extremely small TSD density on the surface.
[0009] It is therefore an object of the present invention to provide a SiC substrate having a very low density of TSDs on its surface.
[0010] According to the present invention, the following aspects are provided. [Aspect 1] A SiC substrate having a biaxially oriented SiC layer, the surface of the biaxially oriented SiC layer being analyzed by photoluminescence (PL) to obtain a graph plotting the distance (μm) in the [11-20] direction on the horizontal axis and the PL intensity I on the vertical axis, (i) the graph has a shape in which maximum and minimum points are repeated, wherein a maximum point is defined as a point that gives a PL intensity I higher than each of the PL intensities I at a total of six points consisting of the first closest point, the second closest point, and the third closest point on the left and right of the maximum point, and a minimum point is defined as a point that gives a PL intensity I lower than each of the PL intensities I at a total of six points consisting of the first closest point, the second closest point, and the third closest point on the left and right of the minimum point, (ii) A certain maximum point P M The maximum value of PL intensity I at M The distance in the horizontal direction is longer than that of the maximum point P M and the closest minimum point P m When the minimum value of PL intensity I is m, the ratio M / m is 1.05 or more, (iii) The maximum point P M and the minimum point P ma distance L in the [11-20] direction between the SiC substrate and the surface of the substrate is 15 to 150 μm. [Aspect 2] 2. The SiC substrate according to aspect 1, wherein the distance L is 50 to 150 μm. [Aspect 3] A SiC substrate having a biaxially oriented SiC layer, wherein an image obtained by analyzing the surface of the biaxially oriented SiC layer by photoluminescence (PL) is processed by a Prewitt filter in the [11-20] direction of the image, and the distance (μm) in the [11-20] direction is plotted as the horizontal axis, and the PL intensity I F If you get a graph plotted on the vertical axis, (i) The graph has a shape in which a plurality of maximum points are repeated, and the maximum point is a total of six PL intensities I F PL intensity I higher than each of F is defined as the point that gives (ii) A certain maximum point P M1 and the maximum point P M1 The distance in the horizontal direction is longer than that of the maximum point P M1 and the closest maximum point P M2 The distance L in the [11-20] direction F A SiC substrate having a thickness of 30 to 300 μm. [Aspect 4] The biaxially oriented SiC layer is oriented in the c-axis direction and the a-axis direction, and the depth (μm) from the (0001) plane, which is the substrate surface, to an arbitrary (000-1) plane in the biaxially oriented SiC layer is taken as the horizontal axis, and the threading screw dislocation (TSD) density (cm -2 ) on the vertical axis, the graph shows that the TSD density decreases with a constant gradient a as the depth decreases, and the absolute value of the gradient a is 5.0 cm -2 4. The SiC substrate according to any one of aspects 1 to 3, comprising a TSD gradient region having a TSD gradient of 1 / μm or more. [Aspect 5] The biaxially oriented SiC layer is doped with boron at 1.0×10 16 ~1.0×10 17 atoms / cm 35. The SiC substrate according to any one of aspects 1 to 4, comprising: [Aspect 6] The absolute value of the slope a is 5.0 to 25 cm -2 5. The SiC substrate of embodiment 4, wherein the SiC surface roughness is 1 / μm. [Aspect 7] A SiC composite substrate comprising: a SiC single crystal substrate; and a SiC substrate according to any one of aspects 1 to 6 on the SiC single crystal substrate. [Brief description of the drawings]
[0011] [Figure 1] 1 is a vertical cross-sectional view of a SiC composite substrate 10. FIG. [Diagram 2] 1 is a manufacturing process diagram of the SiC composite substrate 10. FIG. [Diagram 3] FIG. 1 is a schematic cross-sectional view showing the configuration of an aerosol deposition (AD) device. [Figure 4] 1 is a photoluminescence (PL) imaging image of the (0001) plane of a biaxially oriented SiC layer obtained in Example 1. [Diagram 5] This is a graph created based on the PL imaging image of the (0001) plane of the biaxially oriented SiC layer obtained in Example 1, in which the distance (μm) in the [11-20] direction is plotted on the horizontal axis and the PL intensity I is plotted on the vertical axis. [Figure 6] 1 is an image obtained by processing the PL imaging image of the (0001) plane of the biaxially oriented SiC layer obtained in Example 1 with a Prewitt filter. [Figure 7] This graph was created based on an image obtained by processing the PL imaging image of the (0001) plane of the biaxially oriented SiC layer obtained in Example 1 with a Prewitt filter, and plots the distance (μm) in the [11-20] direction on the horizontal axis and the PL intensity IF on the vertical axis. [Figure 8] 1 is a graph obtained in Example 1, in which the horizontal axis represents the depth (μm) from the (0001) plane to an arbitrary (000-1) plane of the biaxially oriented SiC layer, and the vertical axis represents the threading screw dislocation (TSD) density (cm−2). [Figure 9]1 is an optical microscope image showing the step-terrace structure of the surface of a SiC substrate obtained by a solution growth method, as published in Non-Patent Document 2. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0012] SiC Substrate The SiC substrate according to the present invention is a SiC substrate having a biaxially oriented SiC layer. When the surface of this biaxially oriented SiC layer is analyzed by photoluminescence (PL) to obtain a graph plotting the distance (μm) in the [11-20] direction on the horizontal axis and the PL intensity I on the vertical axis, the biaxially oriented SiC layer has a characteristic shape (i) in which maximum and minimum points are repeated, as shown in FIG. 5 described later. Here, a maximum point is defined as a point that gives a PL intensity I higher than each of the PL intensities I of a total of six points consisting of the first closest point, the second closest point, and the third closest point on the left and right of the maximum point, and a minimum point is defined as a point that gives a PL intensity I lower than each of the PL intensities I of a total of six points consisting of the first closest point, the second closest point, and the third closest point on the left and right of the maximum point. This graph satisfies the following conditions: (ii) A certain maximum point P M The maximum value of PL intensity I at M The distance along the horizontal axis is longer than that of the maximum point P M and the closest minimum point P m When the minimum value of PL intensity I is m, the ratio M / m is 1.05 or more, and (iii) The above maximum point P M and the above minimum point P m The distance L in the [11-20] direction from the Alternatively, the biaxially oriented SiC layer is one that satisfies the following equation: In an image obtained by analyzing the surface of the biaxially oriented SiC layer by PL, the image is processed by a Prewitt filter in the [11-20] direction, the horizontal axis is the distance (μm) in the [11-20] direction, and the PL intensity I FWhen a graph is obtained by plotting PL intensity I on the vertical axis, as shown in FIG. 7 described later, for example, (iv) a shape having a plurality of repeated maximum points is obtained. Here, a maximum point is a total of six points of PL intensity I consisting of the first closest point, the second closest point, and the third closest point on the left and right of the maximum point. F PL intensity I higher than each of F The graph is defined as the point that satisfies the following conditions: (v) A certain maximum point P M1 And its maximum point P M1 The distance along the horizontal axis is longer than that of the maximum point P M1 and the closest maximum point P M2 Distance L in the [11-20] direction from F is 30 to 300 μm It satisfies the following.
