Seed material for liquid phase epitaxial growth of monocrystalline silicon carbide, and method for liquid phase epitaxial growth of monocrystalline silicon carbide

Abstract

Provided is an inexpensive seed material for liquid phase epitaxial growth of silicon carbide. A seed material 12 for liquid phase epitaxial growth of a monocrystalline silicon carbide includes a surface layer containing a polycrystalline silicon carbide with a 3C crystal polymorph. Upon X-ray diffraction of the surface layer thereof, a first-order diffraction peak corresponding to a (111) crystal plane is observed as a diffraction peak corresponding to the polycrystalline silicon carbide with a 3C crystal polymorph but no other first-order diffraction peak having a diffraction intensity of 10% or more of the diffraction intensity of the first-order diffraction peak corresponding to the (111) crystal plane is observed.

Claims

1. A seed material for liquid phase epitaxial growth of a monocrystalline silicon carbide, the seed material being used in a method for liquid phase epitaxial growth of a monocrystalline silicon carbide and including a surface layer containing a polycrystalline silicon carbide with a 3C crystal polymorph, wherein upon X-ray diffraction of the surface layer, a first-order diffraction peak corresponding to a (111) crystal plane is observed as a diffraction peak corresponding to the polycrystalline silicon carbide with a 3C crystal polymorph but no other first-order diffraction peak having a diffraction intensity of 10% or more of the diffraction intensity of the first-order diffraction peak corresponding to the (111) crystal plane is observed; an LO peak derived from a polycrystalline silicon carbide with a 3C crystal polymorph is observed upon Raman spectroscopic analysis of the surface layer with an excitation wavelength of 532 nm and the absolute amount of shift of the LO peak from 972 cm.sup.1 is 4 cm.sup.1 or more; the surface layer is substantially made of the polycrystalline silicon carbide with a 3C crystal polymorph; and a full width at half-maximum of the LO peak is 15 cm.sup.1 or less.

2. The seed material for liquid phase epitaxial growth of a monocrystalline silicon carbide according to claim 1, wherein upon X-ray diffraction of the surface layer at least one first-order diffraction peak is observed, each first-order diffraction peak corresponding to one of a (111) crystal plane, a (200) crystal plane, a (220) crystal plane, and a (311) crystal plane, and the average crystallite diameter calculated from the at least one first-order diffraction peak is more than 700 .

3. The seed material for liquid phase epitaxial growth of a monocrystalline silicon carbide according to claim 1, wherein the proportion of (111) crystal planes having an orientation angle of 67.5 or more in the (111) crystal planes observed by X-ray diffraction of the surface layer is 80% or more.

4. The seed material for liquid phase epitaxial growth of a monocrystalline silicon carbide according to claim 1, wherein the amount of shift of the LO peak from 972 cm.sup.1 is 4 cm.sup.1 or more.

5. The seed material for liquid phase epitaxial growth of a monocrystalline silicon carbide according to claim 1, the seed material including: a support member; and a polycrystalline silicon carbide film formed on the support member and forming the surface layer.

6. The seed material for liquid phase epitaxial growth of a monocrystalline silicon carbide according to claim 5, wherein the polycrystalline silicon carbide film has a thickness within a range of 30 m to 800 m.

7. The seed material for liquid phase epitaxial growth of a monocrystalline silicon carbide according to claim 1, the seed material being formed of a polycrystalline silicon carbide material containing a polycrystalline silicon carbide with a 3C crystal polymorph.

8. A method for liquid phase epitaxial growth of a monocrystalline silicon carbide, wherein the seed material for liquid phase epitaxial growth of a monocrystalline silicon carbide according to claim 1 and a feed material including a surface layer made of silicon carbide are heated with the surface layers of the seed material and the feed material facing each other through a silicon melt layer to epitaxially grow a monocrystalline silicon carbide on the surface layer of the seed material.

Description

BRIEF DESCRIPTION OF DRAWINGS

(1) FIG. 1 is a schematic view for illustrating a method for epitaxial growth of a monocrystalline silicon carbide in one embodiment of the present invention.

(2) FIG. 2 is a schematic cross-sectional view of a feed substrate in the one embodiment of the present invention.

