WO2012157670A1 - Silicon carbide substrate - Google Patents

Silicon carbide substrate Download PDF

Info

Publication number
WO2012157670A1
WO2012157670A1 PCT/JP2012/062507 JP2012062507W WO2012157670A1 WO 2012157670 A1 WO2012157670 A1 WO 2012157670A1 JP 2012062507 W JP2012062507 W JP 2012062507W WO 2012157670 A1 WO2012157670 A1 WO 2012157670A1
Authority
WO
WIPO (PCT)
Prior art keywords
silicon carbide
substrate
stacking fault
transition layer
density
Prior art date
Application number
PCT/JP2012/062507
Other languages
French (fr)
Japanese (ja)
Inventor
邦明 八木
孝光 河原
弘幸 長澤
圭 中筋
晶郎 廣瀬
Original Assignee
Hoya株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hoya株式会社 filed Critical Hoya株式会社
Publication of WO2012157670A1 publication Critical patent/WO2012157670A1/en

Links

Images

Classifications

    • CCHEMISTRY; METALLURGY
    • C30CRYSTAL GROWTH
    • C30BSINGLE-CRYSTAL GROWTH; UNIDIRECTIONAL SOLIDIFICATION OF EUTECTIC MATERIAL OR UNIDIRECTIONAL DEMIXING OF EUTECTOID MATERIAL; REFINING BY ZONE-MELTING OF MATERIAL; PRODUCTION OF A HOMOGENEOUS POLYCRYSTALLINE MATERIAL WITH DEFINED STRUCTURE; SINGLE CRYSTALS OR HOMOGENEOUS POLYCRYSTALLINE MATERIAL WITH DEFINED STRUCTURE; AFTER-TREATMENT OF SINGLE CRYSTALS OR A HOMOGENEOUS POLYCRYSTALLINE MATERIAL WITH DEFINED STRUCTURE; APPARATUS THEREFOR
    • C30B29/00Single crystals or homogeneous polycrystalline material with defined structure characterised by the material or by their shape
    • C30B29/10Inorganic compounds or compositions
    • C30B29/36Carbides
    • HELECTRICITY
    • H01ELECTRIC ELEMENTS
    • H01LSEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/02104Forming layers
    • H01L21/02365Forming inorganic semiconducting materials on a substrate
    • H01L21/02367Substrates
    • H01L21/0237Materials
    • H01L21/02373Group 14 semiconducting materials
    • H01L21/02381Silicon, silicon germanium, germanium
    • HELECTRICITY
    • H01ELECTRIC ELEMENTS
    • H01LSEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/02104Forming layers
    • H01L21/02365Forming inorganic semiconducting materials on a substrate
    • H01L21/02367Substrates
    • H01L21/02433Crystal orientation
    • HELECTRICITY
    • H01ELECTRIC ELEMENTS
    • H01LSEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/02104Forming layers
    • H01L21/02365Forming inorganic semiconducting materials on a substrate
    • H01L21/02518Deposited layers
    • H01L21/02521Materials
    • H01L21/02524Group 14 semiconducting materials
    • H01L21/02529Silicon carbide
    • HELECTRICITY
    • H01ELECTRIC ELEMENTS
    • H01LSEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/02104Forming layers
    • H01L21/02365Forming inorganic semiconducting materials on a substrate
    • H01L21/02612Formation types
    • H01L21/02617Deposition types
    • H01L21/0262Reduction or decomposition of gaseous compounds, e.g. CVD
    • HELECTRICITY
    • H01ELECTRIC ELEMENTS
    • H01LSEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/02104Forming layers
    • H01L21/02365Forming inorganic semiconducting materials on a substrate
    • H01L21/02612Formation types
    • H01L21/02617Deposition types
    • H01L21/02636Selective deposition, e.g. simultaneous growth of mono- and non-monocrystalline semiconductor materials
    • H01L21/02639Preparation of substrate for selective deposition
    • HELECTRICITY
    • H01ELECTRIC ELEMENTS
    • H01LSEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
    • H01L21/00Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
    • H01L21/02Manufacture or treatment of semiconductor devices or of parts thereof
    • H01L21/02104Forming layers
    • H01L21/02365Forming inorganic semiconducting materials on a substrate
    • H01L21/02612Formation types
    • H01L21/02617Deposition types
    • H01L21/02636Selective deposition, e.g. simultaneous growth of mono- and non-monocrystalline semiconductor materials
    • H01L21/02647Lateral overgrowth
    • H01L21/0265Pendeoepitaxy