[0013] In this way, when a SiC substrate with a biaxially oriented SiC layer is analyzed by PL, a biaxially oriented SiC layer that satisfies certain conditions can provide a SiC substrate with an extremely small TSD density on the surface. In addition, since such a SiC substrate can be fabricated without going through a solution growth method, macroscopic defects are unlikely to occur on the substrate surface.
[0014] As mentioned above, conventional SiC substrates, for example, when fabricated by solution growth, have problems such as the tendency for macro defects such as solvent inclusions and heterogeneous polymorphism due to surface roughness to occur. Even when forming an SiC epitaxial layer by chemical vapor deposition, most of the TSDs on the substrate are propagated to the epitaxial layer, so the TSD density on the substrate surface must be kept below 300 cm. -2 In this regard, the SiC substrate of the present invention can advantageously solve the above problem.
[0015] In the biaxially oriented SiC layer constituting the SiC substrate of the present invention, when the surface of the biaxially oriented SiC layer is analyzed by PL and a graph is obtained in which the distance (μm) in the [11-20] direction is plotted on the horizontal axis and the PL intensity I is plotted on the vertical axis, a certain maximum point P MThe maximum value of PL intensity I at M The distance in the [11-20] direction is longer than that of the maximum point P M and the closest minimum point P m When the minimum value of PL intensity I is m, the ratio M / m is 1.05 or more (hereinafter referred to as condition (ii)). There is no particular upper limit to this ratio M / m, but it is preferably 1.05 to 1.80.
[0016] In addition, in the above graph in which the surface of the biaxially oriented SiC layer was analyzed by PL, the maximum point P M and the minimum point P m The distance L in the [11-20] direction between the first and second crystals is 15 to 150 μm (hereinafter referred to as condition (iii)), preferably 50 to 150 μm, more preferably 50 to 120 μm, and further preferably 50 to 100 μm.
[0017] Incidentally, the graph typically has a regular pattern in which a waveform including maximum and minimum points is repeated at a constant period, as shown in Fig. 5 described later. Therefore, if at least one of the maximum points (and the corresponding minimum point) existing on a predetermined measurement line segment (e.g., a 500 μm line segment) parallel to the [11-20] direction satisfies the above conditions (ii) and (iii), the SiC substrate is deemed to satisfy the above conditions (ii) and (iii). However, it is more preferable that at least two of the maximum points (and the corresponding minimum point) satisfy the above conditions (ii) and (iii), and even more preferable that at least three of the maximum points satisfy the above conditions (ii) and (iii).
[0018] Alternatively, the biaxially oriented SiC layer constituting the SiC substrate of the present invention may be obtained by analyzing the surface of the biaxially oriented SiC layer by PL, processing the image in the [11-20] direction with a Prewitt filter, and plotting the distance (μm) in the [11-20] direction as the horizontal axis and the PL intensity I F When a graph is obtained by plotting the vertical axis, a certain maximum point P M1 And its maximum point P M1The distance in the [11-20] direction is longer than that of the maximum point P M1 and the closest maximum point P M2 Distance L in the [11-20] direction from F is 30 to 300 μm (hereinafter referred to as condition (v)), preferably 100 to 300 μm, more preferably 100 to 240 μm, and further preferably 100 to 200 μm.
[0019] Incidentally, the graph obtained by processing with the Prewitt filter has a plurality of repeated maximum points, as shown in FIG. 7, which will be described later. Therefore, among the plurality of maximum points existing on a predetermined measurement line segment (e.g., a 500 μm line segment) parallel to the [11-20] direction, the combination of maximum points (P M1 and P M2 ) is considered to satisfy the above condition (v). However, it is more preferable for the SiC substrate to have at least two of these combinations, and even more preferable for the substrate to have three of these combinations.
[0020] The biaxially oriented SiC layer is preferably oriented in the c-axis direction and the a-axis direction. The SiC substrate is preferably composed of a biaxially oriented SiC layer. The biaxially oriented SiC layer may be a SiC single crystal or a mosaic crystal as long as it is oriented in the two axial directions of the c-axis and the a-axis. The mosaic crystal is a collection of crystals that do not have clear grain boundaries but have slightly different orientations in one or both of the c-axis and the a-axis. The method for evaluating the orientation is not particularly limited, but known analytical methods such as EBSD (Electron Back Scatter Diffraction Patterns) and X-ray pole figures can be used. For example, when the EBSD method is used, inverse pole figure mapping of the surface (plate surface) of the biaxially oriented SiC layer or a cross section perpendicular to the plate surface is measured. In the obtained inverse pole figure mapping, when the following four conditions are satisfied, (A) the SiC layer is oriented in a specific direction (first axis) approximately normal to the plate surface, (B) the SiC layer is oriented in a specific direction (second axis) approximately in the plate surface direction perpendicular to the first axis, (C) the inclination angle from the first axis is distributed within ±10°, and (D) the inclination angle from the second axis is distributed within ±10°, the SiC layer can be defined as being oriented in two axes, the approximately normal direction and the approximately plate surface direction. In other words, when the above four conditions are satisfied, the SiC layer is determined to be oriented in two axes, the c-axis and the a-axis. For example, when the approximately normal direction of the plate surface is oriented to the c-axis, the approximately in-plane direction may be oriented in a specific direction (e.g., the a-axis) perpendicular to the c-axis. The biaxially oriented SiC layer may be oriented in two axes, the approximately normal direction and the approximately in-plane direction, but it is preferable that the approximately normal direction is oriented to the c-axis. The smaller the tilt angle distribution in the approximately normal direction and / or the approximately in-plane direction, the smaller the mosaic property of the biaxially oriented SiC layer, and the closer it is to zero, the closer it is to a single crystal. Therefore, from the viewpoint of the crystallinity of the biaxially oriented SiC layer, it is preferable that the tilt angle distribution is small in both the approximately normal direction and the approximately in-plane direction, for example, ±5° or less is more preferable, and ±3° or less is even more preferable.