(3) FIG. 3 is a schematic cross-sectional view of a seed substrate in the one embodiment of the present invention.

(4) FIG. 4 is a schematic cross-sectional view of a feed substrate in a modification.

(5) FIG. 5 is a schematic cross-sectional view of a seed substrate in the modification.

(6) FIG. 6 is an X-ray diffraction chart relating to Samples 1 to 4.

(7) FIG. 7 is a schematic diagram for illustrating a method for measuring the orientation of a (111) crystal plane.

(8) FIG. 8 is a graph showing the orientation of the (111) crystal plane in Sample 1.

(9) FIG. 9 is a graph showing the orientation of the (111) crystal plane in Sample 2.

(10) FIG. 10 is a graph showing the orientation of the (111) crystal plane in Sample 3.

(11) FIG. 11 is a graph showing the orientation of the (111) crystal plane in Sample 4.

(12) FIG. 12 is a graph showing the amounts () of shift of LO peaks from 972 cm.sup.1 in Samples 1 to 4 and the full widths at half-maximum (FWHM) of the LO peaks.

(13) FIG. 13 is a graph showing the rates of growth of monocrystalline silicon carbide epitaxial growth films in Samples 1, 2, and 4.

(14) FIG. 14 is a graph showing the rates of growth of monocrystalline silicon carbide epitaxial growth films in Samples 1, 2, and 4.

(15) FIG. 15 is a graph showing the rates of growth of monocrystalline silicon carbide epitaxial growth films in Samples 3 and 4.

(16) FIG. 16 is a SEM photograph of a seed substrate (Sample 3) after being subjected to an experiment of liquid phase epitaxial growth in an example.

(17) FIG. 17 is a SEM photograph of a seed substrate (Sample 2) after being subjected to an experiment of liquid phase epitaxial growth in a comparative example.

DESCRIPTION OF EMBODIMENTS

(18) A description will be given below of an example of a preferred embodiment for working of the present invention. However, the following embodiment is simply illustrative. The present invention is not at all limited by the following embodiment.

(19) FIG. 1 is a schematic view for illustrating a method for epitaxial growth of a monocrystalline silicon carbide in this embodiment.

(20) The embodiment describes an example in which an epitaxial growth film of a monocrystalline silicon carbide is formed using MSE.

(21) In this embodiment, as shown in FIG. 1, a seed substrate 12 as the seed material and a feed substrate 11 as the feed material are arranged in a vessel 10 so that a principal surface 12a of the seed substrate 12 and a principal surface 11a of the feed substrate 11 face each other with a silicon plate therebetween. In this position, the seed substrate 12 and the feed substrate 11 are heated to melt the silicon plate. Thus, the seed substrate 12 and the feed substrate 11 are faced each other with a silicon melt layer 13 therebetween. By keeping this state, source materials including silicon, carbon, and silicon carbide are eluted from the seed substrate 12 into the silicon melt layer 13. Thus, a concentration gradient is formed in the silicon melt layer 13. As a result, a monocrystalline silicon carbide epitaxially grows on the principal surface 12a of the seed substrate 12, resulting in the formation of a monocrystalline silicon carbide epitaxial growth film 20. The thickness of the silicon melt layer 13 is extremely small and can be, for example, about 10 m to about 100 m.

(22) (Seed Substrate 12)

(23) The seed substrate 12 includes a surface layer containing a polycrystalline silicon carbide with a 3C crystal polymorph. Specifically, in this embodiment, as shown in FIG. 3, the seed substrate 12 includes a graphite-made support member 12b and a polycrystalline silicon carbide film 12c. The graphite-made support member 12b has high thermal resistance so that it can sufficiently withstand the process for epitaxial growth of silicon carbide. Furthermore, the graphite-made support member 12b has a coefficient of thermal expansion similar to that of the monocrystalline silicon carbide epitaxial growth film 20. Therefore, with the use of the graphite-made support member 12b, the monocrystalline silicon carbide epitaxial growth film 20 can be suitably formed.

(24) Specific examples of graphite include natural graphite, artificial graphite, petroleum coke, coal coke, pitch coke, carbon black, and mesocarbon. An example of a method for producing the graphite-made support member 12b is a production method described in JP-A-2005-132711.