Definitions

  • the present invention relates to a silicon carbide substrate used for a high-performance semiconductor element.
  • the present invention relates to a silicon carbide substrate that can be preferably used as a material for a high-efficiency and high-voltage power semiconductor element because the surface defect density at a specific surface of the crystal is low and the warpage of the substrate is small.
  • Silicon carbide has begun to be used as a compound semiconductor crystal serving as a substrate for high-performance semiconductor elements.
  • Lattice defects contained in the silicon carbide substrate greatly affect the performance of the semiconductor element.
  • surface defects such as an anti-phase region boundary surface (APB: Anti-Phase-Boundary) and a stacking fault (SF: Stacking Fault) cause current leakage and dielectric breakdown, which significantly impairs the performance of the power semiconductor device. For this reason, reduction of a surface defect density is desired for a silicon carbide substrate used as a substrate for a semiconductor element.
  • APIB Anti-Phase-Boundary
  • SF stacking fault
  • Patent Document 1 A method for reducing surface defects when forming a silicon carbide substrate will be described below.
  • Patent Document 2 As described in Patent Document 1, Patent Document 2, and Non-Patent Document 1, as a method for effectively reducing both the anti-phase region boundary surface and the stacking fault, which are surface defects in SiC, parallel to one direction.
  • a technique has been developed for growing silicon carbide on a silicon substrate having undulations with a smooth ridge. Below, the defect elimination mechanism using this technique is demonstrated.
  • the polar surface is the same as the growth on the silicon substrate in which the main surface introduced with the off angle is (001).
  • the crystal orientation directions are integrated in one direction, and the antiphase region boundary surface disappears.
  • the stacking fault density is reduced by the following mechanism.
  • Si-SF a stacking fault that exposes the Si polar face on the (001) plane
  • C-SF C polar face
  • C-SF self-extinguishes by reducing the surface area because the surface energy of the C polar face is relatively lower than the surface energy of the (001) face.
  • Si—SF is stable because the surface energy of the Si polar face is relatively higher than the surface energy of the (001) face. Therefore, the Si polar face continues to be exposed on the outermost surface and remains on the surface.
  • stacking faults that occur on the facing slopes are aligned in a mirror-facing positional relationship, and disappear as a result of the growth of silicon carbide.
  • Non-patent Document 2 As a means for extinguishing such Si-SF, SBE (Switch Back Epitaxy) technology (Non-patent Document 2) has been developed. In the following, the SBE technique will be described.
  • 11 is a silicon carbide crystal
  • 12 is a stacking fault
  • 13 is a Si polar plane of stacking fault
  • 14 is a C polar plane of stacking fault.
  • reference numerals are given in the description using the drawings, but the reference numerals are omitted in other cases.
  • symbol is described in a parenthesis.
  • the reference numerals may be described in parentheses or there may be no parentheses, but there is no change in the meaning that each component corresponds to each reference numeral in the drawings. As shown in FIG.
  • the SF-exposed surface of Si—SF is a C-polar surface on the back side of the substrate. That is, if the back side of the substrate is the growth surface, the exposed surface of Si—SF remaining when SiC is grown on the undulations having a ridge parallel to one direction as described above can be converted to a C polarity surface. Since the C-polar plane self-extinguishes due to growth, cubic silicon carbide is homoepitaxially grown on the back side, and in principle, the remaining stacking faults disappear and the stacking faults are completely eliminated.
  • the above-described surface defect reduction technique significantly reduces the stacking fault density, and a low-defect density cubic silicon carbide substrate suitable for highly functional semiconductor elements is produced. Since the stacking faults existing in the cubic silicon carbide crystal are annihilated or self-annihilated in the silicon carbide growth process and the stacking fault reducing means process, the defect density has a rate of change corresponding to the growth film thickness. However, since the lattice spacing is reduced when stacking faults disappear due to each annihilation mechanism, an intrinsic stress corresponding to the stacking fault density change is generated in the crystal lattice. Thus, since the stress resulting from the stacking fault density change is applied to the silicon carbide substrate, the silicon carbide substrate is finally warped.
  • the warpage of the cubic silicon carbide substrate produced by the conventional technique reaches several hundreds ⁇ m with a 3 inch diameter and several mm with a 6 inch diameter.
  • the warpage of the substrate allowed for the semiconductor manufacturing process, particularly the photolithography process is 50 ⁇ m or less, and therefore the silicon carbide substrate manufactured by the prior art is not suitable for the semiconductor manufacturing process due to the large warpage.
  • the present invention has been made in view of the above problems, and the defect density and the warpage of the substrate are required so that both the defect density necessary for realizing a highly functional semiconductor element and the amount of warpage necessary for adaptation of the semiconductor manufacturing process are compatible. It aims at providing the silicon carbide substrate which reduced both the quantity.
  • the stacking fault density was reduced to a desired density with a gradual change rate by making the growth film thickness very thick. Later, there is a method of cutting out the surface layer and repeating this process several times.
  • silicon carbide is a difficult-to-cut and brittle material, it is difficult to process, and the manufacturing cost increases, so industrial advantages can be found. Absent. Therefore, the details of (b) were studied intensively, and the configuration of the present invention was reached.
  • the problem solving means is as follows. (1) In a silicon carbide substrate having two planes parallel to each other of a first surface that is a main surface and a second surface that is a back surface of the first surface, A surface defect is exposed on the first surface and the second surface, and The surface defect density N1 exposed on the first surface is smaller than the surface defect density N2 exposed on the second surface, and
  • the silicon carbide substrate has one or more transition layers substantially parallel to the first surface having a surface defect density gradient in a direction substantially perpendicular to the first surface;
  • the transition layer has a space inside, and A silicon carbide substrate, wherein a surface defect density gradient in the transition layer is 20 ⁇ (N2-N1) / t or more, where t is a distance between the first surface and the second surface.
  • FIG. 2 is an example in which the transition layer is provided on the first surface side.
  • the crystal structure that lowers the stacking fault density change gradient is also referred to as a mother crystal.
  • crystal structure or “other crystal structure”.
  • the surface defect density on the first surface is N1
  • the surface defect density on the second surface is N2
  • the surface defect density at the interface between the transition layer and the mother crystal is N3
  • the thickness of the transition layer is t T
  • the first surface The distance between the first surface and the second surface, that is, the thickness (total film thickness) of the entire region in the silicon carbide substrate is defined as t.
  • gradient in this specification refers to the rate of change when the surface defect density changes in the thickness direction.
  • the change rate of the surface defect density is called a gradient.
  • the change rate of the surface defect density is referred to as a gradient.
  • control of the surface defect density gradient is important.
  • an inclusion region that prevents propagation of stacking faults is provided in the transition layer.
  • “gradient 1”, which is the surface defect density gradient that the transition layer should have, is set to be “20 ⁇ (N2 ⁇ N1) / t or more”, that is, “20 ⁇ gradient 2” or more. is doing.
  • the surface defect density gradient (gradient 1) of the transition layer is set to be 20 times or more the surface defect density gradient (gradient 2) of the entire region.
  • the “transition layer” in the present specification is a layered portion in the silicon carbide substrate, and an inclusion region (substantially refers to a space.
  • the inclusion region is referred to as a space). It refers to the part inside.
  • the inclusion region (space) may be one or plural.
  • each space is distributed substantially parallel to the first surface.
  • the transition layer is a layered portion parallel to the main surface (52) and including the space (53).
  • FIG. 5 is a schematic diagram (cross-sectional view) for illustrating the space provided in the silicon carbide crystal.
  • the space referred to here does not propagate the stacking fault (55) because a mismatched interface is formed at the interface with silicon carbide inside the silicon carbide substrate. This is because stacking faults propagate in a specific direction because they are in a single crystal.
  • a part of the space (53) in FIG. 5 (the lower part and the part on the second surface side) is formed on the main surface of the silicon carbide substrate, and silicon carbide is further grown thereon. Then, silicon carbide is connected at the main surface (52).
  • the film growth direction can be adjusted as appropriate by adjusting the film forming conditions (adjusting the temperature and pressure, particularly the Si / C raw material supply ratio).
  • the stacking fault is prevented from propagating as soon as it comes into contact with the space (53), regardless of the disappearance of the association.
  • the stacking fault density can be significantly reduced on the main surface (52) regardless of the presence or absence of strain.
  • the Young's modulus of the space (53) is lower than that of silicon carbide (51), the thermal strain and lattice strain generated in the silicon carbide crystal lattice are absorbed, and the generation of stacking faults (55) is suppressed. Furthermore, even if the stacking fault (55) occurs, the space (53) prevents the propagation of them, so that it is possible to reduce the stacking fault density in the upper part of the space (53). However, in order to effectively reduce the stacking fault density, it is necessary to satisfy the condition that all stacking faults (55) meet in the space (53).
  • limiting of (Formula 1) is added with respect to height (H).
  • the side wall of the portion that forms the space must not be parallel to the closest dense surface through which the stacking fault (55) propagates, and the inner angle (54: ⁇ ) of the side wall of the portion that forms the stacking fault and the space is 0. More than 90 degrees and less than 90 degrees, the relationship of Equation 1 must be maintained.
  • the silicon carbide substrate is cubic silicon carbide
  • the main surface is the (001) plane
  • the side wall of the portion forming the space is parallel to the ⁇ 110 ⁇ plane
  • 35.3 degrees.
  • W is 10 ⁇ m
  • H may be 14.1 ⁇ m or more.
  • FIG. 5 is a cross-sectional view in the substrate specific direction, but each space may be discrete or may be connected to each other in the transition layer.
  • the space and the side wall forming the space are distributed over the entire plane substantially parallel to the first surface in the transition layer. Is preferred.
  • the transition layer has a plurality of spaces, it is preferable that the transition layer is distributed substantially in parallel to the main surface. Even in the case of a single space, the side wall forming the space extends over the entire plane substantially parallel to the main surface. Are preferably distributed.
  • off-angle machining is performed by slightly tilting the surface normal axis in order to reduce the stacking fault density gradient and eliminate the anti-phase region boundary surface.
  • Silicon carbide is grown on the formed silicon substrate to form a mother crystal. By introducing the off-angle, the propagation direction of stacking faults becomes unidirectional, the mechanism of disappearance of their association disappears, and the mechanism of decrease of stacking fault density accompanying growth disappears.
  • silicon carbide grows on a silicon substrate whose main surface (the most exposed part of the crystal surface has the largest exposed area) is the (001) plane, an angle of about several degrees is given in the [110] direction.
  • the stacking fault propagating in parallel to the (111) plane is blocked, and the stacking fault propagation direction is limited to the direction parallel to the ( ⁇ 1-11) plane.
  • the stacking fault propagating in parallel to the ( ⁇ 111) plane is cut off, and the stacking fault propagation direction is limited to the direction parallel to the (1-11) plane.
  • the angle applied in the [110] direction needs to be different from the angle applied in the [ ⁇ 110] direction. As a result, anisotropy occurs in the step flow growth direction, the remaining antiphase boundary is suppressed, and the propagation direction of the remaining stacking faults can be controlled.
  • the warpage of the silicon carbide substrate can be reduced by providing a transition layer with a steep stacking fault density gradient at the beginning of growth and then growing a silicon carbide crystal (mother crystal) that maintains a low stacking fault density. . That is, in this case, a mother crystal is formed from the transition layer to the first surface. Even when this means is used, in order to eliminate the anti-phase region boundary surface, a silicon substrate in which off angles are given in advance to the [110] direction and the [ ⁇ 110] direction is used.
  • the inventors of the present invention have a undulation extending direction and residual Si—SF. It is confirmed that there is a correlation in the propagation direction. Specifically, Si—SF is exposed in the direction perpendicular to the undulation extending direction on the silicon carbide growth surface.
  • the present inventors grew silicon carbide on a substrate having such undulations, the amount of warpage of the obtained silicon carbide film was not isotropic in the plane, and the undulation extending direction and its orthogonal direction And found that there is anisotropy (the curvature radius of the warping amount is different).
  • the presence of an in-plane warp anisotropy in the silicon carbide substrate is considered undesirable in the device manufacturing process. That is, since the same radius of curvature in the silicon carbide substrate surface is preferable, the [110] orientation and the [ ⁇ 110] orientation are used rather than using a substrate having undulations extending in one direction as in Patent Document 2. It is preferable to use a substrate having a different off angle.
  • a silicon carbide substrate having a low warpage and a low defect density can be manufactured without complicating the manufacturing process.
  • FIG. 1 It is a schematic diagram showing a stacking fault structure in cubic silicon carbide It is a figure which shows the stacking fault profile of the board
  • this invention is not limited to this. It is a cubic silicon carbide substrate, a main surface having a stacking fault density of N1 is parallel to the ⁇ 001 ⁇ plane, and is a plate-like crystal having a back surface parallel to this and a stacking fault density of N2.
  • the surface defect density gradient in the transition layer having a space is 20 ⁇ (N2-N1) / t or more, preferably 30 ⁇ (N2-N1) / T or more, more desirably 50 ⁇ (N2-N1) / t or more.
  • the numerical range of N1 is preferably 700 / cm or less.
  • FIG. 3 is a stress distribution in the thickness direction of a typical cubic silicon carbide substrate obtained by numerical analysis.
  • FIG. 4 shows a stacking fault density distribution in the thickness direction of a typical cubic silicon carbide substrate. 3 and 4 show that there is a close correlation between stress and stacking fault density distribution. 3 and 4, it can be seen that the stress changes greatly from the back surface of the substrate to 100 ⁇ m, and the stacking fault density also decreases from the back surface of the substrate to a thickness of about 100 ⁇ m.
  • the stress inside the substrate is determined by a stress of up to about 100 ⁇ m with a change in stacking fault density. If such a transition layer is embodied, there is little influence on the warp of the silicon carbide crystal substrate, and a warp of 50 ⁇ m, which is an allowable amount of the semiconductor manufacturing process, is realized.
  • the transition layer thickness t T is preferably t / 20 or less.
  • the thickness t T of the transition layer is t / 20 or less, there is little influence on the warp of the entire silicon carbide crystal substrate, and a warp of 50 ⁇ m, which is an allowable amount of the semiconductor process, can be realized more reliably.
  • the transition layer thickness t T is desirably t / 30 or less.
  • a plurality of spaces are present uniformly within 0.05 ⁇ t, preferably within 0.04 ⁇ t, more preferably within 0.01 ⁇ t from the main surface to the inside of the substrate. is doing.
  • the layer in which the space is distributed is substantially parallel to the main surface, and the side wall of the portion forming the space is substantially parallel to the ⁇ 110 ⁇ plane.
  • the width (S), interval (W), and height (H) of each space do not need to be exactly the same, but W is 100 nm to 100 ⁇ m, preferably 1 ⁇ m to 50 ⁇ m, and more preferably 2 ⁇ m. Within 20 ⁇ m. This is because, as W becomes narrower, not only processing for forming a space becomes difficult, but also the volume occupancy of the space in the silicon carbide substrate increases, and the substrate resistance when operating as a semiconductor device increases. It is. On the other hand, if W becomes too large, the number of spaces in the transition layer will decrease. Also, in order to prevent the propagation of defects, the height H of the space must be increased. Increasing the height H leads to increasing the thickness of the silicon carbide substrate itself.
  • the volume of the silicon carbide substrate is increased more than the height H is increased and the volume of the space is increased.
  • H is specified from the above W according to (Equation 1), but if it is extremely small, not only the processing becomes difficult, but also the volume occupancy of the space decreases, and thermal strain cannot be absorbed completely. .
  • extremely large processing 100 ⁇ m or more is also difficult, and the volume occupancy of the space increases, increasing the transition layer thickness and inhibiting effective warpage reduction.
  • the width (S) of the space is 100 nm to 100 ⁇ m, preferably 1 ⁇ m to 50 ⁇ m, more preferably 2 ⁇ m to 20 ⁇ m. This is because, as S becomes narrower, the processing for forming the space becomes difficult, and the volume occupancy of the space decreases, making it impossible to absorb the thermal strain. On the other hand, if S becomes extremely large, the volume occupancy of the space increases and the substrate resistance when operating as a semiconductor device increases, and it becomes difficult to set W to a desirable value. Alternatively, the present invention can be realized even when a plurality of spaces are uniformly present at 0.95 ⁇ t or more, preferably 0.96 ⁇ t or more from the main surface to the back surface of the substrate.
  • silicon carbide is applied to the silicon substrate that has been subjected to off-angle processing that slightly tilts the surface normal axis in order to eliminate the anti-phase region boundary surface while reducing the stacking fault density gradient.
  • off-angle processing the surface normal axis of the (001) plane is tilted in the range of 0.1 ° to 54.7 °, preferably 1 ° to 6 ° in the [110] direction.
  • the surface normal axis of the (001) plane is also tilted in the range of 0.1 ° to 54.7 °, preferably 1 ° to 6 ° in the [ ⁇ 110] direction.
  • transition layer is present on the first surface or the second surface.
  • the transition layer is present between the first surface and the second surface. It doesn't matter.
  • a plurality of transition layers may exist. After all, a transition layer having a steep surface defect density change gradient and a mother crystal that lowers the stacking fault density change gradient from the first surface or the second surface to the transition layer may be present.
  • a silicon substrate in which the surface normal axis of the (001) plane is tilted by 4 ° in the [110] direction and the surface normal axis of the (001) plane is also tilted by 2 ° in the [ ⁇ 110] direction is placed in a C 2 H in a CVD apparatus. Heating was performed in a mixed atmosphere of 2 and H 2 to form an ultrathin silicon carbide layer. The diameter of the substrate is 6 inches. The substrate was heated to 1350 ° C. The source gas C 2 H 2 and the carrier gas H 2 were supplied to the substrate surface from room temperature. The supply amount and pressure are as shown in the substrate temperature raising conditions in Table 1.
  • silicon carbide was grown by supplying SiH 2 Cl 2 , C 2 H 2 and H 2 at a temperature of 1350 ° C.
  • the growth conditions in this case are as shown in Table 2. The pressure during the growth was adjusted with a pressure regulating valve installed between the reaction chamber and the pump.
  • Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3
  • C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
  • the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate. Near the interface with the silicon substrate, stacking fault layers due to lattice mismatch exist at high density. If this layer is present, the warp cannot be reduced even if the transition layer according to the present invention is provided.
  • a polishing process using 0.1 ⁇ m diamond abrasive grains is performed to remove the 10 ⁇ m stacking fault layer from the silicon substrate interface.
  • it was immersed in molten KOH at 500 ° C. for 5 minutes. Then, when the optical microscope observation was implemented with respect to the board
  • the stacking fault density on the back surface, that is, the second surface was 2.1 ⁇ 10 5 pieces / cm as observed with an optical microscope.
  • the surface defect (stacking defect) density measurement was performed as follows.
  • the (111) plane defect that appears on the (001) surface is mainly parallel to the [ ⁇ 110] direction.
  • the surface defects parallel to the [ ⁇ 110] direction are calculated as the number (unit: lines / cm) intersecting the unit length in the [110] direction perpendicular to the [110] direction. Is calculated as the number (unit: lines / cm) intersecting the unit length in the [ ⁇ 110] direction orthogonal to the surface length, and the average value thereof is defined as the surface defect (stacking defect) density.
  • a metal film layer having a laminated structure of Cr / Au was provided on the main surface using a vacuum deposition method.
  • the thickness of each layer was set to Cr 25 nm and Au 110 nm.
  • a positive resist of 2 ⁇ m was spin-coated on the upper layer, and exposed with ultraviolet rays so as to expose a square pattern array with a photomask.
  • the mask pattern line side was aligned so as to be parallel to the ⁇ 110 ⁇ orientation of the substrate.
  • One side of the square pattern was 2 ⁇ m, and the interval between the patterns was 9 ⁇ m.
  • the resist was developed to provide an opening in the exposed region, and Ni having a thickness of 0.5 ⁇ m was formed in the opening by Ni electroplating.
  • the resist was completely removed by organic cleaning, and a Ni pattern having a thickness of 0.5 ⁇ m was obtained in a square portion with a side of 2 ⁇ m arranged at intervals of 9 ⁇ m.
  • vertical mesa processing was performed by inductively coupled plasma reactive ion etching. Silicon carbide is processed together with chromium and gold other than the square. Thereafter, Cr, Au, and Ni remaining in the square portion were completely removed by wet etching. Further, it was washed with a hydrogen peroxide solution + sulfuric acid mixed solution (1: 1) and an HF solution. After cleaning, a thermal oxide film was formed to a thickness of about 100 nm using a heat treatment apparatus.
  • the formed thermal oxide film was removed with dilute hydrofluoric acid.
  • discrete cubic silicon carbide reference numeral 62 in FIG. 6
  • the start surface of the transition layer is a surface including the bottom of the mesa-processed portion formed by the above method (the bottom of the lower portion of the later space (74)) or the vicinity thereof. It is extremely difficult to measure the stacking fault density of the portion corresponding to the bottom, and there is almost no change in the stacking fault density with a difference of 5 ⁇ m in the thickness direction with respect to the total film thickness of about 300 ⁇ m.
  • the stacking fault density on the main surface before mesa processing is used as the stacking fault density (N3) on the start surface of the transition layer.
  • the “transition layer” includes not only the portion corresponding to the bottom portion but also the vicinity thereof.
  • FIG. 6 shows a cross-sectional structure of the silicon carbide substrate after patterning.
  • 61 is a silicon carbide layer
  • 62 is a discretely arranged silicon carbide.
  • FIG. 7 shows a cross-sectional structure of the silicon carbide grown on the silicon carbide substrate of FIG.
  • Reference numeral 71 denotes a silicon carbide layer
  • 72 denotes silicon carbide discretely grown selectively
  • 73 denotes a silicon carbide layer grown on discretely formed silicon carbide
  • 74 denotes a space.
  • the upper portions of discretely grown silicon carbide (72) are connected to each other, the main surface, that is, the depth of 0.1 ⁇ m below the first surface is 4 ⁇ m wide, 7 ⁇ m apart, A space (74) having a height of 9.9 ⁇ m was formed.
  • 6 and 7 are schematic views only. Originally, when silicon carbide is grown after forming the lower portion of the space as shown in FIG. 6, film growth in both the substrate film thickness direction and the surface in-plane direction occurs, and discretely as shown in FIG. The upper portions of the selectively grown silicon carbide (72) are connected to each other, and the upper portion of the space (74) is also formed.
  • the silicon carbide substrate was immersed in molten KOH at 500 ° C. for 5 minutes.
  • the stacking fault density on the first surface was 525 / cm.
  • a layer that is, a transition layer
  • the stacking fault density is significantly reduced as compared with that before the space is formed. Note that in the region occupying the upper part (the main surface side part) of the space (74) in the transition layer, the propagation of stacking faults is prevented in the space (74) when discretely growing silicon carbide selectively.
  • the stacking fault density is of course greatly reduced. Further, regarding the region that occupies the lower part (the back side part) of the space (74) in the transition layer, a slightly new carbonization is formed in the lower part of the space (74) when silicon carbide is selectively grown discretely. Although silicon is stacked, the stacking fault density is reduced because the propagation of stacking faults is prevented by the space (74) also in this stacked portion. Therefore, in the present specification, a layered portion in the silicon carbide substrate, which is provided with a space (74) therein, is a “transition layer” in which the stacking fault density is sharply reduced.
  • cubic silicon carbide having a (001) plane as the main surface was used as the substrate.
  • the side wall of the portion forming the space is the ⁇ 111 ⁇ plane, ⁇ It is parallel to either the 110 ⁇ plane or the ⁇ 211 ⁇ plane.
  • the ⁇ 0001 ⁇ plane can be selected as the main surface.
  • the side wall of the portion forming the space is parallel to the ⁇ 11-20 ⁇ plane or the ⁇ 1100 ⁇ plane.
  • the most dense surface through which the stacking fault propagates is the (0001) plane, and the stacking fault intersects with the main surface at an angle of 30 to 60 degrees. It is clear that it will result.
  • FIG. 1 the definition of the curvature of a board
  • the substrate surface was scanned horizontally with a laser displacement meter manufactured by Keyence Corporation, and the difference between the maximum value and the minimum value at the substrate height at that time was warped.
  • Table 4 shows the values of the parameters in FIG.
  • t was measured with a micrometer manufactured by Mitutoyo Corporation. N1, N2, and N3 were subjected to a molten KOH treatment (500 ° C. ⁇ 5 min) on the surface of the cubic silicon carbide substrate, and after surface defects were revealed, the number of surface defects per unit length was counted with a microscope. . When the surface defect density gradient was obtained from the obtained numerical value, the gradient 1 was 25.3 times the gradient 2, and the condition of 20 times or more was satisfied. The transition layer thickness t T was about 1/30 of the total film thickness t. It was confirmed by warpage measurement that the warpage of the substrate was 45 ⁇ m.
  • the silicon carbide arranged discretely is formed using the square mask pattern.
  • the silicon carbide arranged discretely may be formed using the mask pattern of the line and space. .
  • a space is formed using discretely arranged silicon carbide formed in this way, a plurality of discrete spaces are formed that are positioned in parallel to each other on a plane substantially parallel to the first surface. . Even in such a form, the effects of the present invention can be sufficiently obtained.
  • ⁇ Comparative Example 1> A silicon substrate in which the surface normal axis of the (001) plane is tilted by 2 ° in the [110] direction and the surface normal axis of the (001) plane is also tilted by 1 ° in the [ ⁇ 110] direction is placed in a C 2 H in a CVD apparatus. Heating was performed in a mixed atmosphere of 2 and H 2 to form an ultrathin silicon carbide layer. The diameter of the substrate is 6 inches. The substrate was heated to 1350 ° C. The source gas C 2 H 2 and the carrier gas H 2 were supplied to the substrate surface from room temperature. The supply amount and pressure are as shown in Table 1. After the substrate surface reached 1350 ° C., the substrate was maintained in the above-described C 2 H 2 and hydrogen atmosphere for 15 minutes.
  • silicon carbide was grown by supplying SiH 2 Cl 2 , C 2 H 2 and H 2 at a temperature of 1350 ° C.
  • the growth conditions in this case are as shown in Table 2.
  • the pressure during the growth was adjusted with a pressure regulating valve installed between the reaction chamber and the pump.
  • Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3
  • C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
  • the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
  • a polishing process using 0.1 ⁇ m diamond abrasive grains is performed to remove a 10 ⁇ m stacking fault layer from the interface.
  • it was immersed in molten KOH at 500 ° C. for 5 minutes. Thereafter, when the substrate on which the defects were revealed was observed with an optical microscope, the stacking fault density on the front surface, that is, the first surface was 4000 / cm, and the stacking fault density on the back surface, that is, the second surface, was 8000. Book / cm.
  • the surface defect (stacking defect) density measurement is calculated as the number (unit: book / cm) of plane defects parallel to the [ ⁇ 110] direction intersecting the unit length in the [110] direction perpendicular to the [ ⁇ 110] direction.
  • the surface defects parallel to the [110] direction are calculated as the number (unit: lines / cm) intersecting the unit length in the [ ⁇ 110] direction orthogonal to the surface defects, and the average value of these is calculated as the surface defect (stacking defect). Density.
  • Table 5 shows the values of the parameters of the obtained substrate. In this comparative example, since no transition layer is provided, N3 is omitted.
  • t and t T was measured by Mitutoyo micrometer.
  • N1, N2, and N3 the surface of the 3C—SiC substrate was subjected to a molten KOH treatment (500 ° C., 5 minutes), and surface defects were revealed, and then the number of surface defects per unit length was counted with a microscope.
  • the gradient 1 was about 13.7 times the gradient 2, and the condition of 20 times or more was not satisfied. It was confirmed by warpage measurement that the warpage of the substrate was 500 ⁇ m. That is, it was found that when the gradient 1 is less than 20 times the gradient 2, the warpage of the substrate becomes large.
  • Comparative Example 2 was performed by a method basically according to Example 1. Note that one side of the square discrete region is 2 ⁇ m and the interval is 16 ⁇ m. First, 300 ⁇ m cubic silicon carbide was grown on a 6-inch diameter silicon substrate. However, a silicon substrate having a (001) surface normal axis inclined by 3 ° in the [110] direction and a (001) surface normal axis inclined by 2 ° in the [ ⁇ 110] direction was used as the silicon substrate. After the growth, the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
  • the surface defect (stacking defect) density measurement is calculated as the number (unit: book / cm) of plane defects parallel to the [ ⁇ 110] direction intersecting the unit length in the [110] direction perpendicular to the [ ⁇ 110] direction.
  • the surface defects parallel to the [110] direction are calculated as the number (unit: lines / cm) intersecting the unit length in the [ ⁇ 110] direction orthogonal to the surface defects, and the average value of these is calculated as the surface defect (stacking defect). Density.
  • the upper portions of discretely grown silicon carbide (72) are connected to each other, the width is 4 ⁇ m, the distance is 14 ⁇ m, and the depth is 30.1 ⁇ m at the lower portion of the main surface (that is, the first surface).
  • a space (74) having a height of 19.9 ⁇ m was formed. In this case, the thickness of the transition layer was 20 ⁇ m.
  • Table 7 shows the value of each parameter for the obtained substrate.
  • t and t T was measured by Mitutoyo micrometer.
  • N1, N2, and N3 the surface of the 3C—SiC substrate was subjected to a molten KOH treatment (500 ° C., 5 min), surface defects were revealed, and the number of surface defects per unit length was counted with a microscope.
  • the gradient 1 was 15.8 times the gradient 2 and did not satisfy the condition of 20 times or more. It was confirmed by warpage measurement that the warpage of the substrate was 250 ⁇ m.
  • a silicon substrate in which the surface normal axis of the (001) plane is tilted by 4 ° in the [110] direction and the surface normal axis of the (001) plane is also tilted by 2 ° in the [ ⁇ 110] direction is placed in a C 2 H in a CVD apparatus. Heating was performed in a mixed atmosphere of 2 and H 2 to form an ultrathin silicon carbide layer. The diameter of the substrate is 6 inches.
  • the substrate was heated to 1350 ° C.
  • the source gas C 2 H 2 and the carrier gas H 2 were supplied to the substrate surface from room temperature. The supply amount and pressure are as shown in Table 1. After the substrate surface reached 1350 ° C., the substrate was maintained in the above-described C 2 H 2 and hydrogen atmosphere for 15 minutes.
  • silicon carbide was grown by supplying SiH 2 Cl 2 , C 2 H 2 and H 2 at a temperature of 1350 ° C.
  • the growth conditions in this case are as shown in Table 2.
  • the pressure during the growth was adjusted with a pressure regulating valve installed between the reaction chamber and the pump.
  • Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3
  • C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
  • a metal film layer having a laminated structure of Cr / Au was provided on the main surface using a vacuum deposition method.
  • the thickness of each layer was set to Cr 25 nm and Au 110 nm.
  • a positive resist of 2 ⁇ m was spin-coated on the upper layer, and exposed with ultraviolet rays so as to expose a square pattern array with a photomask.
  • the mask pattern line side was aligned so as to be parallel to the ⁇ 110 ⁇ orientation of the substrate.
  • One side of the square pattern was 2 ⁇ m, and the distance between the patterns was 12 ⁇ m.
  • the resist was developed to provide an opening in the exposed region, and Ni having a thickness of 0.5 ⁇ m was formed in the opening by Ni electroplating.
  • the resist was completely removed by organic cleaning, and a Ni pattern having a thickness of 0.5 ⁇ m was obtained at a square portion having a side of 2 ⁇ m arranged at intervals of 12 ⁇ m.
  • vertical mesa processing was performed by inductively coupled plasma reactive ion etching. Silicon carbide is processed together with chromium and gold other than the square. Thereafter, Cr, Au, and Ni remaining in the square portion were completely removed by wet etching. Further, it was washed with a hydrogen peroxide solution + sulfuric acid mixed solution (1: 1) and an HF solution. After cleaning, a thermal oxide film was formed to a thickness of about 100 nm using a heat treatment apparatus.
  • the formed thermal oxide film was removed with dilute hydrofluoric acid. Through these steps, discrete cubic silicon carbide (that is, discretely arranged silicon carbide (62)) having a side of 2 ⁇ m and a height of 7 ⁇ m arranged on the main surface at intervals of 12 ⁇ m is formed. It was.
  • FIG. 6 shows a cross-sectional structure of the silicon carbide substrate after patterning. Thereafter, growth was performed for 60 minutes under the conditions in Table 8 to grow 8 ⁇ m cubic silicon carbide ((72) and (73)). At this time, as shown in FIG. 7, the upper parts of discretely grown silicon carbide (72) are connected to each other, the depth of the lower part of the main surface is 0.1 ⁇ m, the width is 4 ⁇ m, the interval is 10 ⁇ m, and the height is 14.9 ⁇ m. A space (74) is formed. In order to measure the stacking fault density on the main surface, that is, the transition layer surface, it was immersed in molten KOH at 500 ° C. for 5 minutes.
  • the stacking fault density in the main surface was 550 pieces / cm.
  • the surface of the transition layer is a surface including a portion immediately after the upper portions of discretely grown silicon carbide (72) are connected to each other, and the stacking fault density of the portion is strictly measured. Since there is almost no change in the stacking fault density when silicon carbide is stacked to the extent that the stacking fault density can be measured immediately after connection, the stacking fault density on the main surface described above Is used as the stacking fault density (N3) on the transition layer surface.
  • the “transition layer” includes not only a portion immediately after the upper portions of discretely grown silicon carbide (72) are connected to each other, but also a portion in which silicon carbide is laminated immediately after the connection so that the stacking fault density can be measured. .
  • a polishing process using 0.1 ⁇ m diamond abrasive grains is performed to remove a 50 ⁇ m stacking fault layer from the interface between the silicon substrate and silicon carbide.
  • it was immersed in molten KOH at 500 ° C. for 5 minutes. Thereafter, when an optical microscope observation was performed on the substrate on which the defects became apparent, the stacking fault density on the front surface (that is, the first surface) was 510 / cm, and the stacking fault on the back surface (that is, the second surface). The density was 1.5 ⁇ 10 5 lines / cm. Further, no anti-phase region interface was observed on the start surface of the transition layer.
  • Table 9 shows the values of the parameters in FIG. The substrate, the transition layer as shown in FIG. 9 (t T region) becomes the second surface area.
  • Total thickness t and the transition layer thickness t T was measured by Mitutoyo micrometer.
  • N1, N2, and N3 were subjected to molten KOH treatment (500 ° C., 5 min) on the surface of the cubic silicon carbide substrate, and surface defects were revealed. The number of defects was counted.
  • the surface defect density gradient (gradient 1) in the transition layer is about 20.6 times the surface defect density gradient (gradient 2) outside the transition layer region, which is 20 times or more.
  • the condition was met.
  • the warpage of the substrate was confirmed by warpage measurement to be 40 ⁇ m. That is, even when the transition layer is located on the second surface side, it was confirmed that a substrate with small warpage can be obtained if the gradient 1 is 20 times or more the gradient 2. In this example, the thickness of the transition layer satisfied 1/20 or less of the entire substrate.
  • Example 1 except that a silicon substrate in which the surface normal axis of the (001) plane is inclined by 3 ° in the [110] direction and the surface normal axis of the (001) plane is also inclined by 2 ° in the [ ⁇ 110] direction is used.
  • an ultrathin silicon carbide layer was formed on the silicon substrate under the conditions shown in Table 1. After the substrate surface reached 1350 ° C., the substrate was maintained in the above-described C 2 H 2 and hydrogen atmosphere for 15 minutes. Thereafter, in the same manner as in Example 1, growth was performed for 6 hours under the conditions shown in Table 2, and 300 ⁇ m cubic silicon carbide was grown on the Si substrate.
  • Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3
  • C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
  • the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
  • a polishing process using 0.1 ⁇ m diamond abrasive grains is performed to remove the 10 ⁇ m stacking fault layer from the silicon substrate interface.
  • the stacking fault density on the front surface that is, the start surface of the transition layer was 7600 / cm
  • the stacking fault density on the back surface that is, the second surface was The number was 8000 / cm. Further, no anti-phase region interface was observed on the start surface of the transition layer.
  • the surface defect (stacking defect) density measurement is calculated as the number (unit: book / cm) of plane defects parallel to the [ ⁇ 110] direction intersecting the unit length in the [110] direction perpendicular to the [ ⁇ 110] direction.
  • the surface defects parallel to the [110] direction are calculated as the number (unit: lines / cm) intersecting the unit length in the [ ⁇ 110] direction orthogonal to the surface defects, and the average value of these is calculated as the surface defect (stacking defect). Density.
  • a metal film layer having a laminated structure of Cr / Au was provided on the main surface using a vacuum deposition method.
  • the thickness of each layer was set to Cr 25 nm and Au 110 nm.
  • a positive resist of 2 ⁇ m was spin-coated on the upper layer, and exposed with ultraviolet rays so as to expose a square pattern array with a photomask.
  • the mask pattern line side was aligned so as to be parallel to the ⁇ 110 ⁇ orientation of the substrate.
  • One side of the square pattern was 2 ⁇ m, and the interval between the patterns was 9 ⁇ m.
  • the resist was developed to provide an opening in the exposed region, and Ni having a thickness of 0.5 ⁇ m was formed in the opening by Ni electroplating.
  • the resist was completely removed by organic cleaning, and a Ni pattern having a thickness of 0.5 ⁇ m was obtained in a square portion with a side of 2 ⁇ m arranged at intervals of 9 ⁇ m.
  • vertical mesa processing was performed by inductively coupled plasma reactive ion etching. Silicon carbide is processed together with chromium and gold other than the square. Thereafter, Cr, Au, and Ni remaining in the square portion were completely removed by wet etching. Further, it was washed with a hydrogen peroxide solution + sulfuric acid mixed solution (1: 1) and an HF solution. After cleaning, a thermal oxide film was formed to a thickness of about 100 nm using a heat treatment apparatus.
  • the formed thermal oxide film was removed with dilute hydrofluoric acid. Through such steps, discrete cubic silicon carbide (62 in FIG. 6) arranged on the main surface at intervals of 9 ⁇ m and having a side of 2 ⁇ m and a height of 5 ⁇ m was formed.
  • the silicon carbide substrate was immersed in molten KOH at 500 ° C. for 5 minutes.
  • the stacking fault density on the first surface was 525 / cm.
  • FIG. 10 shows the definition of the curvature of a board
  • t was measured with a micrometer manufactured by Mitutoyo Corporation. N1, N2, and N3 were subjected to a molten KOH treatment (500 ° C. ⁇ 5 min) on the surface of the cubic silicon carbide substrate, and after surface defects were revealed, the number of surface defects per unit length was counted with a microscope. . When the gradient was obtained from the obtained numerical value, the gradient 1 was about 27.9 times the gradient 2, and the condition of 20 times or more was satisfied. The transition layer thickness t T was about 1/30 of the total film thickness t. Further, it was confirmed by warpage measurement that the warpage of the substrate was 45 ⁇ m.
  • Example 2 except that a silicon substrate in which the surface normal axis of the (001) plane is inclined by 3 ° in the [110] direction and the surface normal axis of the (001) plane is also inclined by 2 ° in the [ ⁇ 110] direction is used. Similarly, heating was performed in a CVD apparatus in a mixed atmosphere of C 2 H 2 and H 2 to form an extremely thin silicon carbide layer on the silicon substrate. The diameter of the substrate is 6 inches. The substrate was heated to 1350 ° C. The source gas C 2 H 2 and the carrier gas H 2 were supplied to the substrate surface from room temperature. The supply amount and pressure are as shown in Table 1.
  • the substrate surface reached 1350 ° C.
  • the substrate was maintained in the above-described C 2 H 2 and hydrogen atmosphere for 15 minutes.
  • silicon carbide was grown by supplying SiH 2 Cl 2 , C 2 H 2 and H 2 at a temperature of 1350 ° C.
  • the growth conditions in this case are as shown in Table 2.
  • the pressure during the growth was adjusted with a pressure regulating valve installed between the reaction chamber and the pump.
  • Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3
  • C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
  • a metal film layer having a laminated structure of Cr / Au was provided on the main surface using a vacuum deposition method.
  • the thickness of each layer was set to Cr 25 nm and Au 110 nm.
  • a positive resist of 2 ⁇ m was spin-coated on the upper layer, and exposed with ultraviolet rays so as to expose a square pattern array with a photomask.
  • the mask pattern line side was aligned so as to be parallel to the ⁇ 110 ⁇ orientation of the substrate.
  • One side of the square pattern was 2 ⁇ m, and the interval between the patterns was 9 ⁇ m.
  • the resist was developed to provide an opening in the exposed region, and Ni having a thickness of 0.5 ⁇ m was formed in the opening by Ni electroplating.
  • the resist was completely removed by organic cleaning, and a Ni pattern having a thickness of 0.5 ⁇ m was obtained in a square portion with a side of 2 ⁇ m arranged at intervals of 9 ⁇ m.
  • vertical mesa processing was performed by inductively coupled plasma reactive ion etching. Silicon carbide is processed together with chromium and gold other than the square. Thereafter, Cr, Au, and Ni remaining in the square portion were completely removed by wet etching. Further, it was washed with a hydrogen peroxide solution + sulfuric acid mixed solution (1: 1) and an HF solution. After cleaning, a thermal oxide film was formed to a thickness of about 100 nm using a heat treatment apparatus.
  • the formed thermal oxide film was removed with dilute hydrofluoric acid. Through these steps, discrete cubic silicon carbide (that is, discretely arranged silicon carbide (62)) having a side of 2 ⁇ m and a height of 5 ⁇ m arranged at intervals of 9 ⁇ m is formed on the main surface. It was.
  • FIG. 6 shows a cross-sectional structure of the silicon carbide substrate after patterning. Thereafter, growth was performed for 60 minutes under the conditions shown in Table 8 to grow 5 ⁇ m cubic silicon carbide ((72) and (73) in FIG. 7). At this time, as shown in FIG. 7, the upper parts of discretely grown silicon carbide (72) are connected to each other, the depth of the lower part of the main surface is 0.1 ⁇ m, the width is 4 ⁇ m, the interval is 7 ⁇ m, and the height is 9.9 ⁇ m. A space (74) is formed. In order to measure the stacking fault density on the main surface, that is, the transition layer surface, it was immersed in molten KOH at 500 ° C. for 5 minutes.
  • the stacking fault density in the main surface was 525 / cm.
  • the surface defect parallel to the [ ⁇ 110] direction is calculated as the number (unit: lines / cm) intersecting the unit length in the [110] direction perpendicular to the [110] direction. ]
  • the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
  • a polishing process using 0.1 ⁇ m diamond abrasive grains is performed to remove the 25 ⁇ m stacking fault layer from the interface between the silicon substrate and silicon carbide.
  • it was immersed in molten KOH at 500 ° C. for 5 minutes. Thereafter, when an optical microscope observation was performed on the substrate in which the defects became apparent, the stacking fault density on the front surface (namely, the first surface) was 500 / cm2, and the stacking fault on the back surface (namely, the second surface). The density was 8000 pieces / cm. Further, no anti-phase region interface was observed on the start surface of the transition layer.
  • Table 11 shows the values of the parameters in FIG. 8 for the obtained substrate.
  • the substrate, the transition layer as shown in FIG. 9 (t T region) becomes the second surface area.
  • Total thickness t and the transition layer thickness t T was measured by Mitutoyo micrometer.
  • N1, N2, and N3 were subjected to a molten KOH treatment (500 ° C., 5 min) on the surface of the cubic silicon carbide substrate, and after revealing surface defects, the number of surface defects per unit length was counted with a microscope. .
  • the surface defect density gradient (gradient 1) in the transition layer is about 30 times the surface defect density gradient (gradient 2) outside the transition layer region, and the condition of 20 times or more is satisfied. I met.
  • the surface defect distribution in the region other than the transition layer, that is, N3 was within the range of N1 ⁇ 5%.
  • the warpage of the substrate was 50 ⁇ m. That is, even when the transition layer is located on the second surface side, it was confirmed that a substrate with small warpage can be obtained if the gradient 1 is 20 times or more the gradient 2. In this example, the thickness of the transition layer satisfied 1/20 or less of the entire substrate.