[0021] For example, as shown in FIG. 8 described later, a biaxially oriented SiC layer oriented in the c-axis and a-axis directions is plotted as a function of the depth (μm) from the (0001) plane, which is the substrate surface, to an arbitrary (000-1) plane, and the TSD density (cm-2 When a graph is obtained by plotting the TSD density on the vertical axis, it is preferable that the graph at least partially includes a TSD gradient region. This TSD gradient region effectively contributes to realizing a SiC substrate with a very small TSD density on the surface. This TSD gradient region is a region in which the TSD density decreases with a constant gradient a as the depth decreases. The absolute value of the gradient a is 5.0 cm -2 / μm or more, preferably 5.0 to 25 cm -2 / μm, more preferably 10 to 25 cm -2 / μm, more preferably 15 to 25 cm -2 / μm. This slope a can be calculated, for example, by finding an approximation line by the least squares method in the TSD slope region of the above graph. "The depth from the (0001) plane, which is the substrate surface, to any (000-1) plane" means, in the biaxially oriented SiC layer on the SiC single crystal substrate, the depth from the (0001) plane of the biaxially oriented SiC layer surface toward the back surface of the biaxially oriented SiC layer on the SiC single crystal substrate.
[0022] The biaxially oriented SiC layer is doped with boron at 1.0×10 16 ~1.0×10 17 atoms / cm 3 and more preferably at a concentration of 1.0×10 16 ~9.5×10 16 atoms / cm 3 In this way, by controlling the boron content, the distribution of PL intensity can be favorably controlled in the above-mentioned PL graph, and a SiC substrate with a reduced TSD density on the substrate surface can be effectively obtained.
[0023] SiC composite substrate The SiC substrate of the present invention may be a free-standing substrate consisting of only a SiC substrate, or may be in the form of a SiC composite substrate. The SiC composite substrate may include a SiC single crystal substrate and the above-mentioned SiC substrate on the SiC single crystal substrate.
[0024] The SiC single crystal substrate is typically a layer made of a SiC single crystal and has a crystal growth surface. The polytype, off-angle, and polarity of the SiC single crystal are not particularly limited, but the polytype is preferably 4H or 6H, the off-angle is preferably 0.1 to 12° from the
[0001] axis of the single crystal SiC, and the polarity is preferably Si-face. More preferably, the polytype is 4H, the off-angle is 1 to 5° from the
[0001] axis of the single crystal SiC, and the polarity is Si-face.
[0025] As described above, the SiC substrate of the present invention may be in the form of a free-standing substrate of a biaxially oriented SiC layer alone, or in the form of a SiC composite substrate accompanied by a SiC single crystal substrate. Therefore, the biaxially oriented SiC layer may be finally separated from the SiC single crystal substrate as necessary. The separation of the SiC single crystal substrate may be performed by a known method, and is not particularly limited. For example, a method of separating the biaxially oriented SiC layer by a wire saw, a method of separating the biaxially oriented SiC layer by electric discharge machining, a method of separating the biaxially oriented SiC layer by using a laser, etc. may be mentioned. In addition, in the case of a form in which the biaxially oriented SiC layer is epitaxially grown on the SiC single crystal substrate, the biaxially oriented SiC layer may be placed on another support substrate after separating the SiC single crystal substrate. The material of the other support substrate is not particularly limited, but a suitable one may be selected from the viewpoint of material properties. For example, from the viewpoint of thermal conductivity, a metal substrate such as Cu, a ceramic substrate such as SiC or AlN, etc. may be mentioned.
[0026] Manufacturing method for SiC composite substrate The SiC composite substrate comprising the SiC substrate of the present invention can be preferably manufactured by (a) forming a predetermined oriented precursor layer on the SiC single crystal substrate, (b) heat-treating the oriented precursor layer on the SiC single crystal substrate to convert at least the portion near the SiC single crystal substrate into a SiC substrate (biaxially oriented SiC layer), and (c) optionally performing processing such as grinding or polishing to expose the surface of the biaxially oriented SiC layer. However, there is no limitation on the manufacturing method of the SiC composite substrate, and it is sufficient to obtain a SiC substrate that gives a specific analysis result as described above when the SiC substrate comprising the biaxially oriented SiC layer is analyzed by PL. For example, the manufacturing method may be a gas phase method such as CVD or sublimation method, a liquid phase method such as solution method, or a solid phase method utilizing grain growth. The distribution of PL intensity in the graph obtained by PL can be controlled by controlling the heat treatment conditions in (b) above, or by controlling the amount of boron or a boron compound (e.g., boron carbide, etc.) added when forming the oriented precursor layer in (a) above. According to this manufacturing method, when a SiC substrate having a biaxially oriented SiC layer is analyzed by PL, it is possible to produce a SiC substrate that gives the specific analysis results described above, and it is possible to significantly reduce the TSD density on the surface of the SiC substrate or a SiC composite substrate using the same.
[0027] A preferred method for producing a SiC composite substrate will now be described. Fig. 1 is a longitudinal cross-sectional view of a SiC composite substrate 10 (a cross-sectional view of the SiC composite substrate 10 cut longitudinally along a plane including the central axis of the SiC composite substrate 10), and Fig. 2 is a process diagram for producing the SiC composite substrate 10.
[0028] As shown in FIG. 1, a SiC composite substrate 10 of this embodiment includes a SiC single crystal substrate 20 and a SiC substrate 30 (corresponding to the SiC substrate of the present invention) on the SiC single crystal substrate.
[0029] (a) Formation process of alignment precursor layer 2(a), the oriented precursor layer 40 becomes a SiC substrate (biaxially oriented SiC layer) 30 by heat treatment described later. In the step of forming the oriented precursor layer 40, the oriented precursor layer 40 is formed on the crystal growth surface of the SiC single crystal substrate 20.