(25) The polycrystalline silicon carbide film 12c is formed to cover the principal and side surfaces of the support member 12b. The polycrystalline silicon carbide film 12c contains a polycrystalline silicon carbide. The surface layer of the seed substrate 12 is formed of the polycrystalline silicon carbide film 12c. The polycrystalline silicon carbide film 12c in this embodiment preferably contains polycrystalline 3CSiC as a major ingredient and is preferably substantially made of polycrystalline 3CSiC. In other words, in this embodiment, the surface layer of the seed substrate 12 preferably contains polycrystalline 3CSiC as a major ingredient and is preferably substantially made of polycrystalline 3CSiC. Thus, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 can be increased.

(26) The thickness t12 of the polycrystalline silicon carbide film 12c is preferably within a range of 30 m to 800 m, more preferably within a range of 40 m to 600 m, and still more preferably within a range of 100 m to 300 m. If the thickness t12 of the polycrystalline silicon carbide film 12c is too small, the graphite-made support member 12b may be exposed during the formation of a monocrystalline silicon carbide epitaxial growth film 20, thus causing elution from the support member 12b and resulting in failure to obtain a suitable monocrystalline silicon carbide epitaxial growth film 20. On the other hand, if the thickness t12 of the polycrystalline silicon carbide film 12c is too large, the polycrystalline silicon carbide film 12c may be likely to produce cracks.

(27) No particular limitation is placed on the method for forming the polycrystalline silicon carbide film 12c. The polycrystalline silicon carbide film 12c can be formed, for example, by CVD (chemical vapor deposition) or sputtering. Particularly in this embodiment, since the polycrystalline silicon carbide film 12c contains polycrystalline 3CSiC, a dense polycrystalline silicon carbide film 12c can be formed with ease at low cost by CVD.

(28) The polycrystalline silicon carbide film 12c forming the surface layer of the seed substrate 12 is such that when it is subjected to X-ray diffraction, a first-order diffraction peak corresponding to the (111) crystal plane is observed as a diffraction peak corresponding to a crystal polymorph of polycrystalline 3CSiC but no other first-order diffraction peak having a diffraction intensity of 10% or more of the diffraction intensity of the first-order diffraction peak corresponding to the (111) crystal plane is observed.

(29) Furthermore, when the polycrystalline silicon carbide film 12c forming the surface layer of the seed substrate 12 is subjected to Raman spectroscopic analysis with an excitation wavelength of 532 nm, an LO peak derived from a polycrystalline silicon carbide with a 3C crystal polymorph is observed. The absolute amount of shift of the LO peak from 972 cm.sup.1 is 4 cm.sup.1 or more.

(30) As described above, it has been heretofore considered that the feed material and the seed material should be selected so that the feed material has a higher free energy than the seed material and that if a 3CSiC substrate having a low free energy is used as the feed substrate, a 3CSiC substrate cannot be used as the seed substrate and a silicon carbide substrate having a different crystal polymorph, such as a 4HSiC substrate or a 6HSiC substrate, should be used as the seed substrate.

(31) However, the inventors have found through intensive studies that, among 3CSiC substrates, there are substrates likely to be eluted into a silicon melt layer and substrates less likely to be eluted into a silicon melt layer and that the 3CSiC substrates less likely to be eluted into a silicon melt layer can be suitably used as the seed material for liquid phase epitaxial growth of silicon carbide. Furthermore, the inventors have found through further intensive studies that if upon X-ray diffraction no other first-order diffraction peak having a diffraction intensity of 10% or more of the diffraction intensity of the first-order diffraction peak corresponding to the (111) crystal plane is observed as a diffraction peak corresponding to a crystal polymorph of polycrystalline 3CSiC, the elution into the silicon melt layer becomes less likely to occur. Based on these findings, the inventors used as the seed substrate 12 a substrate including a surface layer in which upon X-ray diffraction thereof a first-order diffraction peak corresponding to a (111) crystal plane was observed as a diffraction peak corresponding to a crystal polymorph of polycrystalline 3CSiC but no other first-order diffraction peak having a diffraction intensity of 10% or more of the diffraction intensity of the first-order diffraction peak corresponding to the (111) crystal plane was observed. Thus, even with the use of the seed substrate 12 having a surface layer containing polycrystalline 3CSiC and producible at low cost by CVD, silicon carbide could be suitably epitaxially grown. The reason why the elution into the silicon melt layer becomes less likely to occur when no other first-order diffraction peak having a diffraction intensity of 10% or more of the diffraction intensity of the first-order diffraction peak corresponding to the (111) crystal plane is observed can be that the (111) crystal plane is a crystal plane less likely to be eluted into the silicon melt layer than the other crystal planes.