Abstract

Provided is a silicon carbide substrate fulfilling the following: the plane defect density (N1) exposed at a first surface is less than the plane defect density (N2) exposed at a second surface; the silicon carbide substrate has at least one transition layer approximately parallel to a first surface and having a plane defect density gradient in the direction approximately perpendicular to the first surface; the transition layer is provided with an internal space; and when the distance from the first surface to the second surface is defined as t, the plane defect density gradient in the transition layer is at least 20×(N2-N1)/t.

Description

炭化珪素基板Silicon carbide substrate
 本発明は、高機能半導体素子に用いられる炭化珪素基板に関するものである。特に、結晶の特定表面における面欠陥密度が低く、かつ基板の反りが小さいため、高効率かつ高耐圧のパワー半導体素子の材料として好ましく用いることのできる炭化珪素基板に関するものである。 The present invention relates to a silicon carbide substrate used for a high-performance semiconductor element. In particular, the present invention relates to a silicon carbide substrate that can be preferably used as a material for a high-efficiency and high-voltage power semiconductor element because the surface defect density at a specific surface of the crystal is low and the warpage of the substrate is small.
 高機能半導体素子の基板となる化合物半導体結晶として、炭化珪素(SiC)が用いられ始めている。上記炭化珪素基板に含まれる格子欠陥は、半導体素子の性能に多大な影響を与える。たとえば、反位相領域境界面(APB:Anti-Phase-Boundary)や積層欠陥(SF:Stacking Fault)などの面欠陥は電流漏洩や絶縁破壊をもたらし、電力用半導体素子の性能を著しく損なう。このため、半導体素子の基板として用いられる炭化珪素基板に対しては、面欠陥密度の低減が望まれる。 Silicon carbide (SiC) has begun to be used as a compound semiconductor crystal serving as a substrate for high-performance semiconductor elements. Lattice defects contained in the silicon carbide substrate greatly affect the performance of the semiconductor element. For example, surface defects such as an anti-phase region boundary surface (APB: Anti-Phase-Boundary) and a stacking fault (SF: Stacking Fault) cause current leakage and dielectric breakdown, which significantly impairs the performance of the power semiconductor device. For this reason, reduction of a surface defect density is desired for a silicon carbide substrate used as a substrate for a semiconductor element.
 以下には炭化珪素基板を形成する際の面欠陥低減方法を述べる。
 SiC内の面欠陥である反位相領域境界面と積層欠陥の両者を効果的に低減させる方法として、特許文献1、特許文献2、非特許文献1に記載されているように、1方向に平行な尾根を有する起伏が形成された珪素基板上に炭化珪素を成長させる技術が開発された。以下には、この技術を用いた欠陥解消機構を説明する。
A method for reducing surface defects when forming a silicon carbide substrate will be described below.
As described in Patent Document 1, Patent Document 2, and Non-Patent Document 1, as a method for effectively reducing both the anti-phase region boundary surface and the stacking fault, which are surface defects in SiC, parallel to one direction. A technique has been developed for growing silicon carbide on a silicon substrate having undulations with a smooth ridge. Below, the defect elimination mechanism using this technique is demonstrated.
 はじめに、1方向に平行な尾根を有する起伏の斜面が加工された珪素基板に炭化珪素を成長させると、オフ角を導入した主表面が(001)である珪素基板上の成長と同様に極性面の結晶配向方位が1方向に統合されて反位相領域境界面が消滅する。 First, when silicon carbide is grown on a silicon substrate on which a undulating slope having a ridge parallel to one direction is processed, the polar surface is the same as the growth on the silicon substrate in which the main surface introduced with the off angle is (001). The crystal orientation directions are integrated in one direction, and the antiphase region boundary surface disappears.
 そして、1方向に平行な尾根を有する起伏加工された主表面が(001)である珪素基板上に炭化珪素が成長する際には、次の機構により積層欠陥密度が減少する。立方晶炭化珪素(001)面には、2種類の積層欠陥が存在する。1つは、(001)面にSi極性面を露出する積層欠陥(以下、Si-SFとする)であり、もう1つは、C極性面を露出する積層欠陥(以下、C-SFとする)である。このうちC-SFは、C極性面の表面エネルギーが(001)面の表面エネルギーに対して相対的に低いために、表面積を縮小させて自己消滅する。一方、Si-SFは、Si極性面の表面エネルギーが(001)面の表面エネルギーに対して相対的に高いために安定である。そのため、Si極性面が最表面に露出し続け、表面に残存する。このSi-SFに関しては、向かい合った斜面で発生した積層欠陥が鏡面対向の位置関係に揃えられて、炭化珪素の成長に伴って会合消滅する。これらの積層欠陥解消機構が働くと、従来の立方炭化珪素成長では実現できなかった積層欠陥密度減少がもたらされる。 And, when silicon carbide grows on a silicon substrate whose undulated main surface having a ridge parallel to one direction is (001), the stacking fault density is reduced by the following mechanism. There are two types of stacking faults on the cubic silicon carbide (001) plane. One is a stacking fault that exposes the Si polar face on the (001) plane (hereinafter referred to as Si-SF), and the other is a stacking fault that exposes the C polar face (hereinafter referred to as C-SF). ). Among these, C-SF self-extinguishes by reducing the surface area because the surface energy of the C polar face is relatively lower than the surface energy of the (001) face. On the other hand, Si—SF is stable because the surface energy of the Si polar face is relatively higher than the surface energy of the (001) face. Therefore, the Si polar face continues to be exposed on the outermost surface and remains on the surface. With regard to this Si—SF, stacking faults that occur on the facing slopes are aligned in a mirror-facing positional relationship, and disappear as a result of the growth of silicon carbide. When these stacking fault elimination mechanisms work, a stacking fault density reduction that cannot be realized by conventional cubic silicon carbide growth is brought about.
 ところが、1方向に平行な尾根を有する起伏加工された主表面が(001)である珪素基板上に炭化珪素をヘテロエピタキシャル成長すると、C-SFが基板表面にて消滅しているのに対し、会合消滅が発現するにも関わらずSi-SFは依然として残存する。これは、Si-SFは、成長時の格子歪に起因して常に発生しているためである。成長時に炭化珪素内部には、基板面内の温度分布による熱歪や積層欠陥が会合消滅する際の格子整合にともなう歪が生じる。この歪が炭化珪素の弾性限界歪より大きくなると、歪を緩和するために炭化珪素は塑性変形し積層欠陥が生じる。すなわち、会合消滅で積層欠陥密度が減少しながら、一方で歪によって新たな積層欠陥が発生してしまう。 However, when silicon carbide is heteroepitaxially grown on a silicon substrate whose undulated main surface having a ridge parallel to one direction is (001), C-SF disappears on the substrate surface, but the association Despite the disappearance, Si-SF still remains. This is because Si—SF is always generated due to lattice strain during growth. At the time of growth, thermal strain due to temperature distribution in the substrate surface and strain associated with lattice matching when the stacking faults disappear together are generated in the silicon carbide. When this strain becomes larger than the elastic limit strain of silicon carbide, silicon carbide is plastically deformed to cause a stacking fault in order to relax the strain. That is, the stacking fault density decreases due to the disappearance of association, while a new stacking fault occurs due to strain.
 このようなSi-SFを消滅させる手段として、SBE(Switch Back Epitaxy)技術(非特許文献2)が開発された。以下においては、SBE技術を説明する。 As a means for extinguishing such Si-SF, SBE (Switch Back Epitaxy) technology (Non-patent Document 2) has been developed. In the following, the SBE technique will be described.
 図1において、11は炭化珪素結晶、12は積層欠陥、13は積層欠陥のSi極性面、14は積層欠陥のC極性面である。以降、明細書においては、各図を用いた説明の際には符号を付すが、それ以外の際においては符号を省略する。なお、明細書において各構成の名称に対して符号を付す際には、括弧内に符号を記載する。また、各図において符号が括弧内に記載されている場合や括弧が無い場合もあるが、両者とも、図面に記載の各構成と符号とを対応させるという意味に変わりはない。
 図1に示したようにSi-SFのSF露出面は基板裏面側においてC極性面である。すなわち、基板の裏面側を成長面とすれば、先述の1方向に平行な尾根を有する起伏上にSiCを成長させた際に残存したSi-SFの露出面をC極性面に転換させられる。C極性面は成長によって自己消滅することから、この裏面側に立方晶炭化珪素をホモエピタキシャル成長することで、原理上は残存した積層欠陥は消滅し、積層欠陥の完全解消が実現される。
In FIG. 1, 11 is a silicon carbide crystal, 12 is a stacking fault, 13 is a Si polar plane of stacking fault, and 14 is a C polar plane of stacking fault. Hereinafter, in the specification, reference numerals are given in the description using the drawings, but the reference numerals are omitted in other cases. In addition, when attaching | subjecting a code | symbol with respect to the name of each structure in a specification, a code | symbol is described in a parenthesis. Further, in each figure, the reference numerals may be described in parentheses or there may be no parentheses, but there is no change in the meaning that each component corresponds to each reference numeral in the drawings.
As shown in FIG. 1, the SF-exposed surface of Si—SF is a C-polar surface on the back side of the substrate. That is, if the back side of the substrate is the growth surface, the exposed surface of Si—SF remaining when SiC is grown on the undulations having a ridge parallel to one direction as described above can be converted to a C polarity surface. Since the C-polar plane self-extinguishes due to growth, cubic silicon carbide is homoepitaxially grown on the back side, and in principle, the remaining stacking faults disappear and the stacking faults are completely eliminated.
特許第3707726号公報Japanese Patent No. 3707726 特許第3576432号公報Japanese Patent No. 3576432
 上記の面欠陥低減技術によって積層欠陥密度は大幅に減少し、高機能な半導体素子に適した低欠陥密度の立方晶炭化珪素基板が作製される。立方晶炭化珪素結晶内に存在する積層欠陥は炭化珪素成長過程と上記積層欠陥低減手段過程において会合消滅、あるいは自己消滅するため、その欠陥密度は成長膜厚に応じた変化率を有する。しかしながら、各消滅機構によって積層欠陥が消滅すると格子間隔が縮小するため、結晶格子内には積層欠陥密度変化に応じる量の真性応力が生じる。このように、積層欠陥密度変化に起因する応力が炭化珪素基板には加わることより、最終的に炭化珪素基板に反りが生じてしまう。従来の技術で作製した立方晶炭化珪素基板の反りは3インチ口径で数百μm、6インチ口径で数mmにも及ぶ。半導体製造プロセス、とくにフォトリソグラフィープロセスに許容される基板の反りは50μm以下であり、したがって従来技術にて作製された炭化珪素基板は反りが大きいために半導体製造プロセスに適合されない。 The above-described surface defect reduction technique significantly reduces the stacking fault density, and a low-defect density cubic silicon carbide substrate suitable for highly functional semiconductor elements is produced. Since the stacking faults existing in the cubic silicon carbide crystal are annihilated or self-annihilated in the silicon carbide growth process and the stacking fault reducing means process, the defect density has a rate of change corresponding to the growth film thickness. However, since the lattice spacing is reduced when stacking faults disappear due to each annihilation mechanism, an intrinsic stress corresponding to the stacking fault density change is generated in the crystal lattice. Thus, since the stress resulting from the stacking fault density change is applied to the silicon carbide substrate, the silicon carbide substrate is finally warped. The warpage of the cubic silicon carbide substrate produced by the conventional technique reaches several hundreds μm with a 3 inch diameter and several mm with a 6 inch diameter. The warpage of the substrate allowed for the semiconductor manufacturing process, particularly the photolithography process is 50 μm or less, and therefore the silicon carbide substrate manufactured by the prior art is not suitable for the semiconductor manufacturing process due to the large warpage.
 本発明は上記問題点を鑑みてなされたものであり、高機能な半導体素子の実現に必要な欠陥密度と、半導体製造プロセスの適応に必要な反り量を両立するよう、欠陥密度と基板の反り量の両方を低減した炭化珪素基板を提供することを目的とする。 The present invention has been made in view of the above problems, and the defect density and the warpage of the substrate are required so that both the defect density necessary for realizing a highly functional semiconductor element and the amount of warpage necessary for adaptation of the semiconductor manufacturing process are compatible. It aims at providing the silicon carbide substrate which reduced both the quantity.
 炭化珪素基板内部における真性応力と積層欠陥密度分布の相関を検討したところ、積層欠陥密度の変化率を小さくすれば炭化珪素基板の反りは低減することが導き出された。表面において低欠陥密度を実現しつつ、かつ積層欠陥密度の変化率を減じる手段としては、(a)基板裏面にて積層欠陥密度を半導体素子に必要な密度まで低減し、表面まで積層欠陥密度を変化させないこと、(b)基板裏面から表面近傍まで高密度な積層欠陥密度を維持し、表面近傍で急峻に積層欠陥密度を減少させること、が考えられる。(a)の基板裏面にて積層欠陥密度を半導体素子に必要な密度まで低減する手段としては、成長膜厚を非常に厚くして緩やかな変化率で所望の密度まで積層欠陥密度を低減させたのちに表層を切り出す、およびこの工程を幾度と繰り返す方法があるが、炭化珪素は難削かつ脆性材料であるために加工が困難であり、製造コストが嵩んでしまうため、工業的なメリットが見出せない。そこで(b)について詳細を鋭意検討し、本発明の構成に至った。 Examination of the correlation between the intrinsic stress inside the silicon carbide substrate and the stacking fault density distribution revealed that the warpage of the silicon carbide substrate can be reduced if the rate of change in stacking fault density is reduced. As a means of reducing the stacking fault density change rate while realizing a low defect density on the surface, (a) reducing the stacking fault density to the density required for the semiconductor element on the back of the substrate, It is conceivable that (b) a high stacking fault density is maintained from the back surface of the substrate to the vicinity of the front surface, and the stacking fault density is sharply decreased in the vicinity of the front surface. As a means of reducing the stacking fault density to the density required for the semiconductor element on the back surface of the substrate of (a), the stacking fault density was reduced to a desired density with a gradual change rate by making the growth film thickness very thick. Later, there is a method of cutting out the surface layer and repeating this process several times. However, since silicon carbide is a difficult-to-cut and brittle material, it is difficult to process, and the manufacturing cost increases, so industrial advantages can be found. Absent. Therefore, the details of (b) were studied intensively, and the configuration of the present invention was reached.
 本発明による課題解決手段は以下の通りである。
(1)主表面である第1の面と、前記第1の面の裏面である第2の面の、互いに平行な二平面を有する炭化珪素基板において、
 前記第1の面および第2の面には面欠陥が露出し、かつ、
 前記第1の面に露出する面欠陥密度N1は、前記第2の面に露出する面欠陥密度N2より小さく、かつ、
 前記炭化珪素基板は、前記第1の面に略垂直方向に面欠陥密度勾配を有する前記第1の面に略平行な1以上の遷移層を有し、
 前記遷移層は、内部に空間を備え、かつ、
 前記第1の面と前記第2の面との距離をtとすると、前記遷移層における面欠陥密度勾配が20×(N2-N1)/t以上であることを特徴とする炭化珪素基板。
(2)前記遷移層の内部において、前記空間を形成する側壁が、前記第1の面に略平行な方向に分布している(1)の炭化珪素基板。
(3)前記炭化珪素基板中に存在する面欠陥が、主として積層欠陥である(1)又は(2)の炭化珪素基板。
(4)前記遷移層の厚さが、前記第1の面と前記第2の面との距離tの20分の1以下である(1)~(3)いずれかの炭化珪素基板。
(5)前記炭化珪素基板の結晶系が立方晶であり、前記第1の面が{100}面に略平行である(1)~(4)いずれかの炭化珪素基板。
(6)前記炭化珪素基板の結晶系が立方晶であり、前記第1の面が{111}面に略平行である(1)~(4)いずれかの炭化珪素基板。
The problem solving means according to the present invention is as follows.
(1) In a silicon carbide substrate having two planes parallel to each other of a first surface that is a main surface and a second surface that is a back surface of the first surface,
A surface defect is exposed on the first surface and the second surface, and
The surface defect density N1 exposed on the first surface is smaller than the surface defect density N2 exposed on the second surface, and
The silicon carbide substrate has one or more transition layers substantially parallel to the first surface having a surface defect density gradient in a direction substantially perpendicular to the first surface;
The transition layer has a space inside, and
A silicon carbide substrate, wherein a surface defect density gradient in the transition layer is 20 × (N2-N1) / t or more, where t is a distance between the first surface and the second surface.
(2) The silicon carbide substrate according to (1), wherein side walls forming the space are distributed in a direction substantially parallel to the first surface inside the transition layer.
(3) The silicon carbide substrate according to (1) or (2), wherein the surface defects present in the silicon carbide substrate are mainly stacking faults.
(4) The silicon carbide substrate according to any one of (1) to (3), wherein a thickness of the transition layer is equal to or less than 1/20 of a distance t between the first surface and the second surface.
(5) The silicon carbide substrate according to any one of (1) to (4), wherein the silicon carbide substrate has a cubic crystal system and the first surface is substantially parallel to the {100} plane.
(6) The silicon carbide substrate according to any one of (1) to (4), wherein the silicon carbide substrate has a cubic crystal system, and the first surface is substantially parallel to a {111} plane.
 以下、本発明について詳細に説明する。
 炭化珪素結晶では、面欠陥に会合消滅機構あるいは自己消滅が発現し、成長にともなって面欠陥密度が変化するため、結晶内部には積層欠陥密度勾配に応じた応力が生じる。この応力によって、炭化珪素基板には反りが発生する。このような応力発生を抑制し、基板の反りを低減させるために、本発明では一例として、図2に示すように積層欠陥密度変化勾配を急峻とする遷移層、および、その遷移層までは積層欠陥密度変化勾配を低くする結晶構造を設けている。図2は、遷移層が第1の面側に設けられている例である。この場合、上記の積層欠陥密度変化勾配を低くする結晶構造のことを、母結晶とも言う。また、単に「結晶構造」または「その他の結晶構造」とも言う。第1の面における面欠陥密度をN1、第2の面における面欠陥密度をN2、遷移層と母結晶との界面の面欠陥密度をN3、遷移層の厚さをt、第1の面と第2の面との間の距離、即ち炭化珪素基板における全領域の厚さ(全膜厚)をtと定義する。図2では、遷移層の片側は第1の面となっているため、この時の遷移層内の面欠陥密度勾配を勾配1とすると、勾配1=(N3-N1)/t、領域全体の面欠陥密度勾配を勾配2とすると、勾配2=(N2-N1)/tとなる。 一方、遷移層が第2の面側に設けられている場合、N3は遷移層とその上部に設けられた結晶構造(母結晶)との界面の面欠陥密度のことを指し、遷移層内の面欠陥密度勾配を勾配1とすると、勾配1=(N2-N3)/tとなる。なお、勾配2については、遷移層が第1の面側に設けられている場合と同様、勾配2=(N2-N1)/tとなる。
Hereinafter, the present invention will be described in detail.
In silicon carbide crystals, an association annihilation mechanism or self-annihilation occurs in the plane defects, and the plane defect density changes with growth, so that stress corresponding to the stacking fault density gradient is generated inside the crystal. Due to this stress, the silicon carbide substrate is warped. In order to suppress such stress generation and reduce the warpage of the substrate, in the present invention, as an example, a transition layer having a steep stacking fault density change gradient as shown in FIG. A crystal structure is provided to reduce the defect density change gradient. FIG. 2 is an example in which the transition layer is provided on the first surface side. In this case, the crystal structure that lowers the stacking fault density change gradient is also referred to as a mother crystal. It is also simply referred to as “crystal structure” or “other crystal structure”. The surface defect density on the first surface is N1, the surface defect density on the second surface is N2, the surface defect density at the interface between the transition layer and the mother crystal is N3, the thickness of the transition layer is t T , and the first surface The distance between the first surface and the second surface, that is, the thickness (total film thickness) of the entire region in the silicon carbide substrate is defined as t. In FIG. 2, since one side of the transition layer is the first surface, assuming that the plane defect density gradient in the transition layer at this time is gradient 1, the gradient 1 = (N3-N1) / t T , the entire region Assuming that the surface defect density gradient is 2, the gradient 2 = (N2−N1) / t. On the other hand, when the transition layer is provided on the second surface side, N3 indicates the surface defect density at the interface between the transition layer and the crystal structure (mother crystal) provided on the upper side. If the surface defect density gradient is gradient 1, then gradient 1 = (N2−N3) / t T. For the gradient 2, the gradient 2 = (N2−N1) / t, as in the case where the transition layer is provided on the first surface side.
 なお、本明細書における「勾配」とは、面欠陥密度が板厚方向において変化するときの変化率のことを指す。図3や図4(共に後述)に示すように、板厚方向の距離に対して面欠陥密度が曲線状に変化する場合にも、面欠陥密度の変化率のことを勾配と言う。もちろん、図2や図9(共に後述)に示すように、板厚方向の距離に対して面欠陥密度が直線状に変化する場合にも、面欠陥密度の変化率のことを勾配と言う。 In addition, “gradient” in this specification refers to the rate of change when the surface defect density changes in the thickness direction. As shown in FIG. 3 and FIG. 4 (both will be described later), even when the surface defect density changes in a curved shape with respect to the distance in the plate thickness direction, the change rate of the surface defect density is called a gradient. Of course, as shown in FIGS. 2 and 9 (both will be described later), even when the surface defect density changes linearly with respect to the distance in the plate thickness direction, the change rate of the surface defect density is referred to as a gradient.
 本発明では、面欠陥密度勾配の制御が重要である。積層欠陥密度勾配を急峻とするため、遷移層中に、積層欠陥の伝播を阻止する内包領域を設けている。それに加え、遷移層が有すべき面欠陥密度勾配である「勾配1」が、「“20×(N2-N1)/t”以上」即ち「“20×勾配2”以上」となるように設定している。つまり、遷移層の面欠陥密度勾配(勾配1)が、領域全体の面欠陥密度勾配(勾配2)の20倍以上となるように設定している。
 本明細書における「遷移層」とは、炭化珪素基板内における層状の部分であって、内包領域(実質的には、空間のことを指す。以降、内包領域のことを空間と言う。)を内部に有する部分のことを指す。ここで、内包領域(空間)は、1つでも複数でも良い。複数の内包領域(空間)を有する場合、それぞれの空間は、第1の面に略平行に分布されている。なお、図5で言うと、遷移層は、主表面(52)と平行な層状の部分であって空間(53)を含んだ部分である。
In the present invention, control of the surface defect density gradient is important. In order to make the stacking fault density gradient steep, an inclusion region that prevents propagation of stacking faults is provided in the transition layer. In addition, “gradient 1”, which is the surface defect density gradient that the transition layer should have, is set to be “20 × (N2−N1) / t or more”, that is, “20 × gradient 2” or more. is doing. In other words, the surface defect density gradient (gradient 1) of the transition layer is set to be 20 times or more the surface defect density gradient (gradient 2) of the entire region.
The “transition layer” in the present specification is a layered portion in the silicon carbide substrate, and an inclusion region (substantially refers to a space. Hereinafter, the inclusion region is referred to as a space). It refers to the part inside. Here, the inclusion region (space) may be one or plural. In the case of having a plurality of inclusion regions (spaces), each space is distributed substantially parallel to the first surface. In FIG. 5, the transition layer is a layered portion parallel to the main surface (52) and including the space (53).
 なお、空間形成手段は以下のとおりである。図5は、炭化珪素結晶内に設けた空間を説明するための概略図(断面図)である。ここで言う空間は、炭化珪素基板の内部において、炭化珪素との界面に不整合界面を形成しているため、積層欠陥(55)を伝播させない。なぜならば、積層欠陥は単結晶中だからこそ特定の方向に伝播するからである。一例を挙げると、図5の空間(53)の一部分(下部分であり、第2の面側の部分)を炭化珪素基板の主表面に形成しておき、その上に炭化珪素を更に成長させ、主表面(52)において炭化珪素を連結させる。このとき、基板膜厚方向および表面面内方向の両方の膜成長が生じ、それに伴い、空間(53)の一部分(上部分であり、第1の面側の部分)も形成されることになり、炭化珪素基板内に空間(53)が形成される。膜成長方向は、成膜条件の調整(温度、圧力、特にSi/C原料供給比の調整)で適宜調整できる。その際、積層欠陥は会合消滅によらず、空間(53)と接するやいなや伝播が阻止される。その結果、歪の有無に関わらず、主表面(52)上において積層欠陥密度の大幅な低減を導くことができる。また、空間(53)のヤング率は炭化珪素(51)のそれ以下であるため、炭化珪素の結晶格子に生ずる熱歪や格子歪を吸収し、積層欠陥(55)の発生を抑制する。さらに、積層欠陥(55)が発生したとしても空間(53)がそれらの伝播を防ぐため、空間(53)の上部において積層欠陥密度の低減を導くことが可能となる。ただし、効果的に積層欠陥密度の低減を導くためには、すべての積層欠陥(55)が空間(53)に会合する条件を満たす必要があり、空間同士の間の間隔(W)、方向、そして高さ(H)に対して(式1)の制限が加わる。
Figure JPOXMLDOC01-appb-M000001
 