[0030] The method of forming the orientation precursor layer 40 can employ a known method. The method of forming the orientation precursor layer 40 can be, for example, a solid-phase film formation method such as an AD (aerosol deposition) method or an HPPD (supersonic plasma particle deposition) method, a vapor-phase film formation method such as a sputtering method, a vapor deposition method, a sublimation method, or various CVD (chemical vapor deposition) methods, or a liquid-phase film formation method such as a solution growth method, and a method of forming the orientation precursor layer 40 directly on the SiC single crystal substrate 20 can be used. As the CVD method, for example, a thermal CVD method, a plasma CVD method, a mist CVD method, an MO (metal organic) CVD method, or the like can be used. In addition, a method of using a polycrystalline body prepared in advance by a sublimation method, various CVD methods, sintering, or the like as the orientation precursor layer 40 and placing it on the SiC single crystal substrate 20 can also be used. Alternatively, a method of preparing a molded body of the orientation precursor layer 40 in advance and placing this molded body on the SiC single crystal substrate 20 may be used. Such an orientation precursor layer 40 may be a tape-molded body produced by tape casting, or may be a green compact produced by pressure molding using a uniaxial press or the like.
[0031] In forming these oriented precursor layers 40, it is preferable that the raw material of the oriented precursor layer 40 contains a boron compound. The boron compound is not particularly limited, but may be boron carbide or the like. As described above, the concentration of boron in the finally obtained biaxially oriented SiC layer can be controlled by controlling the amount of boron compound added, for example.
[0032] In addition, in the method of forming the oriented precursor layer 40 directly on the SiC single crystal substrate 20, when various CVD methods, sublimation methods, solution growth methods, etc. are used, epitaxial growth may occur on the SiC single crystal substrate 20 without going through the heat treatment process described below, and the SiC substrate 30 may be formed. However, the oriented precursor layer 40 is in an unoriented state when formed, that is, it is an amorphous or unoriented polycrystal, and it is preferable to orient it using the SiC single crystal as a seed in the subsequent heat treatment process. In this way, it is possible to effectively reduce crystal defects that reach the surface of the SiC substrate 30. Although the reason for this is unclear, it is believed that the re-arrangement of the crystal structure of the oriented precursor layer in the solid phase once formed as a film using the SiC single crystal as a seed may also be effective in eliminating crystal defects. Therefore, when various CVD methods, sublimation methods, solution growth methods, etc. are used, it is preferable to select conditions that do not cause epitaxial growth in the formation process of the oriented precursor layer 40. However, since it is difficult to sufficiently reduce the TSD density with this alone, it is preferable to control the heat treatment conditions such as the firing temperature and the amount of boron added.
[0033] However, the method of forming the oriented precursor layer 40 directly on the SiC single crystal substrate 20 by the AD method or various CVD methods, or the method of placing a polycrystalline body separately prepared by sublimation, various CVD methods, sintering, or the like on the SiC single crystal substrate 20 is preferable. By using these methods, it is possible to form the oriented precursor layer 40 in a relatively short time. The AD method is particularly preferable because it does not require a high vacuum process and the film formation speed is relatively fast. In the method of using a polycrystalline body prepared in advance as the oriented precursor layer 40, it is necessary to take measures such as making the surface of the polycrystalline body sufficiently smooth in order to increase the adhesion between the polycrystalline body and the SiC single crystal substrate 20. For this reason, from the viewpoint of cost, the method of directly forming the oriented precursor layer 40 is preferable. In addition, the method of placing a molded body prepared in advance on the SiC single crystal substrate 20 is also preferable as a simple method, but since the oriented precursor layer 40 is made of powder, a sintering process is required in the heat treatment step described later. Both methods can use known conditions, but below we will describe a method of forming an orientation precursor layer 40 directly on a SiC single crystal substrate 20 by the AD method or thermal CVD method, and a method of placing a previously prepared molded body on the SiC single crystal substrate 20.
[0034] The AD method is a technique in which fine particles or fine particle raw material is mixed with gas to form an aerosol, and the aerosol is sprayed at high speed from a nozzle to collide with a substrate to form a coating, and has the characteristic of being able to form a coating at room temperature. An example of a film forming apparatus (AD apparatus) used in such an AD method is shown in FIG. 3. The AD apparatus 50 shown in FIG. 3 is configured as an apparatus used in the AD method in which raw material powder is sprayed onto a substrate under an atmosphere with a pressure lower than atmospheric pressure. This AD apparatus 50 includes an aerosol generating section 52 that generates an aerosol of raw material powder containing raw material components, and a film forming section 60 that sprays the raw material powder onto a SiC single crystal substrate 20 to form a film containing the raw material components. The aerosol generating section 52 includes an aerosol generating chamber 53 that contains raw material powder and generates an aerosol by receiving a carrier gas from a gas cylinder (not shown), a raw material supply pipe 54 that supplies the generated aerosol to the film forming section 60, and a vibrator 55 that applies vibrations at a frequency of 10 to 100 Hz to the aerosol generating chamber 53 and the aerosol therein. The film formation unit 60 includes a film formation chamber 62 for spraying an aerosol onto the SiC single crystal substrate 20, a substrate holder 64 disposed inside the film formation chamber 62 for fixing the SiC single crystal substrate 20, and an XY stage 63 for moving the substrate holder 64 in the X-axis and Y-axis directions. The film formation unit 60 also includes an injection nozzle 66 having a slit 67 formed at its tip for spraying the aerosol onto the SiC single crystal substrate 20, and a vacuum pump 68 for reducing the pressure in the film formation chamber 62. The injection nozzle 66 is attached to the tip of the raw material supply pipe 54.
[0035] It is known that the AD method may cause pores in the film or the film may become a powder compact depending on the film formation conditions. For example, it is easily affected by the collision speed of the raw material powder with the substrate, the particle size of the raw material powder, the agglomeration state of the raw material powder in the aerosol, the amount sprayed per unit time, etc. The collision speed of the raw material powder with the substrate is affected by the pressure difference between the film formation chamber 62 and the spray nozzle 66, the opening area of the spray nozzle, etc. Therefore, in order to obtain a dense oriented precursor layer, it is necessary to appropriately control these factors.
[0036] In the thermal CVD method, a known film forming apparatus such as a commercially available product can be used. The source gas is not particularly limited, but silicon tetrachloride (SiCl4) gas or silane (SiH4) gas can be used as a Si source, and methane (CH4) gas or propane (C3H8) gas can be used as a C source. The film forming temperature is preferably 1000 to 2200°C, more preferably 1100 to 2000°C, and even more preferably 1200 to 1900°C.