(32) Examples of diffraction peaks corresponding to a crystal polymorph of polycrystalline 3CSiC include the peaks shown in Table 1 below.

(33) TABLE-US-00001 TABLE 1 Corresponding Crystal Plane 2 () (111) 35.6 (200) 41.4 (220) 60.0 (311) 71.7

(34) Furthermore, with the use of the seed substrate 12 of this embodiment, an epitaxial growth film of a hexagonal monocrystalline silicon carbide having superior properties can be formed. The reason for this can be that because the (111) crystal plane is equivalent to the (0001) hexagonal crystal plane, a large degree of exposure of the (111) crystal plane suitably promotes the epitaxial growth of a hexagonal monocrystalline silicon carbide.

(35) Typical examples of the hexagonal monocrystalline silicon carbide include a monocrystalline silicon carbide with a 4H crystal polymorph and a monocrystalline silicon carbide with a 6H crystal polymorph. These monocrystalline silicon carbides with a 4H or 6H crystal polymorph (4HSiC and 6HSiC) have the advantage of enabling semiconductor devices having a larger band gap and higher thermal resistance than other silicon carbides with different crystal polymorphs.

(36) Furthermore, the polycrystalline silicon carbide film 12c is preferably such that the average crystallite diameter calculated from the first-order diffraction peak observed by X-ray diffraction and corresponding to the polycrystalline silicon carbide with a 3C crystal polymorph is more than 700 A. In this case, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 can be still further increased. The reason for this can be that the polycrystalline silicon carbide film 12c is decreased in proportion of high-reactivity grain boundaries of polycrystalline silicon carbide crystals, which makes it still less likely to cause the elution of the polycrystalline silicon carbide film 12c into the silicon melt layer.

(37) Moreover, the polycrystalline silicon carbide film 12c is preferably such that the proportion of (111) crystal planes having an orientation angle of 67.5 or more in the (111) crystal planes observed by X-ray diffraction is 80% or more. In this case, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 can be still more effectively increased. The reason for this can be that the degree of exposure of crystal planes having lower stability than the (111) crystal plane in the crystals having (111) crystal planes exposed can be reduced, which makes it possible to reduce the elution of the crystals having (111) crystal planes exposed.

(38) Moreover, in the polycrystalline silicon carbide film 12c, an LO peak derived from a polycrystalline silicon carbide with a 3C crystal polymorph is observed upon Raman spectroscopic analysis with an excitation wavelength of 532 nm and the absolute amount of shift of the LO peak from 972 cm.sup.1 is 4 cm.sup.1 or more. Therefore, the elution into the silicon melt layer is still less likely to occur, so that the liquid phase epitaxial growth of the film can be more suitably performed. The reason why the elution into the silicon melt layer is less likely to occur when the absolute amount of shift of the LO peak from 972 cm.sup.1 is 4 cm.sup.1 or more can be that the internal stress of the surface layer of the seed substrate 12 is increased to increase the denseness of the surface layer.

(39) Furthermore, it can be considered that when the absolute amount of shift of the LO peak from 972 cm.sup.1 is 4 cm.sup.1 or more, the denseness of the polycrystalline silicon carbide film 12c is increased, so that a majority of the crystal planes exposed on the surface of the polycrystalline silicon carbide film 12c have a shape similar to the (0001) hexagonal crystal plane to thereby more suitably promote the epitaxial growth of a hexagonal monocrystalline silicon carbide.

(40) In this embodiment, the amount of shift of the LO peak from 972 cm.sup.1 is preferably 4 cm.sup.1 or more.