The space forming means is as follows. FIG. 5 is a schematic diagram (cross-sectional view) for illustrating the space provided in the silicon carbide crystal. The space referred to here does not propagate the stacking fault (55) because a mismatched interface is formed at the interface with silicon carbide inside the silicon carbide substrate. This is because stacking faults propagate in a specific direction because they are in a single crystal. For example, a part of the space (53) in FIG. 5 (the lower part and the part on the second surface side) is formed on the main surface of the silicon carbide substrate, and silicon carbide is further grown thereon. Then, silicon carbide is connected at the main surface (52). At this time, film growth in both the substrate film thickness direction and the surface in-plane direction occurs, and accordingly, a part of the space (53) (the upper part and the part on the first surface side) is also formed. A space (53) is formed in the silicon carbide substrate. The film growth direction can be adjusted as appropriate by adjusting the film forming conditions (adjusting the temperature and pressure, particularly the Si / C raw material supply ratio). In this case, the stacking fault is prevented from propagating as soon as it comes into contact with the space (53), regardless of the disappearance of the association. As a result, the stacking fault density can be significantly reduced on the main surface (52) regardless of the presence or absence of strain. Further, since the Young's modulus of the space (53) is lower than that of silicon carbide (51), the thermal strain and lattice strain generated in the silicon carbide crystal lattice are absorbed, and the generation of stacking faults (55) is suppressed. Furthermore, even if the stacking fault (55) occurs, the space (53) prevents the propagation of them, so that it is possible to reduce the stacking fault density in the upper part of the space (53). However, in order to effectively reduce the stacking fault density, it is necessary to satisfy the condition that all stacking faults (55) meet in the space (53). The space (W) between the spaces, the direction, And the restriction | limiting of (Formula 1) is added with respect to height (H).
Figure JPOXMLDOC01-appb-M000001
 すなわち、空間を形成する部分の側壁は、積層欠陥(55)が伝播する最稠密面と平行であってはならず、積層欠陥と空間を形成する部分の側壁の内角(54:θ)は0度よりも大きく90度未満であり、式1の関係を保たなければならない。たとえば、炭化珪素基板が立方晶炭化珪素であり、その主表面が(001)面であり、空間を形成する部分の側壁が{110}面に平行な場合には自ずとθ=35.3度となり、たとえばWが10μmの場合にはHは14.1μm以上であれば良いことが分かる。この条件においては、空間下部のいかなる積層欠陥も空間を貫通することができなくなり、積層欠陥密度の低い主表面を得ることができる。
 なお、空間内に存在する物質としては、空間を維持することができ、炭化珪素基板の形成の支障にならず、欠陥密度を低減させることができるものであれば任意のものあっても構わない。
 また、欠陥密度の低減には、空間を形成する部分の側壁が大きく寄与し、遷移層における空間の数や形態は様々なアレンジが可能である。例えば、図5は基板特定方向における断面図であるが、それぞれの空間は離散していても良いし、遷移層内で互いに連結していても良い。第1の面の全体にわたって欠陥密度が低減していることが好ましいため、空間およびこれを形成する側壁は、遷移層における第1の面に略平行な面内全般に亘って分布していることが好ましい。遷移層が複数の空間を有する場合には主表面に略平行に分布されていることが好ましく、空間が一つの場合でも、空間を形成する側壁が、主表面に略平行な面内全体に亘って分布するような形態が好ましい。
That is, the side wall of the portion that forms the space must not be parallel to the closest dense surface through which the stacking fault (55) propagates, and the inner angle (54: θ) of the side wall of the portion that forms the stacking fault and the space is 0. More than 90 degrees and less than 90 degrees, the relationship of Equation 1 must be maintained. For example, when the silicon carbide substrate is cubic silicon carbide, the main surface is the (001) plane, and the side wall of the portion forming the space is parallel to the {110} plane, naturally θ = 35.3 degrees. For example, when W is 10 μm, it is understood that H may be 14.1 μm or more. Under this condition, any stacking fault in the lower part of the space cannot penetrate the space, and a main surface with a low stacking fault density can be obtained.
As the substance existing in the space, any material can be used as long as it can maintain the space, does not hinder the formation of the silicon carbide substrate, and can reduce the defect density. .
Further, the side wall of the portion forming the space greatly contributes to the reduction of the defect density, and various arrangements of the number and form of the spaces in the transition layer are possible. For example, FIG. 5 is a cross-sectional view in the substrate specific direction, but each space may be discrete or may be connected to each other in the transition layer. Since it is preferable that the defect density is reduced over the entire first surface, the space and the side wall forming the space are distributed over the entire plane substantially parallel to the first surface in the transition layer. Is preferred. When the transition layer has a plurality of spaces, it is preferable that the transition layer is distributed substantially in parallel to the main surface. Even in the case of a single space, the side wall forming the space extends over the entire plane substantially parallel to the main surface. Are preferably distributed.
 本発明の一例であるが、第2の面から遷移層までは、積層欠陥密度勾配を低くしつつ、かつ反位相領域境界面を解消させるため、表面法線軸を僅かに傾けるオフ角加工が施された珪素基板へ炭化珪素を成長させ、母結晶を形成する。オフ角の導入によって積層欠陥の伝播方向が一方向化して、それらの会合消滅機構が消失し、成長にともなう積層欠陥密度の減少機構が消滅する。主表面(結晶の表面の中で、最も露出している面積が大きい部分)が(001)面である珪素基板に炭化珪素が成長する場合、[110]方向に数度程度の角度を付けることによって(111)面に平行に伝播する積層欠陥が遮断され、積層欠陥伝播方位は(-1-11)面に平行な方向に限定される。また、[-110]方向に数度程度の角度を付けると(-111)面に平行に伝播する積層欠陥が遮断され、積層欠陥伝播方位は(1-11)面に平行な方向に限定される。ただし、反位相領域境界面を解消させるために、[110]方向へ付与する角度と[-110]方向に付与する角度は異なる角度にする必要がある。これにより、ステップフロー成長方向に異方性が生じ、反位相境界の残存が抑制され、かつ、残存する積層欠陥の伝搬方向を制御可能となる。 As an example of the present invention, from the second surface to the transition layer, off-angle machining is performed by slightly tilting the surface normal axis in order to reduce the stacking fault density gradient and eliminate the anti-phase region boundary surface. Silicon carbide is grown on the formed silicon substrate to form a mother crystal. By introducing the off-angle, the propagation direction of stacking faults becomes unidirectional, the mechanism of disappearance of their association disappears, and the mechanism of decrease of stacking fault density accompanying growth disappears. When silicon carbide grows on a silicon substrate whose main surface (the most exposed part of the crystal surface has the largest exposed area) is the (001) plane, an angle of about several degrees is given in the [110] direction. Thus, the stacking fault propagating in parallel to the (111) plane is blocked, and the stacking fault propagation direction is limited to the direction parallel to the (−1-11) plane. Further, if an angle of about several degrees is formed in the [−110] direction, the stacking fault propagating in parallel to the (−111) plane is cut off, and the stacking fault propagation direction is limited to the direction parallel to the (1-11) plane. The However, in order to eliminate the anti-phase region boundary surface, the angle applied in the [110] direction needs to be different from the angle applied in the [−110] direction. As a result, anisotropy occurs in the step flow growth direction, the remaining antiphase boundary is suppressed, and the propagation direction of the remaining stacking faults can be controlled.
 さらに、炭化珪素基板の反り低減は、成長初期に積層欠陥密度勾配が急峻となる遷移層を設け、その後は低い積層欠陥密度を維持する炭化珪素結晶(母結晶)を成長させることでも実現される。つまり、この場合、遷移層から第1の面までに母結晶が形成されている。この手段を用いる場合にも、反位相領域境界面を解消させるため、あらかじめ[110]方位と[-110]方位にオフ角が付与された珪素基板を用いる。 Furthermore, the warpage of the silicon carbide substrate can be reduced by providing a transition layer with a steep stacking fault density gradient at the beginning of growth and then growing a silicon carbide crystal (mother crystal) that maintains a low stacking fault density. . That is, in this case, a mother crystal is formed from the transition layer to the first surface. Even when this means is used, in order to eliminate the anti-phase region boundary surface, a silicon substrate in which off angles are given in advance to the [110] direction and the [−110] direction is used.
 なお、本発明者らは、特許文献2に記載の1方向に平行な尾根を有する起伏が形成された珪素基板上に炭化珪素を成長させる技術を用いる場合、起伏延在方向と残存Si-SF伝搬方向に相関があることを確認している。具体的には、炭化珪素成長表面において、起伏延在方向と直交方向にSi-SFが露出する。
 本発明者らは、このような起伏を有する基板上に炭化珪素を成長させたところ、得られた炭化珪素膜の反り量が面内で等方的ではなく、起伏延在方向とその直交方向で異方性がある(反り量の曲率半径が異なる)ことを見いだした。これは、起伏延在方向とSi-SF会合消滅機構およびC-SF消滅機構には相関があるため、このような起伏を有する基板上に炭化珪素を成長させた場合、各SFの消滅機構の違いによってその密度勾配が異なり、基板内部の応力は面内で等方的ではなく起伏延在方向に応じて異方性が生じるためであると考えた。
 本発明の目的である低欠陥密度かつ低反り量の基板を、より効果的に実現するためには、このようなSFの会合消滅機構の異方性は避ける方が好ましいと考えられる。また、炭化珪素基板に面内反り量の異方性が存在することは、デバイス製造プロセスにおいても好ましくないと考えられる。すなわち、炭化珪素基板面内での曲率半径は同等の方が好ましいので、特許文献2のような一方向に延在する起伏を有する基板を用いるよりは、[110]方位と[-110]方位に異なるオフ角が付与された基板を用いた方が好ましい。
In addition, when using the technique of growing silicon carbide on a silicon substrate on which undulations having ridges parallel to one direction described in Patent Document 2 are used, the inventors of the present invention have a undulation extending direction and residual Si—SF. It is confirmed that there is a correlation in the propagation direction. Specifically, Si—SF is exposed in the direction perpendicular to the undulation extending direction on the silicon carbide growth surface.
When the present inventors grew silicon carbide on a substrate having such undulations, the amount of warpage of the obtained silicon carbide film was not isotropic in the plane, and the undulation extending direction and its orthogonal direction And found that there is anisotropy (the curvature radius of the warping amount is different). This is because there is a correlation between the undulation extension direction and the Si-SF association annihilation mechanism and the C-SF annihilation mechanism. Therefore, when silicon carbide is grown on a substrate having such undulations, the annihilation mechanism of each SF The density gradient differs depending on the difference, and it is considered that the stress inside the substrate is not isotropic in the plane but anisotropy occurs depending on the undulation extending direction.
In order to more effectively realize the low defect density and low warpage substrate which is the object of the present invention, it is considered preferable to avoid such anisotropy of the SF association / annihilation mechanism. In addition, the presence of an in-plane warp anisotropy in the silicon carbide substrate is considered undesirable in the device manufacturing process. That is, since the same radius of curvature in the silicon carbide substrate surface is preferable, the [110] orientation and the [−110] orientation are used rather than using a substrate having undulations extending in one direction as in Patent Document 2. It is preferable to use a substrate having a different off angle.
 本発明によれば、製造プロセスを複雑化することなく、低反り量、かつ低欠陥密度の炭化珪素基板を製造可能である。 According to the present invention, a silicon carbide substrate having a low warpage and a low defect density can be manufactured without complicating the manufacturing process.
立方晶炭化珪素における積層欠陥構造を示す模式図であるIt is a schematic diagram showing a stacking fault structure in cubic silicon carbide 本発明における基板断面の積層欠陥プロファイルを示す図である。It is a figure which shows the stacking fault profile of the board | substrate cross section in this invention. 本発明における膜厚と応力分布を示す図である。It is a figure which shows the film thickness and stress distribution in this invention. 本発明における膜厚とSF密度分布を示す図である。It is a figure which shows the film thickness and SF density distribution in this invention. 本発明における遷移層中に設けられた空間を説明するための概念図である。It is a conceptual diagram for demonstrating the space provided in the transition layer in this invention. 本発明における実施例を示す断面図である。It is sectional drawing which shows the Example in this invention. 本発明における実施例を示す断面図である。It is sectional drawing which shows the Example in this invention. 本発明における基板反り量の定義を示す概念図である。It is a conceptual diagram which shows the definition of the board | substrate curvature amount in this invention. 本発明における実施例を示す図である。It is a figure which shows the Example in this invention.
 以下、本発明を実施するための形態について説明するが、本発明はこれに限られるものではない。
 立方晶の炭化珪素基板であり、積層欠陥密度がN1である主表面が{001}面に平行であり、これと平行、かつ積層欠陥密度がN2である裏面を有する板状の結晶であり、主表面とこれに平行な裏面との距離がtと定義された場合、空間を有する遷移層における面欠陥密度勾配が20×(N2-N1)/t以上、望ましくは30×(N2-N1)/t以上、さらに望ましくは50×(N2-N1)/t以上である。なお、N1の数値範囲としては700本/cm以下とするのが好ましい。炭化珪素基板の主表面において積層欠陥密度が700本/cm以下ならば、高機能半導体素子としての機能を十分に発揮できる。更に、700本/cmを超える(例えばN1=5000本/cmである)場合、積層欠陥密度を急峻に低減させる必要がそもそもなくなり、上述のような応力の問題が発生しない。そのため、本発明はN1の数値範囲を700本/cm以下とする場合に特に技術的意義がある。
Hereinafter, although the form for implementing this invention is demonstrated, this invention is not limited to this.
It is a cubic silicon carbide substrate, a main surface having a stacking fault density of N1 is parallel to the {001} plane, and is a plate-like crystal having a back surface parallel to this and a stacking fault density of N2. When the distance between the main surface and the back surface parallel to this is defined as t, the surface defect density gradient in the transition layer having a space is 20 × (N2-N1) / t or more, preferably 30 × (N2-N1) / T or more, more desirably 50 × (N2-N1) / t or more. The numerical range of N1 is preferably 700 / cm or less. If the stacking fault density is 700 / cm or less on the main surface of the silicon carbide substrate, the function as a high-performance semiconductor element can be sufficiently exhibited. Furthermore, when it exceeds 700 lines / cm (for example, N1 = 5000 lines / cm), it is not necessary to sharply reduce the stacking fault density in the first place, and the above-described stress problem does not occur. Therefore, the present invention is particularly technically significant when the numerical value range of N1 is 700 lines / cm or less.
 ところで、半導体プロセス許容量である反り量(6インチ口径で50μmの反り)を具現化するためには、基板内の応力を制御する(90MPa以下とする)必要がある。図3は、数値解析によって得られた、代表的な立方晶炭化珪素基板の板厚方向の応力分布である。図4は、同じく代表的な立方晶炭化珪素基板の板厚方向の積層欠陥密度分布である。図3、図4から、応力と積層欠陥密度分布には密接な相関関係があることがわかった。図3、図4から、応力は基板裏面より100μmまでに大幅に変化するとともに、積層欠陥密度も同様に基板裏面より100μm程度の厚さまでに大幅に減少することがわかる。基板内部の応力は、積層欠陥密度変化に伴って、ほぼ100μmまでの応力で決定されることが明らかとなっている。
 このような遷移層を具現化すれば、炭化珪素結晶基板の反りに与える影響は少なく、半導体製造プロセスの許容量である反り50μmは実現される。
By the way, in order to realize a warpage amount (a warp of 50 μm with a 6-inch diameter) that is a semiconductor process allowable amount, it is necessary to control the stress in the substrate (90 MPa or less). FIG. 3 is a stress distribution in the thickness direction of a typical cubic silicon carbide substrate obtained by numerical analysis. FIG. 4 shows a stacking fault density distribution in the thickness direction of a typical cubic silicon carbide substrate. 3 and 4 show that there is a close correlation between stress and stacking fault density distribution. 3 and 4, it can be seen that the stress changes greatly from the back surface of the substrate to 100 μm, and the stacking fault density also decreases from the back surface of the substrate to a thickness of about 100 μm. It has become clear that the stress inside the substrate is determined by a stress of up to about 100 μm with a change in stacking fault density.
If such a transition layer is embodied, there is little influence on the warp of the silicon carbide crystal substrate, and a warp of 50 μm, which is an allowable amount of the semiconductor manufacturing process, is realized.
 主表面とこれに平行な裏面との距離がtと定義された場合、遷移層の厚さtはt/20以下が好ましい。遷移層内では、当然のことながら積層欠陥密度変化勾配が存在する。この勾配によって遷移層内には応力が発生する。ただし、遷移層の厚さtがt/20以下であれば、炭化珪素結晶基板全体の反りに与える影響は少なく、より確実に、半導体プロセスの許容量である反り50μmは実現される。なお、遷移層の厚さtは、望ましくはt/30以下である。 When the distance between the main surface and the back surface parallel thereto is defined as t, the transition layer thickness t T is preferably t / 20 or less. In the transition layer, there is of course a stacking fault density change gradient. This gradient generates stress in the transition layer. However, if the thickness t T of the transition layer is t / 20 or less, there is little influence on the warp of the entire silicon carbide crystal substrate, and a warp of 50 μm, which is an allowable amount of the semiconductor process, can be realized more reliably. Note that the transition layer thickness t T is desirably t / 30 or less.
 また、断面視において、主表面から基板内部に向け0.05×t以内、望ましくは0.04×t以内、さらに望ましくは0.01×t以内に複数の空間(又は側壁)が均等に存在している。そして、断面視において、空間が分布している層は、主表面と略平行であり、空間を形成する部分の側壁は{110}面に略平行である。この組み合わせにおいては、立方晶炭化珪素に含まれる積層欠陥は最稠密面である{111}面と平行に伝播するので、積層欠陥と空間はθ=35.3度で交わる。個々の空間の幅(S)、間隔(W)、高さ(H)は厳密に一致している必要は無いが、Wは100nm以上100μm以下、望ましくは1μm以上50μm以下、さらに望ましくは2μmから20μm以内である。なぜならば、Wが狭くなるにつれて空間形成のための加工が困難になるばかりか、炭化珪素基板における空間の体積占有率が増えてしまい、半導体デバイスとして動作させる場合の基板抵抗が増大してしまうためである。一方、Wが大きくなりすぎると、遷移層内における空間の数が減ってしまう。また、欠陥の伝播を阻止するためには、空間の高さHを大きくしなければならない。高さHを大きくすることは、炭化珪素基板自体の厚さを大きくすることにつながる。そうなると、高さHを大きくして空間の体積が増加する以上に、炭化珪素基板の体積増加につながる。その結果、炭化珪素基板における空間の体積占有率が減少してしまい、熱歪を吸収しきれなくなってしまう。Hは上記Wから(式1)に従って特定されるが、極端に小さい場合にはその加工が困難となるばかりか、空間の体積占有率が減少してしまい、熱歪を吸収しきれなくなってしまう。一方、極端に大きい(100μm以上)加工も困難であり、かつ、空間の体積占有率が増えてしまい、遷移層厚さを増やしてしまい効果的な反り低減を阻害してしまう。空間の幅(S)に関しては100nm以上100μm以下、望ましくは1μm以上50μm以下、さらにのぞましくは2μmから20μm以内である。なぜならば、Sが狭くなるにつれて空間形成のための加工が困難になるばかりか、空間の体積占有率が減少し熱歪を吸収しきれなくなってしまう。一方、Sが極端に大きくなると空間の体積占有率が増加し半導体デバイスとして動作させる場合の基板抵抗が増大してしまうばかりか、Wを望ましい値に設定することが困難となる。
 あるいは、主表面から基板裏面に向け0.95×t以上、望ましくは0.96×t以上に複数の空間が均等に存在しても本発明は実現される。
Also, in a cross-sectional view, a plurality of spaces (or side walls) are present uniformly within 0.05 × t, preferably within 0.04 × t, more preferably within 0.01 × t from the main surface to the inside of the substrate. is doing. In a sectional view, the layer in which the space is distributed is substantially parallel to the main surface, and the side wall of the portion forming the space is substantially parallel to the {110} plane. In this combination, the stacking fault contained in cubic silicon carbide propagates in parallel with the {111} plane which is the most dense surface, so the stacking fault and the space intersect at θ = 35.3 degrees. The width (S), interval (W), and height (H) of each space do not need to be exactly the same, but W is 100 nm to 100 μm, preferably 1 μm to 50 μm, and more preferably 2 μm. Within 20 μm. This is because, as W becomes narrower, not only processing for forming a space becomes difficult, but also the volume occupancy of the space in the silicon carbide substrate increases, and the substrate resistance when operating as a semiconductor device increases. It is. On the other hand, if W becomes too large, the number of spaces in the transition layer will decrease. Also, in order to prevent the propagation of defects, the height H of the space must be increased. Increasing the height H leads to increasing the thickness of the silicon carbide substrate itself. Then, the volume of the silicon carbide substrate is increased more than the height H is increased and the volume of the space is increased. As a result, the volume occupancy of the space in the silicon carbide substrate is reduced, and thermal strain cannot be absorbed. H is specified from the above W according to (Equation 1), but if it is extremely small, not only the processing becomes difficult, but also the volume occupancy of the space decreases, and thermal strain cannot be absorbed completely. . On the other hand, extremely large processing (100 μm or more) is also difficult, and the volume occupancy of the space increases, increasing the transition layer thickness and inhibiting effective warpage reduction. The width (S) of the space is 100 nm to 100 μm, preferably 1 μm to 50 μm, more preferably 2 μm to 20 μm. This is because, as S becomes narrower, the processing for forming the space becomes difficult, and the volume occupancy of the space decreases, making it impossible to absorb the thermal strain. On the other hand, if S becomes extremely large, the volume occupancy of the space increases and the substrate resistance when operating as a semiconductor device increases, and it becomes difficult to set W to a desirable value.
Alternatively, the present invention can be realized even when a plurality of spaces are uniformly present at 0.95 × t or more, preferably 0.96 × t or more from the main surface to the back surface of the substrate.
 主表面と平行な裏面から遷移層までは積層欠陥密度勾配を低くしつつ、反位相領域境界面を解消させるため、表面法線軸を僅かに傾けるオフ角加工が施された珪素基板へ炭化珪素を成長させる。オフ角加工としては、(001)面の表面法線軸を[110]方向に0.1°から54.7°、望ましくは1°から6°の範囲で傾ける。同様に、(001)面の表面法線軸を[-110]方向にも0.1°から54.7°、望ましくは1°から6°の範囲で傾ける。ただし、それぞれの方位に付与する角度が同等であると反位相領域境界面は残存するため、[110]方向へ付与する角度と[-110]方向に付与する角度は1°以上の差をつける。それぞれの角度が浅い(オフ角が小さい)方が、対向する反位相領域境界どうしが会合する確率が高くなるため、少ない成長膜厚で欠陥低減効果が得られやすい。 From the back surface parallel to the main surface to the transition layer, silicon carbide is applied to the silicon substrate that has been subjected to off-angle processing that slightly tilts the surface normal axis in order to eliminate the anti-phase region boundary surface while reducing the stacking fault density gradient. Grow. As the off-angle processing, the surface normal axis of the (001) plane is tilted in the range of 0.1 ° to 54.7 °, preferably 1 ° to 6 ° in the [110] direction. Similarly, the surface normal axis of the (001) plane is also tilted in the range of 0.1 ° to 54.7 °, preferably 1 ° to 6 ° in the [−110] direction. However, if the angle applied to each direction is equal, the anti-phase region boundary surface remains, so that the angle applied in the [110] direction and the angle applied in the [−110] direction have a difference of 1 ° or more. . When each angle is shallower (off angle is smaller), the probability that the opposite anti-phase region boundaries meet each other increases, so that the defect reduction effect can be easily obtained with a small growth film thickness.
 なお、上記の記載においては、第1の面または第2の面に遷移層が存在する場合について述べたが、もちろん、第1の面と第2の面との間に遷移層が存在しても構わない。また、複数の遷移層が存在していても構わない。結局のところ、面欠陥密度変化勾配を急峻とする遷移層、および、第1の面または第2の面から遷移層までは積層欠陥密度変化勾配を低くする母結晶が存在すれば良い。 In the above description, the case where the transition layer is present on the first surface or the second surface has been described. Of course, the transition layer is present between the first surface and the second surface. It doesn't matter. A plurality of transition layers may exist. After all, a transition layer having a steep surface defect density change gradient and a mother crystal that lowers the stacking fault density change gradient from the first surface or the second surface to the transition layer may be present.
 (001)面の表面法線軸を[110]方向に4°傾け、(001)面の表面法線軸を[-110]方向にも2°傾けた珪素基板を、CVD装置内にてCとHの混合雰囲気で加熱し、極薄の炭化珪素層を形成した。基板の口径は6インチである。基板は1350℃まで加熱した。原料ガスであるC、およびキャリアガスであるHは、室温より基板表面に供給した。供給量と圧力は、表1の基板昇温条件に示したとおりである。 A silicon substrate in which the surface normal axis of the (001) plane is tilted by 4 ° in the [110] direction and the surface normal axis of the (001) plane is also tilted by 2 ° in the [−110] direction is placed in a C 2 H in a CVD apparatus. Heating was performed in a mixed atmosphere of 2 and H 2 to form an ultrathin silicon carbide layer. The diameter of the substrate is 6 inches. The substrate was heated to 1350 ° C. The source gas C 2 H 2 and the carrier gas H 2 were supplied to the substrate surface from room temperature. The supply amount and pressure are as shown in the substrate temperature raising conditions in Table 1.
Figure JPOXMLDOC01-appb-T000002
 