[0037] It is known that when the oriented precursor layer 40 is formed on the SiC single crystal substrate 20 by using the thermal CVD method, epitaxial growth may occur on the SiC single crystal substrate 20 to form the SiC substrate 30. However, the oriented precursor layer 40 is not oriented when it is produced, that is, it is amorphous or non-oriented polycrystalline, and it is preferable to use the SiC single crystal as a seed crystal during the heat treatment process to cause crystal rearrangement. It is known that the formation of an amorphous or polycrystalline layer on the SiC single crystal by using the thermal CVD method is affected by the film formation temperature, the gas flow rate of the Si source and the C source and their ratio, the film formation pressure, etc. The film formation temperature has a large effect, and from the viewpoint of forming an amorphous or polycrystalline layer, a lower film formation temperature is preferable, preferably less than 1700°C, more preferably 1500°C or less, and even more preferably 1400°C or less. However, if the film formation temperature is too low, the film formation rate itself also decreases, so from the viewpoint of the film formation rate, a higher film formation temperature is preferable.
[0038] When a pre-prepared molded body is used as the orientation precursor layer 40, it can be produced by molding raw material powder of the orientation precursor. For example, when press molding is used, the orientation precursor layer 40 is a press-molded body. The press-molded body can be produced by press molding the raw material powder of the orientation precursor based on a known method, for example, by putting the raw material powder into a mold and pressing it under a pressure of preferably 100 to 400 kgf / cm. 2 , more preferably 150 to 300 kgf / cm 2The forming method is not particularly limited, and in addition to press forming, tape forming, extrusion forming, casting, doctor blade method, and any combination thereof can be used. For example, when tape forming is used, it is preferable to add additives such as binder, plasticizer, dispersant, and dispersion medium to the raw material powder to make a slurry, and then discharge and form the slurry into a sheet by passing it through a thin slit-shaped discharge port. There is no limitation on the thickness of the molded body formed into a sheet, but from the viewpoint of handling, it is preferable to have a thickness of 5 to 500 μm. In addition, if a thick orientation precursor layer is required, a large number of these sheet molded bodies can be stacked and used to the desired thickness. These molded bodies are then heat-treated on the SiC single crystal substrate 20, and the portion near the SiC single crystal substrate 20 becomes the SiC substrate 30. In such a method, it is necessary to sinter the molded body in the heat treatment process described later. It is preferable to form the SiC substrate 30 after the molded body is sintered and integrated with the SiC single crystal substrate 20 as a polycrystalline body. If the molded body is not sintered, epitaxial growth using the SiC single crystal as a seed may not occur sufficiently. For this reason, the molded body may contain additives such as sintering aids in addition to the SiC raw material.
[0039] (b) Heat treatment process As shown in FIG. 2(b), in the heat treatment step, the SiC substrate 30 is generated by heat-treating a laminate in which the orientation precursor layer 40 is laminated or placed on the SiC single crystal substrate 20. The heat treatment method is not particularly limited as long as epitaxial growth occurs using the SiC single crystal substrate 20 as a seed, and can be performed in a known heat treatment furnace such as a tubular furnace or a hot plate. In addition to these normal pressure (pressless) heat treatments, pressurized heat treatments such as hot pressing and HIP, and combinations of normal pressure heat treatment and pressurized heat treatment can also be used. The heat treatment atmosphere can be selected from vacuum, nitrogen, and inert gas atmospheres. The heat treatment temperature is preferably 1700 to 2700°C. By increasing the temperature, the orientation precursor layer 40 is more likely to grow while being oriented to the c-axis and a-axis using the SiC single crystal substrate 20 as a seed crystal. Therefore, the heat treatment temperature is preferably 1700°C or higher, more preferably 1850°C or higher, even more preferably 2000°C or higher, and particularly preferably 2200°C or higher. On the other hand, if the temperature is too high, a part of the SiC may be lost due to sublimation, or the SiC may be plastically deformed, causing defects such as warping. Therefore, the heat treatment temperature is preferably 2700°C or less, more preferably 2500°C or less. However, since the heat treatment conditions affect the distribution of PL intensity in the graph obtained by PL of the biaxially oriented SiC layer surface, it is preferable to appropriately control the conditions (for example, heat treatment temperature and holding time). From this viewpoint, the heat treatment temperature is preferably 2000 to 2700°C, more preferably 2200 to 2600°C, and even more preferably 2400 to 2500°C. In addition, the holding time is preferably 2 to 30 hours, more preferably 4 to 20 hours. In addition, the heat treatment temperature and holding time are related to the thickness of the SiC substrate 30 generated by epitaxial growth, and can be appropriately adjusted.
[0040] However, when a preformed molded body is used as the orientation precursor layer 40, it is necessary to sinter it during the heat treatment, and high-temperature atmospheric sintering, hot pressing, HIP, or a combination of these is suitable. For example, when using hot pressing, the surface pressure is 50 kgf / cm 2 More preferably, 100 kgf / cm 2More preferably, 200 kgf / cm 2 The temperature is preferably 1700 to 2700°C. The holding time is preferably 2 to 18 hours. The atmosphere during firing can be selected from vacuum, nitrogen, inert gas atmosphere, or a mixture of nitrogen and inert gas. The SiC powder as the raw material may be either α-SiC powder or β-SiC powder, but is preferably β-SiC powder. The SiC powder is preferably composed of SiC particles having an average particle size of 0.01 to 100 μm. The average particle size refers to the average value of the maximum diameter in a certain direction measured by observing the powder with a scanning electron microscope and measuring the maximum diameter in a certain direction for 100 primary particles.
[0041] In the heat treatment process, the crystals in the oriented precursor layer 40 grow from the crystal growth surface of the SiC single crystal substrate 20 while being oriented to the c-axis and a-axis, so that the oriented precursor layer 40 gradually changes from the crystal growth surface to the SiC substrate 30. The SiC composite substrate including the generated SiC substrate 30 has a reduced TSD density on the substrate surface. The reason for this is unclear, but it is thought that the distribution of PL intensity reflects the step-terrace structure that occurs during crystal growth, and that the TSDs are converted to stacking faults as the crystal grows.
[0042] (c) Grinding and polishing process 2(c), in the grinding and polishing step, the orientation precursor layer 40 remaining on the SiC substrate 30 after the heat treatment step is ground away to expose the surface of the SiC substrate 30, which is then polished using diamond abrasive grains and optionally finished by CMP (chemical mechanical polishing). In this manner, the SiC composite substrate 10 is obtained. If it is desired to remove the SiC single crystal substrate 20 of the SiC composite substrate 10, the substrate may be surface ground to a predetermined thickness using a grinder, and then polished using diamond abrasive grains.