(41) Furthermore, the full width at half-maximum of the LO peak is preferably 15 cm.sup.1 or less. In this case, the rate of epitaxial growth of the monocrystalline silicon carbide can be still further increased.

(42) The reason why the rate of epitaxial growth of the monocrystalline silicon carbide can be further increased when the full width at half-maximum of the LO peak is 15 cm.sup.1 or less can be that smaller full widths at half-maximum of the LO peak can be caused by higher degrees of crystallinity or lower impurity concentrations of the polycrystalline silicon carbide in the surface layer of the seed substrate 12, which makes it still less likely to cause the elution from the surface layer of the seed substrate 12.

(43) (Feed Substrate 11)

(44) In this embodiment, no particular limitation is placed on the material of the feed substrate 11 so long as it is less likely to cause the elution into the silicon melt layer 13 than the seed substrate 12. Therefore, an example of the feed substrate 11 that can be suitably used is a substrate which includes a surface layer containing a polycrystalline silicon carbide with a 3C crystal polymorph and in which upon X-ray diffraction of the surface layer a diffraction peak corresponding to the (111) crystal plane and a diffraction peak other than the diffraction peak corresponding to the (111) crystal plane are observed.

(45) Furthermore, an example of the feed substrate 11 that can be more suitably used is a substrate which includes a surface layer containing a polycrystalline silicon carbide with a 3C crystal polymorph and in which the absolute amount of shift of the LO peak observed by Raman spectroscopic analysis of the surface layer with an excitation wavelength of 532 nm from 972 cm.sup.1 is less than 4 cm.sup.1.

(46) FIG. 2 shows a schematic cross-sectional view of the feed substrate 11 in this embodiment. Specifically, in this embodiment, as shown in FIG. 2, the feed substrate 11 includes a graphite-made support member 11b and a polycrystalline silicon carbide film 11c. The graphite-made support member 11b has high thermal resistance so that it can sufficiently withstand the process for epitaxial growth of silicon carbide. Furthermore, the graphite-made support member 11b has a coefficient of thermal expansion similar to that of the monocrystalline silicon carbide epitaxial growth film 20. Therefore, with the use of the graphite-made support member 11b, the silicon carbide epitaxial growth film 20 can be suitably formed.

(47) Specific examples of graphite include natural graphite, artificial graphite, petroleum coke, coal coke, pitch coke, carbon black, and mesocarbon. An example of a method for producing the graphite-made support member 11b is a production method described in JP-A-2005-132711.

(48) The polycrystalline silicon carbide film 11c is formed to cover the principal and side surfaces of the support member 11b. The polycrystalline silicon carbide film 11c contains a polycrystalline silicon carbide. The surface layer of the feed substrate 11 is formed of the polycrystalline silicon carbide film 11c. The polycrystalline silicon carbide film 11c preferably contains polycrystalline 3CSiC as a major ingredient and is preferably substantially made of polycrystalline 3CSiC. In other words, the surface layer of the feed substrate 11 preferably contains polycrystalline 3CSiC as a major ingredient and is preferably substantially made of polycrystalline 3CSiC. Thus, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 can be increased.

(49) The thickness t11 of the polycrystalline silicon carbide film 11c is preferably within a range of 30 m to 800 m, more preferably within a range of 40 m to 600 m, and still more preferably within a range of 100 m to 300 m. If the thickness t11 of the polycrystalline silicon carbide film 11c is too small, the graphite-made support member 11b may be exposed during the formation of a monocrystalline silicon carbide epitaxial growth film 20, thus causing elution from the support member 11b and resulting in failure to obtain a suitable monocrystalline silicon carbide epitaxial growth film 20. On the other hand, if the thickness t11 of the polycrystalline silicon carbide film 11c is too large, the polycrystalline silicon carbide film 11c may be likely to produce cracks.

(50) No particular limitation is placed on the method for forming the polycrystalline silicon carbide film 11c. The polycrystalline silicon carbide film 11c can be formed, for example, by CVD (chemical vapor deposition) or sputtering. Particularly in this embodiment, since the polycrystalline silicon carbide film 11c contains polycrystalline 3CSiC, a dense polycrystalline silicon carbide film 11c can be formed with ease at low cost by CVD.