Figure JPOXMLDOC01-appb-T000002
 
 基板表面が1350℃に到達した後は、基板を15分間、前述したCと水素雰囲気を保持した。
 上記方法で極薄炭化珪素層を形成させた後に、温度1350℃でSiHCl、C そしてHを供給することにより炭化珪素を成長させた。この場合の成長条件は、表2に示したとおりである。成長時の圧力は、反応室-ポンプ間に設置した圧力調整バルブにて調整した。
After the substrate surface reached 1350 ° C., the substrate was maintained in the above-described C 2 H 2 and hydrogen atmosphere for 15 minutes.
After forming an ultrathin silicon carbide layer by the above method, silicon carbide was grown by supplying SiH 2 Cl 2 , C 2 H 2 and H 2 at a temperature of 1350 ° C. The growth conditions in this case are as shown in Table 2. The pressure during the growth was adjusted with a pressure regulating valve installed between the reaction chamber and the pump.
Figure JPOXMLDOC01-appb-T000003
 
Figure JPOXMLDOC01-appb-T000003
 
 以上の条件で6時間成長をおこない、300μmの立方晶炭化珪素をSi基板上に成長させた。なお、炭化珪素の成長に用いるためのSiの原料ガスとしてはSiH、SiCl、SiHClを、Cの原料ガスとしてはCH、C、C、Cを用いても差し支えない。
 成長後、珪素基板をフッ酸と硝酸の混酸でエッチングし、単独の立方晶炭化珪素基板を作製した。
 珪素基板との界面近傍には格子ミスマッチに起因する積層欠陥層が高密度で存在する。この層が存在すると、本発明記載の遷移層を設けても反りを低減することができない。そこで本発明では、0.1μmダイヤモンド砥粒を用いた研磨処理を施し、珪素基板界面から10μmの積層欠陥層を除去する。
 主表面と、この研磨処理によって得られた主表面に平行な裏面の積層欠陥密度を測定するために500℃溶融KOHに5分間浸漬させた。その後、欠陥が顕在化した基板に対して、光学顕微鏡観察を実施したところ表面における積層欠陥密度は1.8×10本/cmであった。この表面は、遷移層開始面(厳密にはこの表面の下部5μm)の積層欠陥密度および反位相領域境界とほぼ同程度であることを、別の基板により確認済であるため、ここでは遷移層の開始面における特性として近似することにする。すなわち、遷移層の開始面における積層欠陥密度は1.8×10本/cmであった。また、遷移層の開始面上において反位相領域境界面は認められなかった。
 また、光学顕微鏡観察により裏面、すなわち第2の面の積層欠陥密度は2.1×10本/cmであった。
 なお、ここで、面欠陥(積層欠陥)密度測定は、以下のように行った。[110]方向への4°オフかつ[-110]方向への2°オフ基板を用いた場合、(001)表面に顕在化する(111)面欠陥は、主として[-110]方向に平行に生じている。[-110]方向に平行な面欠陥は、これと直交する[110]方向の単位長と交差する本数(単位:本/cm)として算出し、[110]方向に平行な面欠陥は、これと直交する[-110]方向の単位長と交差する本数(単位:本/cm)として算出し、これらの平均値を面欠陥(積層欠陥)密度とした。
Growth was performed for 6 hours under the above conditions, and 300 μm cubic silicon carbide was grown on the Si substrate. Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3 , and C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
After growth, the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
Near the interface with the silicon substrate, stacking fault layers due to lattice mismatch exist at high density. If this layer is present, the warp cannot be reduced even if the transition layer according to the present invention is provided. Therefore, in the present invention, a polishing process using 0.1 μm diamond abrasive grains is performed to remove the 10 μm stacking fault layer from the silicon substrate interface.
In order to measure the stacking fault density of the main surface and the back surface parallel to the main surface obtained by this polishing treatment, it was immersed in molten KOH at 500 ° C. for 5 minutes. Then, when the optical microscope observation was implemented with respect to the board | substrate with which the defect became obvious, the stacking fault density in the surface was 1.8 * 10 < 5 > / cm. Since it has been confirmed by another substrate that this surface is almost the same as the stacking fault density and antiphase region boundary of the transition layer starting surface (strictly, the lower part 5 μm of this surface), here the transition layer Approximate the characteristics on the starting surface of That is, the stacking fault density at the start surface of the transition layer was 1.8 × 10 5 / cm. Further, no anti-phase region interface was observed on the start surface of the transition layer.
The stacking fault density on the back surface, that is, the second surface was 2.1 × 10 5 pieces / cm as observed with an optical microscope.
Here, the surface defect (stacking defect) density measurement was performed as follows. When using a 4 ° off substrate in the [110] direction and a 2 ° off substrate in the [−110] direction, the (111) plane defect that appears on the (001) surface is mainly parallel to the [−110] direction. Has occurred. The surface defects parallel to the [−110] direction are calculated as the number (unit: lines / cm) intersecting the unit length in the [110] direction perpendicular to the [110] direction. Is calculated as the number (unit: lines / cm) intersecting the unit length in the [−110] direction orthogonal to the surface length, and the average value thereof is defined as the surface defect (stacking defect) density.
 次いで、真空蒸着法を用いて主表面上にCr/Auの積層構造の金属膜層を設けた。それぞれの層の厚さはCr25nm、Au110nmとした。さらにその上層に2μmのポジ型レジストをスピン塗布し、フォトマスクにて正方形のパターン配列を露出するように紫外線で露光した。マスクパターンのラインの辺は基板の{110}方位に平行になるようにアライメントした。正方形パターンの1辺は2μmであり、パターン同士の間隔は9μmとした。次いで、レジストを現像して露光領域に開口部を設け、Ni電解メッキにより開口部に0.5μmの厚さのNiを形成した。その後、有機洗浄にてレジストを完全に除去し、間隔が9μm間隔で配列された1辺が2μmの正方形の箇所に厚さ0.5μmのNiパターンを得た。その後、誘導結合プラズマ反応性イオンエッチングにより垂直形状のメサ加工を行った。正方形以外の部分のクロム、金と共に炭化珪素を加工する。その後正方形の箇所に残るCr、Au、Niをウェットエッチングにより完全に除去した。さらに過酸化水素水+硫酸混合溶液(1:1)、HF溶液にて洗浄した。洗浄した後、熱処理装置を用いて熱酸化膜を約100nm形成した。形成した熱酸化膜を、希フッ酸により除去した。このような工程を経て、主表面に9μm間隔に配列された、1辺が2μm、高さ5μmの離散した立方晶炭化珪素(図6の符号62)が形成された。なお、厳密に言うと遷移層の開始面は、上記の手法で形成されたメサ加工部分の底部(後の空間(74)の下部分の底部)又はその近傍を含む面であるが、厳密に当該底部に該当する部分の積層欠陥密度を測定するのが極めて困難であること、そして、約300μmの全膜厚に対し板厚方向における5μmの違いでは、積層欠陥密度においてほとんど変化がないことから、メサ加工前の主表面における積層欠陥密度を、遷移層の開始面における積層欠陥密度(N3)として用いている。「遷移層」は、当該底部に該当する部分に加え、その近傍部分も含む。 Next, a metal film layer having a laminated structure of Cr / Au was provided on the main surface using a vacuum deposition method. The thickness of each layer was set to Cr 25 nm and Au 110 nm. Further, a positive resist of 2 μm was spin-coated on the upper layer, and exposed with ultraviolet rays so as to expose a square pattern array with a photomask. The mask pattern line side was aligned so as to be parallel to the {110} orientation of the substrate. One side of the square pattern was 2 μm, and the interval between the patterns was 9 μm. Next, the resist was developed to provide an opening in the exposed region, and Ni having a thickness of 0.5 μm was formed in the opening by Ni electroplating. Thereafter, the resist was completely removed by organic cleaning, and a Ni pattern having a thickness of 0.5 μm was obtained in a square portion with a side of 2 μm arranged at intervals of 9 μm. Thereafter, vertical mesa processing was performed by inductively coupled plasma reactive ion etching. Silicon carbide is processed together with chromium and gold other than the square. Thereafter, Cr, Au, and Ni remaining in the square portion were completely removed by wet etching. Further, it was washed with a hydrogen peroxide solution + sulfuric acid mixed solution (1: 1) and an HF solution. After cleaning, a thermal oxide film was formed to a thickness of about 100 nm using a heat treatment apparatus. The formed thermal oxide film was removed with dilute hydrofluoric acid. Through these steps, discrete cubic silicon carbide (reference numeral 62 in FIG. 6) arranged on the main surface at intervals of 9 μm and having a side of 2 μm and a height of 5 μm was formed. Strictly speaking, the start surface of the transition layer is a surface including the bottom of the mesa-processed portion formed by the above method (the bottom of the lower portion of the later space (74)) or the vicinity thereof. It is extremely difficult to measure the stacking fault density of the portion corresponding to the bottom, and there is almost no change in the stacking fault density with a difference of 5 μm in the thickness direction with respect to the total film thickness of about 300 μm. The stacking fault density on the main surface before mesa processing is used as the stacking fault density (N3) on the start surface of the transition layer. The “transition layer” includes not only the portion corresponding to the bottom portion but also the vicinity thereof.
 図6には、パターニング後の炭化珪素基板の断面構造を示す。61は炭化珪素層、62は離散的に配列された炭化珪素である。図7には、図6の炭化珪素基板上へ炭化珪素成長させたものの断面構造を示す。71は炭化珪素層、72は離散的に選択成長した炭化珪素、73は離散的に形成された炭化珪素上に成長した炭化珪素層、74は空間である。図6の後、表3の条件で60分間の成長をおこない、5μmの立方晶炭化珪素((72)および(73))を成長させた。この際、図7に示すように離散的に選択成長した炭化珪素(72)の上部同士が連結し、主表面、すなわち第1の面の下部0.1μmの深さに幅4μm、間隔7μm、高さ9.9μmの空間(74)を形成した。
 なお、図6及び図7はあくまで概略図である。本来ならば、図6のように空間の下部分を形成した後、炭化珪素を成長させると、基板膜厚方向および表面面内方向の両方の膜成長が生じ、図7のように離散的に選択成長した炭化珪素(72)の上部同士が連結すると共に、空間(74)の上部分も形成されることになる。
FIG. 6 shows a cross-sectional structure of the silicon carbide substrate after patterning. 61 is a silicon carbide layer, and 62 is a discretely arranged silicon carbide. FIG. 7 shows a cross-sectional structure of the silicon carbide grown on the silicon carbide substrate of FIG. Reference numeral 71 denotes a silicon carbide layer, 72 denotes silicon carbide discretely grown selectively, 73 denotes a silicon carbide layer grown on discretely formed silicon carbide, and 74 denotes a space. After FIG. 6, the growth was performed for 60 minutes under the conditions shown in Table 3, and 5 μm cubic silicon carbide ((72) and (73)) was grown. At this time, as shown in FIG. 7, the upper portions of discretely grown silicon carbide (72) are connected to each other, the main surface, that is, the depth of 0.1 μm below the first surface is 4 μm wide, 7 μm apart, A space (74) having a height of 9.9 μm was formed.
6 and 7 are schematic views only. Originally, when silicon carbide is grown after forming the lower portion of the space as shown in FIG. 6, film growth in both the substrate film thickness direction and the surface in-plane direction occurs, and discretely as shown in FIG. The upper portions of the selectively grown silicon carbide (72) are connected to each other, and the upper portion of the space (74) is also formed.
 その後、炭化珪素基板を500℃溶融KOHに5分間浸漬させた。次いで、欠陥が顕在化した基板の主表面すなわち第1面に対して、光学顕微鏡観察を実施したところ第1面の積層欠陥密度は525本/cmであった。以上のように、炭化珪素基板に対して空間が設けられた層(即ち遷移層)を形成することにより、積層欠陥密度は空間形成前に比べて大幅に低減する。なお、遷移層において空間(74)の上部分(主表面側の部分)を占める領域においては、離散的に炭化珪素を選択成長させる際に、空間(74)において積層欠陥の伝搬が阻止されているため、積層欠陥密度はもちろん大幅に低減している。更に、遷移層において空間(74)の下部分(裏面側の部分)を占める領域に関して言うと、離散的に炭化珪素を選択成長させる際に、空間(74)の下部分にわずかに新たな炭化珪素が積層するが、この積層部分においても空間(74)によって積層欠陥の伝搬が阻止されるため、積層欠陥密度は低減している。そのため、本明細書においては、炭化珪素基板内における層状の部分であって、内部に空間(74)を備えた部分を、積層欠陥密度が急峻に低減する「遷移層」としている。 Thereafter, the silicon carbide substrate was immersed in molten KOH at 500 ° C. for 5 minutes. Next, when the main surface of the substrate, ie, the first surface, on which the defects became apparent was observed with an optical microscope, the stacking fault density on the first surface was 525 / cm. As described above, by forming a layer (that is, a transition layer) in which a space is provided with respect to the silicon carbide substrate, the stacking fault density is significantly reduced as compared with that before the space is formed. Note that in the region occupying the upper part (the main surface side part) of the space (74) in the transition layer, the propagation of stacking faults is prevented in the space (74) when discretely growing silicon carbide selectively. Therefore, the stacking fault density is of course greatly reduced. Further, regarding the region that occupies the lower part (the back side part) of the space (74) in the transition layer, a slightly new carbonization is formed in the lower part of the space (74) when silicon carbide is selectively grown discretely. Although silicon is stacked, the stacking fault density is reduced because the propagation of stacking faults is prevented by the space (74) also in this stacked portion. Therefore, in the present specification, a layered portion in the silicon carbide substrate, which is provided with a space (74) therein, is a “transition layer” in which the stacking fault density is sharply reduced.
Figure JPOXMLDOC01-appb-T000004
 