[0043] It is to be noted that the present invention is not limited to the above-mentioned embodiment, and can be embodied in various forms as long as it falls within the technical scope of the present invention. For example, in the above-mentioned embodiment, only one layer of SiC substrate 30 is provided on SiC single crystal substrate 20, but two or more layers may be provided. Specifically, a second layer of SiC substrate 30 can be provided on SiC substrate 30 by stacking an orientation precursor layer 40 on SiC substrate 30 of SiC composite substrate 10, and performing heat treatment and grinding in this order. EXAMPLES
[0044] The present invention will be described in more detail with reference to the following examples, but the present invention is not limited to the following examples.
[0045] Example 1 (1) Preparation of the oriented precursor layer Raw material powders containing 90.8% by weight of commercially available fine β-SiC powder (volume-based D50 particle size: 0.7 μm), 8.1% by weight of yttrium oxide powder (volume-based D50 particle size: 0.1 μm), and 1.1% by weight of silicon dioxide powder (volume-based D50 particle size: 0.7 μm) were ball-milled in ethanol using SiC balls for 24 hours and dried to obtain a mixed powder. A commercially available SiC single crystal substrate (n-type 4H-SiC, diameter 100 mm (4 inches), Si face, (0001) face, off-angle 4°, thickness 0.35 mm, no orientation flat) was prepared, and the mixed powder was sprayed onto the SiC single crystal substrate using the AD device 50 shown in FIG. 3 to form an AD film (oriented precursor layer).
[0046] The AD film formation conditions were as follows. First, the carrier gas was N2, and the film was formed using a ceramic nozzle with a slit of 5 mm long side x 0.4 mm short side. The nozzle scan conditions were a scan speed of 0.5 mm / s, and the nozzle was moved 105 mm in the direction perpendicular to the long side of the slit and forward, 5 mm in the direction of the long side of the slit, 105 mm in the direction perpendicular to the long side of the slit and returning, and 5 mm in the direction of the long side of the slit and in the opposite direction to the initial position, and when the nozzle moved 105 mm from the initial position in the direction of the long side of the slit, the nozzle was scanned in the opposite direction to the direction up to that point, and the cycle of returning to the initial position was counted as one cycle, and this cycle was repeated 4000 times. The thickness of the AD film formed in this way was about 400 μm.
[0047] (2) Heat treatment of the orientation precursor layer The SiC single crystal substrate on which the AD film, which is the orientation precursor layer, was formed was removed from the AD device and annealed in an argon atmosphere at 2400°C for 10 hours. In other words, the orientation precursor layer was heat-treated to form a heat-treated layer. In this way, a SiC composite substrate was produced in which a heat-treated layer was formed on a SiC single crystal substrate.
[0048] (3) Grinding and polishing (3-1) Surface polishing The (0001) surface (i.e., the surface of the heat-treated layer) of the SiC composite substrate, which had a heat-treated layer formed on a SiC single crystal substrate, was polished using diamond abrasive grains (grain sizes of 3.0 μm, 1.0 μm, 0.5 μm, and 0.1 μm) in descending order of grain size, until the target thickness and surface condition were achieved.
[0049] (3-2) Backside polishing After the above (3-1), the (000-1) surface of the SiC composite substrate in which the heat treatment layer was formed on the SiC single crystal substrate was surface ground to a predetermined thickness using a grinder (diamond wheel with grit size of 1000 to 6000). Then, diamond abrasive grains (grain sizes of 3.0 μm, 1.0 μm, 0.5 μm, and 0.1 μm) were used in descending order of grain size to perform polishing. In this way, the entire SiC single crystal substrate was ground to obtain a substrate (SiC substrate) consisting only of the heat treatment layer.
[0050] (4) Cutting process of the sample As the sample, the SiC composite substrate or SiC substrate obtained in the above (3-1) or (3-2) was cut with a diamond cutter and processed into a chip shape of 5 mm × 6 mm.
[0051] (5) Evaluation (5-1) Biaxial orientation of the heat treatment layer Using the EBSD (Electron Back Scatter Diffraction Patterns) method, under the conditions shown below, the inverse pole figure mapping of the surface (plate surface) of the heat treatment layer fabricated in the above (3-1) and (3-2) and the cross section perpendicular to the plate surface was measured. As a result, since the tilt angle distribution was 0.01° or less in both the substantially normal direction and the substantially plate surface direction, it was determined that the heat treatment layer was a biaxially oriented SiC layer oriented with the c-axis and a-axis.
[0052] <EBSD measurement conditions> · Acceleration voltage: 15 kv · Spot intensity: 70 · Working distance: 22.5 mm · Step size: 0.5 μm · Sample tilt angle: 70° · Measurement program: Aztec (version 3.3)
[0053] (5-2) Photoluminescence (PL) measurement of the biaxially oriented SiC layer A sample obtained by processing the heat treatment layer (biaxially oriented SiC layer) produced in the above (1) to (4) into a chip shape was used as an evaluation sample, and a PL imaging image of the surface of the evaluation sample was acquired. At this time, as shown in FIG. 4, an arbitrary region of 3.0 mm × 2.2 mm was cut out from the acquired PL imaging image. A line segment of 500 μm parallel to the [11-20] direction (the portion indicated by the lower left line segment in FIG. 4) was drawn at an arbitrary position in this region. As shown in FIG. 5, a graph was created in which the distance (μm) in the [11-20] direction was plotted on the horizontal axis and the PL intensity I was plotted on the vertical axis in the line segment region. This graph had a shape repeating maximum points and minimum points. And the maximum point is a point giving a PL intensity I higher than each of the total 6 points of PL intensity I of the first closest point, the second closest point, and the third closest point on its left and right, and the minimum point is a point giving a PL intensity I lower than each of the total 6 points of PL intensity I of the first closest point, the second closest point, and the third closest point on its left and right. Also, from FIG. 5, the maximum value of the PL intensity I at a certain maximum point P M is denoted as M, and the distance in the horizontal axis direction is longer than this maximum point P M , and the minimum value of the PL intensity I at the minimum point P M closest to this maximum point P m was denoted as m, and M / m was calculated. Further, the distance L in the [11-20] direction between the maximum point P M and the minimum point P m was measured. The results were as shown in Table 1. The measurement conditions at this time were as shown below. From FIG. 4, it was found that the step-terrace structure on the surface of the heat treatment layer had a wide step width of about 100 to 200 μm.