Figure JPOXMLDOC01-appb-T000004
 
 本実施例では基板として(001)面を主表面とした立方晶炭化珪素を用いたが、(111)面を主表面とした場合は、空間を形成する部分の側壁は{111}面、{110}面あるいは{211}面のいずれかと平行である。また、六方晶炭化珪素であっても主表面として{0001}面を選ぶことができる。この場合は、空間を形成する部分の側壁は{11-20}面あるいは{-1100}面と平行である。この場合、積層欠陥が伝播する最稠密面は(0001)面であり、積層欠陥は主表面と30度から60度の角度で交わるので、本実施例の操作により同様の積層欠陥密度の低減がもたらされることは明白である。
 また、基板の反りの定義を図8に示す。測定方法は、キーエンス社製レーザー変位計にて基板表面を水平にスキャンし、その際の基板高さにおける最大値と最小値の差分を反りとした。
 得られた基板について、図2における各パラメータの値を表4に示す。
In this example, cubic silicon carbide having a (001) plane as the main surface was used as the substrate. However, when the (111) plane is set as the main surface, the side wall of the portion forming the space is the {111} plane, { It is parallel to either the 110} plane or the {211} plane. Further, even if it is hexagonal silicon carbide, the {0001} plane can be selected as the main surface. In this case, the side wall of the portion forming the space is parallel to the {11-20} plane or the {−1100} plane. In this case, the most dense surface through which the stacking fault propagates is the (0001) plane, and the stacking fault intersects with the main surface at an angle of 30 to 60 degrees. It is clear that it will result.
Moreover, the definition of the curvature of a board | substrate is shown in FIG. As a measuring method, the substrate surface was scanned horizontally with a laser displacement meter manufactured by Keyence Corporation, and the difference between the maximum value and the minimum value at the substrate height at that time was warped.
Table 4 shows the values of the parameters in FIG.
Figure JPOXMLDOC01-appb-T000005
 
Figure JPOXMLDOC01-appb-T000005
 
 tは、ミツトヨ社製マイクロメーターにて測定した。N1,N2,N3は、立方晶炭化珪素基板表面に対して溶融KOH処理(500℃×5min)を実施し、面欠陥を顕在化した後に顕微鏡にて単位長さ当たりの面欠陥数をカウントした。得られた数値から面欠陥密度勾配を求めたところ、勾配1は勾配2の25.3倍であり、20倍以上という条件を満たしていた。また、遷移層の厚さtは全膜厚tの約30分の1であった。この基板の反りは45μmであることを反り測定にて確認した。 t was measured with a micrometer manufactured by Mitutoyo Corporation. N1, N2, and N3 were subjected to a molten KOH treatment (500 ° C. × 5 min) on the surface of the cubic silicon carbide substrate, and after surface defects were revealed, the number of surface defects per unit length was counted with a microscope. . When the surface defect density gradient was obtained from the obtained numerical value, the gradient 1 was 25.3 times the gradient 2, and the condition of 20 times or more was satisfied. The transition layer thickness t T was about 1/30 of the total film thickness t. It was confirmed by warpage measurement that the warpage of the substrate was 45 μm.
 また、本実施例では正方形マスクパターンを使用して離散的に配列された炭化珪素を形成したが、ラインアンドスペースのマスクパターンを使用して離散的に配列された炭化珪素を形成しても良い。このようにして形成された、離散的に配列された炭化珪素を用いて空間を形成した場合、第1の面に略平行な面において互いに平行に位置し、かつ離散した空間が複数形成される。このような形態においても、本発明の効果は十分に得られる。 In this embodiment, the silicon carbide arranged discretely is formed using the square mask pattern. However, the silicon carbide arranged discretely may be formed using the mask pattern of the line and space. . When a space is formed using discretely arranged silicon carbide formed in this way, a plurality of discrete spaces are formed that are positioned in parallel to each other on a plane substantially parallel to the first surface. . Even in such a form, the effects of the present invention can be sufficiently obtained.
<比較例1>
 (001)面の表面法線軸を[110]方向に2°傾け、(001)面の表面法線軸を[-110]方向にも1°傾けた珪素基板を、CVD装置内にてCとHの混合雰囲気で加熱し、極薄の炭化珪素層を形成した。基板の口径は6インチである。基板は1350℃まで加熱した。原料ガスであるC、およびキャリアガスであるHは、室温より基板表面に供給した。供給量と圧力は、表1に示したとおりである。基板表面が1350℃に到達した後は、基板を15分間、前述したCと水素雰囲気を保持した。
<Comparative Example 1>
A silicon substrate in which the surface normal axis of the (001) plane is tilted by 2 ° in the [110] direction and the surface normal axis of the (001) plane is also tilted by 1 ° in the [−110] direction is placed in a C 2 H in a CVD apparatus. Heating was performed in a mixed atmosphere of 2 and H 2 to form an ultrathin silicon carbide layer. The diameter of the substrate is 6 inches. The substrate was heated to 1350 ° C. The source gas C 2 H 2 and the carrier gas H 2 were supplied to the substrate surface from room temperature. The supply amount and pressure are as shown in Table 1. After the substrate surface reached 1350 ° C., the substrate was maintained in the above-described C 2 H 2 and hydrogen atmosphere for 15 minutes.
 上記方法で極薄炭化珪素層を形成させた後に、温度1350℃でSiHCl、C そしてHを供給することにより炭化珪素を成長させた。この場合の成長条件は、表2に示したとおりである。成長時の圧力は、反応室-ポンプ間に設置した圧力調整バルブにて調整した。 After forming an ultrathin silicon carbide layer by the above method, silicon carbide was grown by supplying SiH 2 Cl 2 , C 2 H 2 and H 2 at a temperature of 1350 ° C. The growth conditions in this case are as shown in Table 2. The pressure during the growth was adjusted with a pressure regulating valve installed between the reaction chamber and the pump.
 以上の条件で6時間成長をおこない、300μmの立方晶炭化珪素をSi基板上に成長させた。なお、炭化珪素の成長に用いるためのSiの原料ガスとしてはSiH、SiCl、SiHClを、Cの原料ガスとしてはCH、C、C、Cを用いても差し支えない。 Growth was performed for 6 hours under the above conditions, and 300 μm cubic silicon carbide was grown on the Si substrate. Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3 , and C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
 成長後に珪素基板をフッ酸と硝酸の混酸でエッチングし、単独の立方晶炭化珪素基板を作製した。 After the growth, the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
 次に、0.1μmダイヤモンド砥粒を用いた研磨処理を施し、界面から10μmの積層欠陥層を除去する。主表面と、この研磨処理によって得られた主表面に平行な裏面の積層欠陥密度を測定するために500℃溶融KOHに5分間浸漬させた。その後、欠陥が顕在化した基板に対して、光学顕微鏡観察を実施したところ表面すなわち第1の面における積層欠陥密度は4000本/cmであり、裏面、すなわち第2の面の積層欠陥密度は8000本/cmであった。
 ここで、面欠陥(積層欠陥)密度測定は、[-110]方向に平行な面欠陥は、これと直交する[110]方向の単位長と交差する本数(単位:本/cm)として算出し、[110]方向に平行な面欠陥は、これと直交する[-110]方向の単位長と交差する本数(単位:本/cm)として算出し、これらの平均値を面欠陥(積層欠陥)密度とした。
 得られた基板について、各パラメータの値を表5に示す。本比較例では遷移層を設けていないため、N3は省略する。
Next, a polishing process using 0.1 μm diamond abrasive grains is performed to remove a 10 μm stacking fault layer from the interface. In order to measure the stacking fault density of the main surface and the back surface parallel to the main surface obtained by this polishing treatment, it was immersed in molten KOH at 500 ° C. for 5 minutes. Thereafter, when the substrate on which the defects were revealed was observed with an optical microscope, the stacking fault density on the front surface, that is, the first surface was 4000 / cm, and the stacking fault density on the back surface, that is, the second surface, was 8000. Book / cm.
Here, the surface defect (stacking defect) density measurement is calculated as the number (unit: book / cm) of plane defects parallel to the [−110] direction intersecting the unit length in the [110] direction perpendicular to the [−110] direction. The surface defects parallel to the [110] direction are calculated as the number (unit: lines / cm) intersecting the unit length in the [−110] direction orthogonal to the surface defects, and the average value of these is calculated as the surface defect (stacking defect). Density.
Table 5 shows the values of the parameters of the obtained substrate. In this comparative example, since no transition layer is provided, N3 is omitted.
Figure JPOXMLDOC01-appb-T000006
 
Figure JPOXMLDOC01-appb-T000006
 
 tは、ミツトヨ社製マイクロメーターにて測定した。N1,N2は、3C-SiC基板表面に対して溶融KOH処理(500℃×5min)を実施し、面欠陥を顕在化した後に顕微鏡にて単位長さあたりの面欠陥数をカウントした。この基板の反りは450μmであることを反り測定にて確認した。
 上述した方法にて得られた基板について、さらに遷移層を設けた。遷移層の形成は実施例1の方法に準じた。その結果、得られた基板における図2における各パラメータ値を表6に示す。
t was measured with a micrometer manufactured by Mitutoyo Corporation. N1 and N2 were subjected to molten KOH treatment (500 ° C. × 5 min) on the surface of the 3C—SiC substrate, and surface defects were revealed, and the number of surface defects per unit length was counted with a microscope. The warpage of the substrate was confirmed to be 450 μm by warpage measurement.
A transition layer was further provided on the substrate obtained by the method described above. The transition layer was formed according to the method of Example 1. As a result, each parameter value in FIG. 2 for the obtained substrate is shown in Table 6.
Figure JPOXMLDOC01-appb-T000007
 
Figure JPOXMLDOC01-appb-T000007
 
 tおよびtは、ミツトヨ社製マイクロメーターにて測定した。N1、N2、N3は、3C-SiC基板表面に対して溶融KOH処理(500℃、5min)を実施し、面欠陥を顕在化した後に顕微鏡にて単位長さあたりの面欠陥数をカウントした。得られた数値から勾配を求めたところ、勾配1は勾配2の約13.7倍であり、20倍以上という条件を満たしていなかった。この基板の反りは500μmであることを反り測定にて確認した。
 すなわち、勾配1が勾配2の20倍未満の場合、基板の反りは大きくなることがわかった。
t and t T was measured by Mitutoyo micrometer. For N1, N2, and N3, the surface of the 3C—SiC substrate was subjected to a molten KOH treatment (500 ° C., 5 minutes), and surface defects were revealed, and then the number of surface defects per unit length was counted with a microscope. When the gradient was obtained from the obtained numerical value, the gradient 1 was about 13.7 times the gradient 2, and the condition of 20 times or more was not satisfied. It was confirmed by warpage measurement that the warpage of the substrate was 500 μm.
That is, it was found that when the gradient 1 is less than 20 times the gradient 2, the warpage of the substrate becomes large.
<比較例2>
 比較例2は、基本的に実施例1に準じる方法によって行った。なお、正方形の離散領域の1辺を2μm、間隔を16μmとしている。
 まず、300μmの立方晶炭化珪素を6インチ口径の珪素基板上に成長させた。ただし、珪素基板は、(001)面の表面法線軸を[110]方向に3°傾け、(001)面の表面法線軸を[-110]方向に2°傾けた珪素基板を用いた。
 成長後に珪素基板をフッ酸と硝酸の混酸でエッチングし、単独の立方晶炭化珪素基板を作製した。
<Comparative Example 2>
Comparative Example 2 was performed by a method basically according to Example 1. Note that one side of the square discrete region is 2 μm and the interval is 16 μm.
First, 300 μm cubic silicon carbide was grown on a 6-inch diameter silicon substrate. However, a silicon substrate having a (001) surface normal axis inclined by 3 ° in the [110] direction and a (001) surface normal axis inclined by 2 ° in the [−110] direction was used as the silicon substrate.
After the growth, the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
 次に、0.1μmダイヤモンド砥粒を用いた研磨処理を施し、界面から10μmの積層欠陥層を除去した。主表面と、この研磨処理によって得られた主表面に平行な裏面の積層欠陥密度を測定するために500℃溶融KOHに5分間浸漬させた。その後、欠陥が顕在化した基板に対して、光学顕微鏡観察を実施したところ表面(すなわち第1の面)における積層欠陥密度は7600本/cmであり、裏面(すなわち第2の面)の積層欠陥密度は8000本/cmであった。
 ここで、面欠陥(積層欠陥)密度測定は、[-110]方向に平行な面欠陥は、これと直交する[110]方向の単位長と交差する本数(単位:本/cm)として算出し、[110]方向に平行な面欠陥は、これと直交する[-110]方向の単位長と交差する本数(単位:本/cm)として算出し、これらの平均値を面欠陥(積層欠陥)密度とした。
Next, a polishing process using 0.1 μm diamond abrasive grains was performed, and a 10 μm stacking fault layer was removed from the interface. In order to measure the stacking fault density of the main surface and the back surface parallel to the main surface obtained by this polishing treatment, it was immersed in molten KOH at 500 ° C. for 5 minutes. Thereafter, when an optical microscope observation was performed on the substrate on which the defects became apparent, the stacking fault density on the front surface (that is, the first surface) was 7600 / cm, and the stacking fault on the back surface (that is, the second surface). The density was 8000 pieces / cm.
Here, the surface defect (stacking defect) density measurement is calculated as the number (unit: book / cm) of plane defects parallel to the [−110] direction intersecting the unit length in the [110] direction perpendicular to the [−110] direction. The surface defects parallel to the [110] direction are calculated as the number (unit: lines / cm) intersecting the unit length in the [−110] direction orthogonal to the surface defects, and the average value of these is calculated as the surface defect (stacking defect). Density.
 そして、図7に示すように離散的に選択成長した炭化珪素(72)の上部同士が連結し、主表面(すなわち第1の面)の下部30.1μmの深さに幅4μm、間隔14μm、高さ19.9μmの空間(74)を形成した。この場合、遷移層の厚さは20μmであった。
 得られた基板について、各パラメータの値を表7に示す。
Then, as shown in FIG. 7, the upper portions of discretely grown silicon carbide (72) are connected to each other, the width is 4 μm, the distance is 14 μm, and the depth is 30.1 μm at the lower portion of the main surface (that is, the first surface). A space (74) having a height of 19.9 μm was formed. In this case, the thickness of the transition layer was 20 μm.
Table 7 shows the value of each parameter for the obtained substrate.
Figure JPOXMLDOC01-appb-T000008
 
Figure JPOXMLDOC01-appb-T000008
 
 tおよびtは、ミツトヨ社製マイクロメーターにて測定した。N1,N2,N3は、3C-SiC基板表面に対して溶融KOH処理(500℃、5min)を実施し、面欠陥を顕在化した後に顕微鏡にて単位長さ当たりの面欠陥数をカウントした。得られた数値から、面欠陥密度勾配を求めたところ、勾配1は勾配2の15.8倍であり、20倍以上という条件を満たしていなかった。この基板の反りは250μmであることを反り測定にて確認した。 t and t T was measured by Mitutoyo micrometer. For N1, N2, and N3, the surface of the 3C—SiC substrate was subjected to a molten KOH treatment (500 ° C., 5 min), surface defects were revealed, and the number of surface defects per unit length was counted with a microscope. When the surface defect density gradient was determined from the obtained numerical values, the gradient 1 was 15.8 times the gradient 2 and did not satisfy the condition of 20 times or more. It was confirmed by warpage measurement that the warpage of the substrate was 250 μm.
 (001)面の表面法線軸を[110]方向に4°傾け、(001)面の表面法線軸を[-110]方向にも2°傾けた珪素基板を、CVD装置内にてCとHの混合雰囲気で加熱し、極薄の炭化珪素層を形成した。基板の口径は6インチである。基板は1350℃まで加熱した。原料ガスであるC、およびキャリアガスであるHは、室温より基板表面に供給した。供給量と圧力は、表1に示したとおりである。基板表面が1350℃に到達した後は、基板を15分間、前述したCと水素雰囲気を保持した。 A silicon substrate in which the surface normal axis of the (001) plane is tilted by 4 ° in the [110] direction and the surface normal axis of the (001) plane is also tilted by 2 ° in the [−110] direction is placed in a C 2 H in a CVD apparatus. Heating was performed in a mixed atmosphere of 2 and H 2 to form an ultrathin silicon carbide layer. The diameter of the substrate is 6 inches. The substrate was heated to 1350 ° C. The source gas C 2 H 2 and the carrier gas H 2 were supplied to the substrate surface from room temperature. The supply amount and pressure are as shown in Table 1. After the substrate surface reached 1350 ° C., the substrate was maintained in the above-described C 2 H 2 and hydrogen atmosphere for 15 minutes.
 上記方法で極薄炭化珪素層を形成させた後に、温度1350℃でSiHCl、C そしてHを供給することにより炭化珪素を成長させた。この場合の成長条件は、表2に示したとおりである。成長時の圧力は、反応室-ポンプ間に設置した圧力調整バルブにて調整した。 After forming an ultrathin silicon carbide layer by the above method, silicon carbide was grown by supplying SiH 2 Cl 2 , C 2 H 2 and H 2 at a temperature of 1350 ° C. The growth conditions in this case are as shown in Table 2. The pressure during the growth was adjusted with a pressure regulating valve installed between the reaction chamber and the pump.
 以上の条件で2時間成長をおこない、60μmの立方晶炭化珪素をSi基板上に成長させた。なお、炭化珪素の成長に用いるためのSiの原料ガスとしてはSiH、SiCl、SiHClを、Cの原料ガスとしてはCH、C、C、Cを用いても差し支えない。 Growth was performed for 2 hours under the above conditions, and 60 μm cubic silicon carbide was grown on the Si substrate. Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3 , and C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
 次いで、真空蒸着法を用いて主表面上にCr/Auの積層構造の金属膜層を設けた。それぞれの層の厚さはCr25nm、Au110nmとした。さらにその上層に2μmのポジ型レジストをスピン塗布し、フォトマスクにて正方形のパターン配列を露出するように紫外線で露光した。マスクパターンのラインの辺は基板の{110}方位に平行になるようにアライメントした。正方形パターンの1辺は2μmであり、パターン同士の間隔は12μmとした。次いで、レジストを現像して露光領域に開口部を設け、Ni電解メッキにより開口部に0.5μmの厚さのNiを形成した。その後、有機洗浄にてレジストを完全に除去し、間隔が12μm間隔で配列された1辺が2μmの正方形の箇所に厚さ0.5μmのNiパターンを得た。その後、誘導結合プラズマ反応性イオンエッチングにより垂直形状のメサ加工を行った。正方形以外の部分のクロム、金と共に炭化珪素を加工する。その後正方形の箇所に残るCr、Au、Niをウェットエッチングにより完全に除去した。さらに過酸化水素水+硫酸混合溶液(1:1)、HF溶液にて洗浄した。洗浄した後、熱処理装置を用いて熱酸化膜を約100nm形成した。形成した熱酸化膜を、希フッ酸により除去した。このような工程を経て、主表面に12μm間隔に配列された、1辺が2μm、高さ7μmの離散した立方晶炭化珪素(即ち、離散的に配列された炭化珪素(62))が形成された。 Next, a metal film layer having a laminated structure of Cr / Au was provided on the main surface using a vacuum deposition method. The thickness of each layer was set to Cr 25 nm and Au 110 nm. Further, a positive resist of 2 μm was spin-coated on the upper layer, and exposed with ultraviolet rays so as to expose a square pattern array with a photomask. The mask pattern line side was aligned so as to be parallel to the {110} orientation of the substrate. One side of the square pattern was 2 μm, and the distance between the patterns was 12 μm. Next, the resist was developed to provide an opening in the exposed region, and Ni having a thickness of 0.5 μm was formed in the opening by Ni electroplating. Thereafter, the resist was completely removed by organic cleaning, and a Ni pattern having a thickness of 0.5 μm was obtained at a square portion having a side of 2 μm arranged at intervals of 12 μm. Thereafter, vertical mesa processing was performed by inductively coupled plasma reactive ion etching. Silicon carbide is processed together with chromium and gold other than the square. Thereafter, Cr, Au, and Ni remaining in the square portion were completely removed by wet etching. Further, it was washed with a hydrogen peroxide solution + sulfuric acid mixed solution (1: 1) and an HF solution. After cleaning, a thermal oxide film was formed to a thickness of about 100 nm using a heat treatment apparatus. The formed thermal oxide film was removed with dilute hydrofluoric acid. Through these steps, discrete cubic silicon carbide (that is, discretely arranged silicon carbide (62)) having a side of 2 μm and a height of 7 μm arranged on the main surface at intervals of 12 μm is formed. It was.
 図6には、パターニング後の炭化珪素基板の断面構造を示す。その後、まず表8の条件で60分間の成長をおこない、8μmの立方晶炭化珪素((72)および(73))を成長させた。この際、図7に示すように離散的に選択成長した炭化珪素(72)の上部同士が連結し、主表面の下部0.1μmの深さに幅4μm、間隔10μm、高さ14.9μmの空間(74)が形成される。主表面、すなわち遷移層表面の積層欠陥密度を測定するために500℃溶融KOHに5分間浸漬させた。その後、欠陥が顕在化した基板に対して、光学顕微鏡観察を実施したところ主表面における積層欠陥密度は550本/cmであった。厳密に言うと、上記の遷移層表面は、離散的に選択成長した炭化珪素(72)の上部同士が連結した直後の部分を含む面であるが、当該部分の積層欠陥密度を厳密に測定するのが極めて困難であること、そして、連結直後から積層欠陥密度が測定可能な程度まで炭化珪素を積層させた程度では、積層欠陥密度においてほとんど変化がないことから、上記の主表面における積層欠陥密度を、遷移層表面における積層欠陥密度(N3)として用いている。「遷移層」は、離散的に選択成長した炭化珪素(72)の上部同士が連結した直後の部分に加え、連結直後から積層欠陥密度が測定可能な程度まで炭化珪素を積層させた部分も含む。 FIG. 6 shows a cross-sectional structure of the silicon carbide substrate after patterning. Thereafter, growth was performed for 60 minutes under the conditions in Table 8 to grow 8 μm cubic silicon carbide ((72) and (73)). At this time, as shown in FIG. 7, the upper parts of discretely grown silicon carbide (72) are connected to each other, the depth of the lower part of the main surface is 0.1 μm, the width is 4 μm, the interval is 10 μm, and the height is 14.9 μm. A space (74) is formed. In order to measure the stacking fault density on the main surface, that is, the transition layer surface, it was immersed in molten KOH at 500 ° C. for 5 minutes. Then, when the optical microscope observation was implemented with respect to the board | substrate with which the defect became obvious, the stacking fault density in the main surface was 550 pieces / cm. Strictly speaking, the surface of the transition layer is a surface including a portion immediately after the upper portions of discretely grown silicon carbide (72) are connected to each other, and the stacking fault density of the portion is strictly measured. Since there is almost no change in the stacking fault density when silicon carbide is stacked to the extent that the stacking fault density can be measured immediately after connection, the stacking fault density on the main surface described above Is used as the stacking fault density (N3) on the transition layer surface. The “transition layer” includes not only a portion immediately after the upper portions of discretely grown silicon carbide (72) are connected to each other, but also a portion in which silicon carbide is laminated immediately after the connection so that the stacking fault density can be measured. .
Figure JPOXMLDOC01-appb-T000009
 
Figure JPOXMLDOC01-appb-T000009
 
 次に、表2の条件で4時間の成長をおこない、最終的に珪素基板上に360μmの立方晶炭化珪素を成長させた。
 成長後に珪素基板をフッ酸と硝酸の混酸でエッチングし、単独の立方晶炭化珪素基板を作製した。
Next, growth was performed for 4 hours under the conditions shown in Table 2, and finally 360 μm cubic silicon carbide was grown on the silicon substrate.
After the growth, the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
 次に、0.1μmダイヤモンド砥粒を用いた研磨処理を施し、珪素基板と炭化珪素との界面から50μmの積層欠陥層を除去する。主表面と、この研磨処理によって得られた主表面に平行な裏面の積層欠陥密度を測定するために500℃溶融KOHに5分間浸漬させた。その後、欠陥が顕在化した基板に対して、光学顕微鏡観察を実施したところ表面(すなわち第1の面)における積層欠陥密度は510本/cmであり、裏面(すなわち第2の面)の積層欠陥密度は1.5×10本/cmであった。また、遷移層の開始面上において反位相領域境界面は認められなかった。
 得られた基板について、図8における各パラメータの値を表9に示す。当該基板は、図9に示す様に遷移層(t領域)が第2の面側の領域となる。
Next, a polishing process using 0.1 μm diamond abrasive grains is performed to remove a 50 μm stacking fault layer from the interface between the silicon substrate and silicon carbide. In order to measure the stacking fault density of the main surface and the back surface parallel to the main surface obtained by this polishing treatment, it was immersed in molten KOH at 500 ° C. for 5 minutes. Thereafter, when an optical microscope observation was performed on the substrate on which the defects became apparent, the stacking fault density on the front surface (that is, the first surface) was 510 / cm, and the stacking fault on the back surface (that is, the second surface). The density was 1.5 × 10 5 lines / cm. Further, no anti-phase region interface was observed on the start surface of the transition layer.
Table 9 shows the values of the parameters in FIG. The substrate, the transition layer as shown in FIG. 9 (t T region) becomes the second surface area.
Figure JPOXMLDOC01-appb-T000010
 
Figure JPOXMLDOC01-appb-T000010
 
 全膜厚tと遷移層厚さtは、ミツトヨ社製マイクロメーターにて測定した。N1,N2,N3は、立方晶炭化珪素基板表面に対して溶融KOH処理(500℃、5min)を実施し、面欠陥を顕在化した後に・BR>ー微鏡にて単位長さあたりの面欠陥数をカウントした。得られた数値から勾配を求めたところ、遷移層内における面欠陥密度勾配(勾配1)は遷移層領域外の面欠陥密度勾配(勾配2)の約20.6倍であり、20倍以上という条件を満たしていた。この基板の反りは40μmであることを反り測定にて確認した。
 すなわち、遷移層が第2の面側に位置していても、勾配1が勾配2の20倍以上であれば、反りが小さい基板が得られることを確認した。なお、本実施例において、遷移層の厚さが基板全体の20分の1以下を満たしていた。
Total thickness t and the transition layer thickness t T was measured by Mitutoyo micrometer. N1, N2, and N3 were subjected to molten KOH treatment (500 ° C., 5 min) on the surface of the cubic silicon carbide substrate, and surface defects were revealed. The number of defects was counted. When the gradient is obtained from the obtained numerical value, the surface defect density gradient (gradient 1) in the transition layer is about 20.6 times the surface defect density gradient (gradient 2) outside the transition layer region, which is 20 times or more. The condition was met. The warpage of the substrate was confirmed by warpage measurement to be 40 μm.
That is, even when the transition layer is located on the second surface side, it was confirmed that a substrate with small warpage can be obtained if the gradient 1 is 20 times or more the gradient 2. In this example, the thickness of the transition layer satisfied 1/20 or less of the entire substrate.
 (001)面の表面法線軸を[110]方向に3°傾け、(001)面の表面法線軸を[-110]方向にも2°傾けた珪素基板を用いたこと以外は実施例1と同様にして、表1に示した条件で珪素基板上に極薄の炭化珪素層を形成した。基板表面が1350℃に到達した後は、基板を15分間、前述したCと水素雰囲気を保持した。その後、実施例1と同様に、表2に示す条件で6時間成長を行い、300μmの立方晶炭化珪素をSi基板上に成長させた。なお、炭化珪素の成長に用いるためのSiの原料ガスとしてはSiH、SiCl、SiHClを、Cの原料ガスとしてはCH、C、C、Cを用いても差し支えない。 Example 1 except that a silicon substrate in which the surface normal axis of the (001) plane is inclined by 3 ° in the [110] direction and the surface normal axis of the (001) plane is also inclined by 2 ° in the [−110] direction is used. Similarly, an ultrathin silicon carbide layer was formed on the silicon substrate under the conditions shown in Table 1. After the substrate surface reached 1350 ° C., the substrate was maintained in the above-described C 2 H 2 and hydrogen atmosphere for 15 minutes. Thereafter, in the same manner as in Example 1, growth was performed for 6 hours under the conditions shown in Table 2, and 300 μm cubic silicon carbide was grown on the Si substrate. Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3 , and C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
 成長後、珪素基板をフッ酸と硝酸の混酸でエッチングし、単独の立方晶炭化珪素基板を作製した。 After growth, the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
 珪素基板との界面近傍には格子ミスマッチに起因する積層欠陥層が高密度で存在する。この層が存在すると、本発明記載の遷移層を設けても反りを低減することができない。そこで本発明では、0.1μmダイヤモンド砥粒を用いた研磨処理を施し、珪素基板界面から10μmの積層欠陥層を除去する。 Near the interface with the silicon substrate, stacking fault layers due to lattice mismatch exist at high density. If this layer is present, the warp cannot be reduced even if the transition layer according to the present invention is provided. Therefore, in the present invention, a polishing process using 0.1 μm diamond abrasive grains is performed to remove the 10 μm stacking fault layer from the silicon substrate interface.
 主表面と、この研磨処理によって得られた主表面に平行な裏面の積層欠陥密度を測定するために500℃溶融KOHに5分間浸漬させた。その後、欠陥が顕在化した基板に対して、光学顕微鏡観察を実施したところ表面すなわち遷移層の開始面における積層欠陥密度は7600本/cmであり、裏面、すなわち第2の面の積層欠陥密度は8000本/cmであった。また、遷移層の開始面上において反位相領域境界面は認められなかった。
ここで、面欠陥(積層欠陥)密度測定は、[-110]方向に平行な面欠陥は、これと直交する[110]方向の単位長と交差する本数(単位:本/cm)として算出し、[110]方向に平行な面欠陥は、これと直交する[-110]方向の単位長と交差する本数(単位:本/cm)として算出し、これらの平均値を面欠陥(積層欠陥)密度とした。
In order to measure the stacking fault density of the main surface and the back surface parallel to the main surface obtained by this polishing treatment, it was immersed in molten KOH at 500 ° C. for 5 minutes. Thereafter, when the substrate on which the defects were revealed was observed with an optical microscope, the stacking fault density on the front surface, that is, the start surface of the transition layer was 7600 / cm, and the stacking fault density on the back surface, that is, the second surface was The number was 8000 / cm. Further, no anti-phase region interface was observed on the start surface of the transition layer.
Here, the surface defect (stacking defect) density measurement is calculated as the number (unit: book / cm) of plane defects parallel to the [−110] direction intersecting the unit length in the [110] direction perpendicular to the [−110] direction. The surface defects parallel to the [110] direction are calculated as the number (unit: lines / cm) intersecting the unit length in the [−110] direction orthogonal to the surface defects, and the average value of these is calculated as the surface defect (stacking defect). Density.
 次いで、真空蒸着法を用いて主表面上にCr/Auの積層構造の金属膜層を設けた。それぞれの層の厚さはCr25nm、Au110nmとした。さらにその上層に2μmのポジ型レジストをスピン塗布し、フォトマスクにて正方形のパターン配列を露出するように紫外線で露光した。マスクパターンのラインの辺は基板の{110}方位に平行になるようにアライメントした。正方形パターンの1辺は2μmであり、パターン同士の間隔は9μmとした。次いで、レジストを現像して露光領域に開口部を設け、Ni電解メッキにより開口部に0.5μmの厚さのNiを形成した。その後、有機洗浄にてレジストを完全に除去し、間隔が9μm間隔で配列された1辺が2μmの正方形の箇所に厚さ0.5μmのNiパターンを得た。その後、誘導結合プラズマ反応性イオンエッチングにより垂直形状のメサ加工を行った。正方形以外の部分のクロム、金と共に炭化珪素を加工する。その後正方形の箇所に残るCr、Au、Niをウェットエッチングにより完全に除去した。さらに過酸化水素水+硫酸混合溶液(1:1)、HF溶液にて洗浄した。洗浄した後、熱処理装置を用いて熱酸化膜を約100nm形成した。形成した熱酸化膜を、希フッ酸により除去した。このような工程を経て、主表面に9μm間隔に配列された、1辺が2μm、高さ5μmの離散した立方晶炭化珪素(図6の62)が形成された。 Next, a metal film layer having a laminated structure of Cr / Au was provided on the main surface using a vacuum deposition method. The thickness of each layer was set to Cr 25 nm and Au 110 nm. Further, a positive resist of 2 μm was spin-coated on the upper layer, and exposed with ultraviolet rays so as to expose a square pattern array with a photomask. The mask pattern line side was aligned so as to be parallel to the {110} orientation of the substrate. One side of the square pattern was 2 μm, and the interval between the patterns was 9 μm. Next, the resist was developed to provide an opening in the exposed region, and Ni having a thickness of 0.5 μm was formed in the opening by Ni electroplating. Thereafter, the resist was completely removed by organic cleaning, and a Ni pattern having a thickness of 0.5 μm was obtained in a square portion with a side of 2 μm arranged at intervals of 9 μm. Thereafter, vertical mesa processing was performed by inductively coupled plasma reactive ion etching. Silicon carbide is processed together with chromium and gold other than the square. Thereafter, Cr, Au, and Ni remaining in the square portion were completely removed by wet etching. Further, it was washed with a hydrogen peroxide solution + sulfuric acid mixed solution (1: 1) and an HF solution. After cleaning, a thermal oxide film was formed to a thickness of about 100 nm using a heat treatment apparatus. The formed thermal oxide film was removed with dilute hydrofluoric acid. Through such steps, discrete cubic silicon carbide (62 in FIG. 6) arranged on the main surface at intervals of 9 μm and having a side of 2 μm and a height of 5 μm was formed.
 図6の状況へと炭化珪素を成長させた後、表3の条件で60分間の成長をおこない、5μmの立方晶炭化珪素(図7の(72)および(73))を成長させた。この際、図7に示すように離散的に選択成長した炭化珪素(72)の上部同士が連結し、主表面、すなわち第1の面の下部0.1μmの深さに幅4μm、間隔7μm、高さ9.9μmの空間(74)を形成した。 6 After growing silicon carbide to the situation of FIG. 6, growth was performed for 60 minutes under the conditions of Table 3 to grow 5 μm cubic silicon carbide ((72) and (73) of FIG. 7). At this time, as shown in FIG. 7, the upper portions of discretely grown silicon carbide (72) are connected to each other, the main surface, that is, the depth of 0.1 μm below the first surface is 4 μm wide, 7 μm apart, A space (74) having a height of 9.9 μm was formed.
 その後、炭化珪素基板を500℃溶融KOHに5分間浸漬させた。次いで、欠陥が顕在化した基板の主表面すなわち第1面に対して、光学顕微鏡観察を実施したところ第1面の積層欠陥密度は525本/cmであった。以上のように、炭化珪素基板に対して空間が設けられた層(即ち遷移層)を形成することにより、積層欠陥密度は空間形成前に比べて1桁以上低減する。 Thereafter, the silicon carbide substrate was immersed in molten KOH at 500 ° C. for 5 minutes. Next, when the main surface of the substrate, ie, the first surface, on which the defects became apparent was observed with an optical microscope, the stacking fault density on the first surface was 525 / cm. As described above, by forming a layer (that is, a transition layer) in which a space is provided with respect to the silicon carbide substrate, the stacking fault density is reduced by an order of magnitude or more compared with that before the space is formed.
 また、基板の反りの定義を図8に示す。測定方法は、キーエンス社製レーザー変位計にて基板表面を水平にスキャンし、その際の基板高さにおける最大値と最小値の差分を反りとした。
 得られた基板について、図2における各パラメータの値を表10に示す。
Moreover, the definition of the curvature of a board | substrate is shown in FIG. As a measuring method, the substrate surface was scanned horizontally with a laser displacement meter manufactured by Keyence Corporation, and the difference between the maximum value and the minimum value at the substrate height at that time was warped.
Table 10 shows the values of the parameters in FIG.
Figure JPOXMLDOC01-appb-T000011
 