[0054] <PL Measurement Conditions> · Excitation wavelength: 313 nm · Detection wavelength: 750 nm or more · Measurement temperature: Room temperature · Detector: CCD camera (Roper Scientific)
[0055] Also, in the image obtained by cutting out an arbitrary 3.0 mm × 2.2 mm region in the above PL imaging image, using image processing software (product name: WinROOF2015), it was processed with a Prewitt filter using a 5 × 5 kernel in the [11-20] direction. The result is shown in Fig. 6. At an arbitrary position in the image after the filter processing, a 500-μm line segment parallel to the [11-20] direction (the part indicated by the lower left line segment in Fig. 6) was drawn. In the line segment region, a graph was created with the distance (μm) in the [11-20] direction on the horizontal axis and the PL intensity I F on the vertical axis. The graph is shown in Fig. 7. This graph had a shape with multiple maximum points repeating. And the maximum points were confirmed to be points giving a PL intensity I F higher than each of the total six PL intensities I F from the first closest point, the second closest point, and the third closest point on its left and right. Also, from this image, a certain maximum point P M1 and a point having a longer distance in the horizontal axis direction than this maximum point P M1 , and the distance L M1 in the [11-20] direction between this maximum point P M2 and the maximum point P F closest to it were measured. The results were as shown in Table 1.
[0056] (5-3) Boron Concentration in Biaxially Oriented SiC Layer Using the sample obtained by processing the heat treatment layer (biaxially oriented SiC layer) fabricated in the above (1) to (4) into a chip shape as an evaluation sample, the boron concentration inside the biaxially oriented SiC layer was quantified by secondary ion mass spectrometry (SIMS). When measuring the boron concentration from the (0001) plane, which is the substrate surface of the evaluation sample, to a depth of 10 μm, the boron concentration was almost constant regardless of the depth of the biaxially oriented SiC layer, and the boron concentration was as shown in Table 1. In this example, although boron was detected in the biaxially oriented SiC layer despite no boron compound being added, this is only because boron as an inevitable impurity was detected. The analysis conditions at this time were as shown below.
[0057] <SIMS Analysis Conditions> Primary ion species: O2 + Primary acceleration voltage: 11.0kV Detection area: diameter 30μm Detection depth: 10μm
[0058] (5-4) Threading screw dislocation (TSD) density in heat-treated layer (biaxially oriented SiC layer) The SiC composite substrate obtained by the above (1-3-1) was processed into chips by the above (4). The chips were used as evaluation samples, and the TSD density (cm) of the biaxially oriented SiC layer was measured. -2 ) was measured. The evaluation sample was placed in a nickel crucible together with KOH crystals, and etched in an electric furnace at 500°C for 10 minutes. The evaluation sample after etching was washed, and its surface was observed under an optical microscope to determine the type of dislocation from the shape of the pits. Here, the TSD density was measured by regarding shell-shaped pits as basal plane dislocations, small hexagonal pits as threading edge dislocations, and medium to large hexagonal pits as TSDs. Next, the surface of this evaluation sample was polished in the same manner as in (3-1) above, and the etching process and TSD density measurement were repeated multiple times in the same manner as above to obtain TSD densities at multiple depths. Thus, as shown in FIG. 8, the horizontal axis represents the depth (polished thickness) (μm) from the (0001) plane, which is the substrate surface of the evaluation sample, to an arbitrary (000-1) plane, and the TSD density (cm -2 ) on the vertical axis. In FIG. 8, the region where the TSD density increases from the (0001) plane to the (000-1) plane (the region where the polishing thickness is about 380 to 400 μm), that is, the region where the TSD density decreases with a constant slope a as the depth from the (0001) plane to an arbitrary (000-1) plane decreases, was defined as the TSD gradient region. In the graph of this TSD gradient region, an approximate straight line was obtained by the least squares method, and the absolute value of the slope a of the TSD density against the polishing thickness was obtained. In addition, the TSD density of the (0001) plane of the evaluation sample with a polishing thickness of 50 μm was measured as the TSD density of the SiC substrate surface. The results are shown in Table 1.
[0059] Example 2 (comparison) In the above (2), a SiC substrate was produced and evaluated in the same manner as in Example 1, except that the annealing temperature was set to 2200° C. The heat-treated layer of the obtained SiC substrate was confirmed to be a biaxially oriented SiC layer. The results are shown in Table 1.
[0060] Example 3 In the above (1), the raw powder contained 85.8% by weight of fine β-SiC powder (volume-based D50 particle size: 0.7 μm), 8.1% by weight of yttrium oxide powder (volume-based D50 particle size: 0.1 μm), 1.1% by weight of silicon dioxide powder (volume-based D50 particle size: 0.7 μm), and 5.0% by weight of boron carbide powder (volume-based D50 particle size: 0.5 μm). Except for this, the SiC substrate was produced and evaluated in the same manner as in Example 1. It was confirmed that the heat-treated layer of the obtained SiC substrate was a biaxially oriented SiC layer. The results were as shown in Table 1.
[0061] Example 4 In the above (1), the raw powder contained 80.8% by weight of fine β-SiC powder (volume-based D50 particle size: 0.7 μm), 8.1% by weight of yttrium oxide powder (volume-based D50 particle size: 0.1 μm), 1.1% by weight of silicon dioxide powder (volume-based D50 particle size: 0.7 μm), and 10.0% by weight of boron carbide powder (volume-based D50 particle size: 0.5 μm). Except for this, the SiC substrate was produced and evaluated in the same manner as in Example 1. It was confirmed that the heat-treated layer of the obtained SiC substrate was a biaxially oriented SiC layer. The results were as shown in Table 1.
[0062] Example 5 In the above (1), the raw powder contained 70.8% by weight of fine β-SiC powder (volume-based D50 particle size: 0.7 μm), 8.1% by weight of yttrium oxide powder (volume-based D50 particle size: 0.1 μm), 1.1% by weight of silicon dioxide powder (volume-based D50 particle size: 0.7 μm), and 20.0% by weight of boron carbide powder (volume-based D50 particle size: 0.5 μm). Except for this, the SiC substrate was produced and evaluated in the same manner as in Example 1. It was confirmed that the heat-treated layer of the obtained SiC substrate was a biaxially oriented SiC layer. The results were as shown in Table 1.
[0063] Example 6 (comparison) In the above (1), the raw powder contained 60.8% by weight of fine β-SiC powder (volume-based D50 particle size: 0.7 μm), 8.1% by weight of yttrium oxide powder (volume-based D50 particle size: 0.1 μm), 1.1% by weight of silicon dioxide powder (volume-based D50 particle size: 0.7 μm), and 30.0% by weight of boron carbide powder (volume-based D50 particle size: 0.5 μm). Except for this, the SiC substrate was produced and evaluated in the same manner as in Example 1. It was confirmed that the heat-treated layer of the obtained SiC substrate was a biaxially oriented SiC layer. The results were as shown in Table 1.