Figure JPOXMLDOC01-appb-T000011
 
 tは、ミツトヨ社製マイクロメーターにて測定した。N1,N2,N3は、立方晶炭化珪素基板表面に対して溶融KOH処理(500℃×5min)を実施し、面欠陥を顕在化した後に顕微鏡にて単位長さ当たりの面欠陥数をカウントした。得られた数値から勾配を求めたところ、勾配1は勾配2の約27.9倍であり、20倍以上という条件を満たしていた。また、遷移層の厚さtは全膜厚tの約30分の1であった。また、この基板の反りは45μmであることを反り測定にて確認した。 t was measured with a micrometer manufactured by Mitutoyo Corporation. N1, N2, and N3 were subjected to a molten KOH treatment (500 ° C. × 5 min) on the surface of the cubic silicon carbide substrate, and after surface defects were revealed, the number of surface defects per unit length was counted with a microscope. . When the gradient was obtained from the obtained numerical value, the gradient 1 was about 27.9 times the gradient 2, and the condition of 20 times or more was satisfied. The transition layer thickness t T was about 1/30 of the total film thickness t. Further, it was confirmed by warpage measurement that the warpage of the substrate was 45 μm.
 (001)面の表面法線軸を[110]方向に3°傾け、(001)面の表面法線軸を[-110]方向にも2°傾けた珪素基板を用いたこと以外は実施例2と同様にして、CVD装置内にてCとHの混合雰囲気で加熱し、珪素基板上に極薄の炭化珪素層を形成した。基板の口径は6インチである。基板は1350℃まで加熱した。原料ガスであるC、およびキャリアガスであるHは、室温より基板表面に供給した。供給量と圧力は、表1に示したとおりである。 Example 2 except that a silicon substrate in which the surface normal axis of the (001) plane is inclined by 3 ° in the [110] direction and the surface normal axis of the (001) plane is also inclined by 2 ° in the [−110] direction is used. Similarly, heating was performed in a CVD apparatus in a mixed atmosphere of C 2 H 2 and H 2 to form an extremely thin silicon carbide layer on the silicon substrate. The diameter of the substrate is 6 inches. The substrate was heated to 1350 ° C. The source gas C 2 H 2 and the carrier gas H 2 were supplied to the substrate surface from room temperature. The supply amount and pressure are as shown in Table 1.
 基板表面が1350℃に到達した後は、基板を15分間、前述したCと水素雰囲気を保持した。 After the substrate surface reached 1350 ° C., the substrate was maintained in the above-described C 2 H 2 and hydrogen atmosphere for 15 minutes.
 上記方法で極薄炭化珪素層を形成させた後に、温度1350℃でSiHCl、C そしてHを供給することにより炭化珪素を成長させた。この場合の成長条件は、表2に示したとおりである。成長時の圧力は、反応室-ポンプ間に設置した圧力調整バルブにて調整した。 After forming an ultrathin silicon carbide layer by the above method, silicon carbide was grown by supplying SiH 2 Cl 2 , C 2 H 2 and H 2 at a temperature of 1350 ° C. The growth conditions in this case are as shown in Table 2. The pressure during the growth was adjusted with a pressure regulating valve installed between the reaction chamber and the pump.
 以上の条件で2時間成長をおこない、120μmの立方晶炭化珪素をSi基板上に成長させた。なお、炭化珪素の成長に用いるためのSiの原料ガスとしてはSiH、SiCl、SiHClを、Cの原料ガスとしてはCH、C、C、Cを用いても差し支えない。 Growth was performed for 2 hours under the above conditions, and 120 μm cubic silicon carbide was grown on the Si substrate. Si source gas used for the growth of silicon carbide is SiH 4 , SiCl 4 , SiHCl 3 , and C source gas is CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . It can be used.
 次いで、真空蒸着法を用いて主表面上にCr/Auの積層構造の金属膜層を設けた。それぞれの層の厚さはCr25nm、Au110nmとした。さらにその上層に2μmのポジ型レジストをスピン塗布し、フォトマスクにて正方形のパターン配列を露出するように紫外線で露光した。マスクパターンのラインの辺は基板の{110}方位に平行になるようにアライメントした。正方形パターンの1辺は2μmであり、パターン同士の間隔は9μmとした。次いで、レジストを現像して露光領域に開口部を設け、Ni電解メッキにより開口部に0.5μmの厚さのNiを形成した。その後、有機洗浄にてレジストを完全に除去し、間隔が9μm間隔で配列された1辺が2μmの正方形の箇所に厚さ0.5μmのNiパターンを得た。その後、誘導結合プラズマ反応性イオンエッチングにより垂直形状のメサ加工を行った。正方形以外の部分のクロム、金と共に炭化珪素を加工する。その後正方形の箇所に残るCr、Au、Niをウェットエッチングにより完全に除去した。さらに過酸化水素水+硫酸混合溶液(1:1)、HF溶液にて洗浄した。洗浄した後、熱処理装置を用いて熱酸化膜を約100nm形成した。形成した熱酸化膜を、希フッ酸により除去した。このような工程を経て、主表面に9μm間隔に配列された、1辺が2μm、高さ5μmの離散した立方晶炭化珪素(即ち、離散的に配列された炭化珪素(62))が形成された。 Next, a metal film layer having a laminated structure of Cr / Au was provided on the main surface using a vacuum deposition method. The thickness of each layer was set to Cr 25 nm and Au 110 nm. Further, a positive resist of 2 μm was spin-coated on the upper layer, and exposed with ultraviolet rays so as to expose a square pattern array with a photomask. The mask pattern line side was aligned so as to be parallel to the {110} orientation of the substrate. One side of the square pattern was 2 μm, and the interval between the patterns was 9 μm. Next, the resist was developed to provide an opening in the exposed region, and Ni having a thickness of 0.5 μm was formed in the opening by Ni electroplating. Thereafter, the resist was completely removed by organic cleaning, and a Ni pattern having a thickness of 0.5 μm was obtained in a square portion with a side of 2 μm arranged at intervals of 9 μm. Thereafter, vertical mesa processing was performed by inductively coupled plasma reactive ion etching. Silicon carbide is processed together with chromium and gold other than the square. Thereafter, Cr, Au, and Ni remaining in the square portion were completely removed by wet etching. Further, it was washed with a hydrogen peroxide solution + sulfuric acid mixed solution (1: 1) and an HF solution. After cleaning, a thermal oxide film was formed to a thickness of about 100 nm using a heat treatment apparatus. The formed thermal oxide film was removed with dilute hydrofluoric acid. Through these steps, discrete cubic silicon carbide (that is, discretely arranged silicon carbide (62)) having a side of 2 μm and a height of 5 μm arranged at intervals of 9 μm is formed on the main surface. It was.
 図6には、パターニング後の炭化珪素基板の断面構造を示す。その後、まず表8の条件で60分間の成長をおこない、5μmの立方晶炭化珪素(図7の(72)および(73))を成長させた。この際、図7に示すように離散的に選択成長した炭化珪素(72)の上部同士が連結し、主表面の下部0.1μmの深さに幅4μm、間隔7μm、高さ9.9μmの空間(74)が形成する。主表面、すなわち遷移層表面の積層欠陥密度を測定するために500℃溶融KOHに5分間浸漬させた。その後、欠陥が顕在化した基板に対して、光学顕微鏡観察を実施したところ主表面における積層欠陥密度は525本/cmであった。面欠陥(積層欠陥)密度測定は、[-110]方向に平行な面欠陥は、これと直交する[110]方向の単位長と交差する本数(単位:本/cm)として算出し、[110]方向に平行な面欠陥は、これと直交する[-110]方向の単位長と交差する本数(単位:本/cm)として算出し、これらの平均値を面欠陥(積層欠陥)密度とした。 FIG. 6 shows a cross-sectional structure of the silicon carbide substrate after patterning. Thereafter, growth was performed for 60 minutes under the conditions shown in Table 8 to grow 5 μm cubic silicon carbide ((72) and (73) in FIG. 7). At this time, as shown in FIG. 7, the upper parts of discretely grown silicon carbide (72) are connected to each other, the depth of the lower part of the main surface is 0.1 μm, the width is 4 μm, the interval is 7 μm, and the height is 9.9 μm. A space (74) is formed. In order to measure the stacking fault density on the main surface, that is, the transition layer surface, it was immersed in molten KOH at 500 ° C. for 5 minutes. Then, when the optical microscope observation was implemented with respect to the board | substrate with which the defect became obvious, the stacking fault density in the main surface was 525 / cm. In the surface defect (stacking defect) density measurement, the surface defect parallel to the [−110] direction is calculated as the number (unit: lines / cm) intersecting the unit length in the [110] direction perpendicular to the [110] direction. ] Are calculated as the number of units (unit: lines / cm) intersecting the unit length in the [−110] direction orthogonal to this, and the average value of these is defined as the surface defect (stacking defect) density. .
 次に、表2の条件で4時間の成長をおこない、最終的に珪素基板上に325μmの立方晶炭化珪素を成長させた。 Next, growth was performed for 4 hours under the conditions shown in Table 2, and finally 325 μm cubic silicon carbide was grown on the silicon substrate.
 成長後に珪素基板をフッ酸と硝酸の混酸でエッチングし、単独の立方晶炭化珪素基板を作製した。 After the growth, the silicon substrate was etched with a mixed acid of hydrofluoric acid and nitric acid to produce a single cubic silicon carbide substrate.
 次に、0.1μmダイヤモンド砥粒を用いた研磨処理を施し、珪素基板と炭化珪素との界面から25umの積層欠陥層を除去する。主表面と、この研磨処理によって得られた主表面に平行な裏面の積層欠陥密度を測定するために500℃溶融KOHに5分間浸漬させた。その後、欠陥が顕在化した基板に対して、光学顕微鏡観察を実施したところ表面(すなわち第1の面)における積層欠陥密度は500本/cmであり、裏面(すなわち第2の面)の積層欠陥密度は8000本/cmであった。また、遷移層の開始面上において反位相領域境界面は認められなかった。
 得られた基板について、図8における各パラメータの値を表11に示す。当該基板は、図9に示す様に遷移層(t領域)が第2の面側の領域となる。
Next, a polishing process using 0.1 μm diamond abrasive grains is performed to remove the 25 μm stacking fault layer from the interface between the silicon substrate and silicon carbide. In order to measure the stacking fault density of the main surface and the back surface parallel to the main surface obtained by this polishing treatment, it was immersed in molten KOH at 500 ° C. for 5 minutes. Thereafter, when an optical microscope observation was performed on the substrate in which the defects became apparent, the stacking fault density on the front surface (namely, the first surface) was 500 / cm2, and the stacking fault on the back surface (namely, the second surface). The density was 8000 pieces / cm. Further, no anti-phase region interface was observed on the start surface of the transition layer.
Table 11 shows the values of the parameters in FIG. 8 for the obtained substrate. The substrate, the transition layer as shown in FIG. 9 (t T region) becomes the second surface area.
Figure JPOXMLDOC01-appb-T000012
 
Figure JPOXMLDOC01-appb-T000012
 
 全膜厚tと遷移層厚さtは、ミツトヨ社製マイクロメーターにて測定した。N1,N2,N3は、立方晶炭化珪素基板表面に対して溶融KOH処理(500℃、5min)を実施し、面欠陥を顕在化した後に顕微鏡にて単位長さあたりの面欠陥数をカウントした。得られた数値から勾配を求めたところ、遷移層内における面欠陥密度勾配(勾配1)は遷移層領域外の面欠陥密度勾配(勾配2)の約30倍であり、20倍以上という条件を満たしていた。そして、遷移層以外の領域の面欠陥分布、すなわち、N3はN1±5%の範囲内であった。この基板の反りは50μmであることを反り測定にて確認した。
 すなわち、遷移層が第2の面側に位置していても、勾配1が勾配2の20倍以上であれば、反りが小さい基板が得られることを確認した。なお、本実施例において、遷移層の厚さが基板全体の20分の1以下を満たしていた。
Total thickness t and the transition layer thickness t T was measured by Mitutoyo micrometer. N1, N2, and N3 were subjected to a molten KOH treatment (500 ° C., 5 min) on the surface of the cubic silicon carbide substrate, and after revealing surface defects, the number of surface defects per unit length was counted with a microscope. . When the gradient is obtained from the obtained numerical value, the surface defect density gradient (gradient 1) in the transition layer is about 30 times the surface defect density gradient (gradient 2) outside the transition layer region, and the condition of 20 times or more is satisfied. I met. The surface defect distribution in the region other than the transition layer, that is, N3 was within the range of N1 ± 5%. It was confirmed by warpage measurement that the warpage of the substrate was 50 μm.
That is, even when the transition layer is located on the second surface side, it was confirmed that a substrate with small warpage can be obtained if the gradient 1 is 20 times or more the gradient 2. In this example, the thickness of the transition layer satisfied 1/20 or less of the entire substrate.
 11 炭化珪素結晶
 12 積層欠陥
 13 積層欠陥のSi極性面
 14 積層欠陥のC極性面
 51 炭化珪素基板
 52 主表面
 53 空間
 54 空間を形成する部分の側壁と積層欠陥の内角
 55 積層欠陥
 61 炭化珪素層
 62 離散的に配列された炭化珪素
 71 炭化珪素層
 72 離散的に選択成長した炭化珪素
 73 離散的に形成された炭化珪素上に成長した炭化珪素層
 74 空間
DESCRIPTION OF SYMBOLS 11 Silicon carbide crystal 12 Stacking fault 13 Si polar face of stacking fault 14 C polar face of stacking fault 51 Silicon carbide substrate 52 Main surface 53 Space 54 Side wall of space forming portion and inner angle of stacking fault 55 Stacking fault 61 Silicon carbide layer 62 Discretely arranged silicon carbide 71 Silicon carbide layer 72 Discrete and selectively grown silicon carbide 73 Silicon carbide layer grown on discretely formed silicon carbide 74 Space

Claims (6)

  1.  主表面である第1の面と、前記第1の面の裏面である第2の面の、互いに平行な二平面を有する炭化珪素基板において、
     前記第1の面および第2の面には面欠陥が露出し、かつ、
     前記第1の面に露出する面欠陥密度N1は、前記第2の面に露出する面欠陥密度N2より小さく、かつ、
     前記炭化珪素基板は、前記第1の面に略垂直方向に面欠陥密度勾配を有する前記第1の面に略平行な1以上の遷移層を有し、
     前記遷移層は、内部に空間を備え、かつ、
     前記第1の面と前記第2の面との距離をtとすると、前記遷移層における面欠陥密度勾配が20×(N2-N1)/t以上であることを特徴とする炭化珪素基板。
    In a silicon carbide substrate having two planes parallel to each other of a first surface that is a main surface and a second surface that is a back surface of the first surface,
    A surface defect is exposed on the first surface and the second surface, and
    The surface defect density N1 exposed on the first surface is smaller than the surface defect density N2 exposed on the second surface, and
    The silicon carbide substrate has one or more transition layers substantially parallel to the first surface having a surface defect density gradient in a direction substantially perpendicular to the first surface;
    The transition layer has a space inside, and
    A silicon carbide substrate, wherein a surface defect density gradient in the transition layer is 20 × (N2-N1) / t or more, where t is a distance between the first surface and the second surface.
  2.  前記遷移層の内部において、前記空間を形成する側壁が、前記第1の面に略平行な方向に分布していることを特徴とする請求項1に記載の炭化珪素基板。 2. The silicon carbide substrate according to claim 1, wherein in the transition layer, side walls forming the space are distributed in a direction substantially parallel to the first surface.
  3.  前記炭化珪素基板中に存在する面欠陥が、主として積層欠陥であることを特徴とする請求項1又は2に記載の炭化珪素基板。 3. The silicon carbide substrate according to claim 1, wherein the surface defects present in the silicon carbide substrate are mainly stacking faults.
  4.  前記遷移層の厚さが、前記第1の面と前記第2の面との距離tの20分の1以下であることを特徴とする請求項1~3のいずれかに記載の炭化珪素基板。 The silicon carbide substrate according to any one of claims 1 to 3, wherein a thickness of the transition layer is equal to or less than 1/20 of a distance t between the first surface and the second surface. .
  5.  前記炭化珪素基板の結晶系が立方晶であり、前記第1の面が{100}面に略平行であることを特徴とする請求項1~4のいずれかに記載の炭化珪素基板。 The silicon carbide substrate according to any one of claims 1 to 4, wherein a crystal system of the silicon carbide substrate is a cubic system, and the first surface is substantially parallel to a {100} plane.
  6.  前記炭化珪素基板の結晶系が立方晶であり、前記第1の面が{111}面に略平行であることを特徴とする請求項1~4のいずれかに記載の炭化珪素基板。 The silicon carbide substrate according to any one of claims 1 to 4, wherein a crystal system of the silicon carbide substrate is a cubic system, and the first surface is substantially parallel to a {111} plane.
PCT/JP2012/062507 2011-05-18 2012-05-16 Silicon carbide substrate WO2012157670A1 (en)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
JP2011111842 2011-05-18
JP2011-111842 2011-05-18
JP2011172570 2011-08-08
JP2011-172570 2011-08-08

Publications (1)

Publication Number Publication Date
WO2012157670A1 true WO2012157670A1 (en) 2012-11-22

Family

ID=47176987

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2012/062507 WO2012157670A1 (en) 2011-05-18 2012-05-16 Silicon carbide substrate

Country Status (1)

Country Link
WO (1) WO2012157670A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113035709A (en) * 2021-03-01 2021-06-25 同辉电子科技股份有限公司 Method for improving interface characteristics of SiC device

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002217116A (en) * 2001-01-18 2002-08-02 Sony Corp Method for manufacturing crystal film, crystal substrate, and semiconductor device
JP2002217115A (en) * 2001-01-18 2002-08-02 Sony Corp Crystal film, crystal substrate, and semiconductor device
JP2004200234A (en) * 2002-12-16 2004-07-15 Hoya Advanced Semiconductor Technologies Co Ltd Semiconductor and semiconductor substrate, its manufacturing method and semiconductor element
JP2008027969A (en) * 2006-07-18 2008-02-07 Fuji Electric Holdings Co Ltd Method of manufacturing single crystal wafer
JP2011084435A (en) * 2009-10-15 2011-04-28 Hoya Corp Compound single crystal and method for producing the same
WO2012067079A1 (en) * 2010-11-15 2012-05-24 Hoya株式会社 Silicon carbide substrate and semiconductor element

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002217116A (en) * 2001-01-18 2002-08-02 Sony Corp Method for manufacturing crystal film, crystal substrate, and semiconductor device
JP2002217115A (en) * 2001-01-18 2002-08-02 Sony Corp Crystal film, crystal substrate, and semiconductor device
JP2004200234A (en) * 2002-12-16 2004-07-15 Hoya Advanced Semiconductor Technologies Co Ltd Semiconductor and semiconductor substrate, its manufacturing method and semiconductor element
JP2008027969A (en) * 2006-07-18 2008-02-07 Fuji Electric Holdings Co Ltd Method of manufacturing single crystal wafer
JP2011084435A (en) * 2009-10-15 2011-04-28 Hoya Corp Compound single crystal and method for producing the same
WO2012067079A1 (en) * 2010-11-15 2012-05-24 Hoya株式会社 Silicon carbide substrate and semiconductor element
WO2012067105A1 (en) * 2010-11-15 2012-05-24 Hoya株式会社 Silicon carbide substrate, and method for producing semi-conductor element and silicon carbide substrate

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
HIROYUKI NAGASAWA ET AL.: "Heteroepitaxial Growth and Characteristics of 3C-SiC on Large- Diameter Si(001) Substrates", MATERIALS SCIENCE FORUM, vol. 389-393, 2002, pages 319 - 322 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113035709A (en) * 2021-03-01 2021-06-25 同辉电子科技股份有限公司 Method for improving interface characteristics of SiC device
CN113035709B (en) * 2021-03-01 2022-11-08 同辉电子科技股份有限公司 Method for improving interface characteristics of SiC device

Similar Documents

Publication Publication Date Title
JP5945505B2 (en) Silicon carbide substrate, semiconductor element, and method for manufacturing silicon carbide substrate
JP5218047B2 (en) Method for producing gallium nitride crystal and gallium nitride wafer
US8221546B2 (en) Epitaxial growth on low degree off-axis SiC substrates and semiconductor devices made thereby
US8847236B2 (en) Semiconductor substrate and semiconductor substrate manufacturing method
JP5345499B2 (en) Compound single crystal and method for producing the same
JP2011121847A5 (en)
JP2008004888A (en) Manufacturing method for silicon carbide semiconductor epitaxial substrate
KR20080056347A (en) Nitride semiconductor substrate and manufacturing method thereof
TW201006973A (en) Nitride semiconductor layer-containing structure, nitride semiconductor layer-containing composite substrate and production methods of these
TW201133555A (en) A method for reducing internal mechanical stresses in a semiconductor structure and a low mechanical stress semiconductor structure
US9758902B2 (en) Method for producing 3C-SiC epitaxial layer, 3C-SiC epitaxial substrate, and semiconductor device
JPWO2016140051A1 (en) SiC epitaxial wafer and method for producing SiC epitaxial wafer
JP5996406B2 (en) Method for manufacturing silicon carbide epitaxial wafer
JP4628189B2 (en) Method for producing silicon carbide single crystal
JP5545310B2 (en) Silicon carbide epitaxial wafer manufacturing method, silicon carbide epitaxial wafer, and silicon carbide semiconductor device
WO2012157670A1 (en) Silicon carbide substrate
WO2016013259A1 (en) Gallium nitride substrate
JP2010118616A (en) Method of manufacturing semiconductor substrate, and semiconductor substrate
JP2013035731A (en) Manufacturing method for single crystal silicon carbide film and manufacturing method for substrate with single crystal silicon carbide film
US20140290566A1 (en) Process of Surface Treatment for Wafer
JPH11181567A (en) Production of silicon carbide
JP2019009400A (en) Silicon carbide epitaxial wafer and method for manufacturing the same
JP6796407B2 (en) Manufacturing method of SiC epitaxial wafer
WO2013150587A1 (en) METHOD FOR MANUFACTURING MONOCRYSTALLINE SiC EPITAXIAL SUBSTRATE, AND MONOCRYSTALLINE SiC EPITAXIAL SUBSTRATE
JP5363370B2 (en) Method for producing highly oriented diamond film

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 12785154

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 12785154

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: JP