[0064] Example 7 In the above (2), a SiC substrate was produced and evaluated in the same manner as in Example 1, except that the annealing temperature was set to 2350° C. The heat-treated layer of the obtained SiC substrate was confirmed to be a biaxially oriented SiC layer. The results are shown in Table 1.
[0065] [Table 1]
[0066] From Examples 1 to 7, although the cause is unclear, it was found that the distribution of PL intensity with respect to distance in the [11-20] direction can be controlled by the annealing temperature and boron content. In addition, as shown in Examples 1, 3 to 5, and 7, in the graph obtained by PL, the ratio of M / m is 1.05 or more and the distance L is 15 to 150 μm, or in the graph obtained by processing the image obtained by PL with a Prewitt filter, the distance L is F It was found that by setting the polishing depth to 30-300μm, the TSD density of the (0001) plane of the evaluation sample with a polishing thickness of 50μm, i.e., the TSD density of the biaxially oriented SiC layer surface, can be reduced. In addition, in a graph plotting the depth from the (0001) plane, which is the substrate surface, to an arbitrary (000-1) plane on the horizontal axis and the TSD density on the vertical axis, the TSD density decreased with a constant slope a as the depth decreased, and the absolute value of the slope a was 5.0cm. -2 It was found that SiC substrates with biaxially-oriented SiC layers that have TSD gradient regions with a TSD gradient of 100 μm or more can effectively reduce the TSD density on the surface of the biaxially-oriented SiC layer. This is thought to be because the distribution of PL intensity versus distance in the [11-20] direction reflects the step-terrace structure that occurs during crystal growth, and the TSDs are converted into stacking faults as the crystal grows. On the other hand, when the sintering temperature is low and the ratio of M / m is below 1.05, when the distance L exceeds 150 μm, or when the distance L exceeds 150 μm, as in Examples 2 and 6, the TSD density is reduced. FWhen the thickness of the biaxially oriented SiC layer exceeded 300 μm, the absolute value of the slope a of the TSD density significantly decreased, and the TSD density on the biaxially oriented SiC layer surface increased. This is believed to be because the step-terrace structure collapsed and the TSDs were no longer converted to stacking faults. In this regard, as shown in FIG. 4, it was found that the step-terrace structure on the biaxially oriented SiC layer surface of the SiC substrate showing an example of the present invention has a wide step width of about 100 to 200 μm. On the other hand, according to Non-Patent Document 2 (Toru Ujihara et al. “Conversion Mechanism of Threading Screw Dislocation during SiC Solution Growth” Materials Science Forum Vols. 717-720, pp. 351-354 (2012)), as shown in FIG. 9, the step width on the surface of the SiC substrate produced by the solution growth method is very narrow, less than 20 μm. It is believed that such differences in the step-terrace structure affect the TSD density on the SiC substrate surface.
Claims
1. A SiC substrate having a biaxially oriented SiC layer, the surface of the biaxially oriented SiC layer being analyzed by photoluminescence (PL) to obtain a graph plotting the distance (μm) in the [11-20] direction on the horizontal axis and the PL intensity I on the vertical axis, (i) the graph has a shape in which maximum and minimum points are repeated, and a maximum point is defined as a point that gives a PL intensity I higher than each of the PL intensities I at a total of six points consisting of the first, second, and third closest points on the left and right of the maximum point, and a minimum point is defined as a point that gives a PL intensity I lower than each of the PL intensities I at a total of six points consisting of the first, second, and third closest points on the left and right of the minimum point, (ii) A certain maximum point P M The maximum value of the PL intensity I at M The distance in the horizontal direction is longer than that of the maximum point P M and the closest minimum point P m When the minimum value of the PL intensity I is m, the ratio M / m is 1.05 or more, (iii) The maximum point P M and the minimum point P m the distance L in the [11-20] direction is 15 to 150 μm; The biaxially oriented SiC layer contains boron at a concentration of 1.0×10 16 to 1.0×10 17 atoms / cm 3 .
2. The SiC substrate according to claim 1, wherein the distance L is 50 to 150 μm.
3. A SiC substrate having a biaxially oriented SiC layer, wherein an image obtained by analyzing the surface of the biaxially oriented SiC layer by photoluminescence (PL) is processed by a Prewitt filter in the [11-20] direction of the image, and the distance (μm) in the [11-20] direction is plotted as the horizontal axis, and the PL intensity I F If you get a graph plotted on the vertical axis, (i) The graph has a shape in which a plurality of maximum points are repeated, and the maximum point is a total of six PL intensities I consisting of the first closest point, the second closest point, and the third closest point on the left and right of the maximum point. F A PL intensity I higher than each of F is defined as the point that gives (ii) A certain maximum point P M1 and the maximum point P M1 The distance in the horizontal direction is longer than that of the maximum point P M1 and the closest maximum point P M2 The distance L in the [11-20] direction F is 30 to 300 μm, The biaxially oriented SiC layer contains boron at a concentration of 1.0×10 16 to 1.0×10 17 atoms / cm 3 .
4. The biaxially oriented SiC layer is oriented in the c-axis direction and the a-axis direction, and the depth (μm) from the (0001) plane, which is the substrate surface, to an arbitrary (000-1) plane in the biaxially oriented SiC layer is plotted as the horizontal axis, and the threading screw dislocation (TSD) density (cm -2 ) on the vertical axis, the graph shows that the TSD density decreases with a constant gradient a as the depth decreases, and the absolute value of the gradient a is 5.0 cm -2 / μm or more, The depth from the (0001) plane, which is the substrate surface, to any (000-1) plane means the depth from the (0001) plane of the biaxially oriented SiC layer surface toward the back surface of the biaxially oriented SiC layer. The SiC substrate according to claim 1 or 3.
5. The absolute value of the inclination a is 5.0 to 25 cm -2 The SiC substrate according to claim 4 , wherein the surface roughness is 1 μm.
6. A SiC composite substrate comprising: a SiC single crystal substrate; and the SiC substrate according to claim 1 or 3 on the SiC single crystal substrate.
Citation Information
Patent Citations
Single crystal silicon carbide and its production
JP1999228295A
High resistivity silicon carbide substrate for semiconductor devices with high breakdown voltage
JP2004533720A
METHOD FOR MAKING SiC SINGLE CRYSTAL AND SiC SINGLE CRYSTAL
JP2014043369A