(51) The polycrystalline silicon carbide film 11c forming the surface layer of the feed substrate 11 is such that when it is subjected to X-ray diffraction, a diffraction peak corresponding to the (111) crystal plane and a diffraction peak other than the diffraction peak corresponding to the (111) crystal plane are observed as diffraction peaks corresponding to a crystal polymorph of polycrystalline 3CSiC. More specifically, the polycrystalline silicon carbide film 11c is such that when it is subjected to X-ray diffraction, a diffraction peak corresponding to the (111) crystal plane and at least one diffraction peak each corresponding to one of the (200) crystal plane, the (220) crystal plane, and the (311) crystal plane are observed as diffraction peaks corresponding to a crystal polymorph of polycrystalline 3CSiC. Therefore, in this embodiment, the monocrystalline silicon carbide epitaxial growth film 20 can be formed at a high rate of growth. A possible reason for this is that the crystal planes other than the (111) crystal plane are more likely to cause the elution into the silicon melt layer 13 than the (111) crystal plane.

(52) The polycrystalline silicon carbide film 11c is such that when subjected to X-ray diffraction, in addition to a diffraction peak corresponding to the (111) crystal plane, at least one diffraction peak each corresponding to one of the (200) crystal plane, the (220) crystal plane, and the (311) crystal plane is preferably observed as a diffraction peak corresponding to a crystal polymorph of polycrystalline 3CSiC and, more preferably, diffraction peaks each corresponding to one of the (200) crystal plane, the (220) crystal plane, and the (311) crystal plane are observed. In this case, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 can be further increased. The reason for this can be that, among the crystal planes other than the (111) crystal plane, the (200) crystal plane, the (220) crystal plane, and the (311) crystal plane have particularly high reactivity and thus are more likely to cause the elution into the silicon melt layer 13.

(53) Moreover, the sum of the intensities of first-order diffraction peaks other than the first-order diffraction peak corresponding to the (111) crystal plane is preferably 10% or more of the sum of the intensities of all the first-order diffraction peaks and more preferably 20% or more thereof. In this case, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 can be still further increased.

(54) Among a plurality of first-order diffraction peaks observed as diffraction peaks corresponding to a crystal polymorph of polycrystalline 3CSiC when the polycrystalline silicon carbide film 11c is subjected to X-ray diffraction, a first-order diffraction peak corresponding to the (111) crystal plane is preferably a main diffraction peak having the highest diffraction intensity.

(55) Furthermore, the polycrystalline silicon carbide film 11c is preferably such that the average crystallite diameter calculated from the first-order diffraction peak observed by X-ray diffraction and corresponding to the polycrystalline silicon carbide with a 3C crystal polymorph is 700 A or less. In this case, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 can be still further increased. The reason for this can be that the polycrystalline silicon carbide film 11c is increased in proportion of high-reactivity grain boundaries of polycrystalline silicon carbide crystals, which makes it more likely to cause the elution from the polycrystalline silicon carbide film 11c.

(56) Furthermore, the polycrystalline silicon carbide film 11c is preferably such that when subjected to X-ray diffraction, a first-order diffraction peak corresponding to the (111) crystal plane and at least one first-order diffraction peak each corresponding to one of the (200) crystal plane, the (220) crystal plane, and the (311) crystal plane are observed and (I.sub.1/I.sub.0).sup.1.Math.D.sup.2 is equal to or smaller than 10.sup.8,

(57) where:

(58) I.sub.0 represents the sum of the intensity of the first-order diffraction peak corresponding to the (111) crystal plane and the total intensity of the at least one first-order diffraction peak each corresponding to one of the (200) crystal plane, the (220) crystal plane, and the (311) crystal plane;

(59) I.sub.1 represents the total intensity of the at least one first-order diffraction peak each corresponding to one of the (200) crystal plane, the (220) crystal plane, and the (311) crystal plane; and

(60) D represents the average crystallite diameter calculated using the Hall's equation from the at least one first-order diffraction peak each corresponding to one of the (200) crystal plane, the (220) crystal plane, and the (311) crystal plane.

(61) In this case, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 can be still more effectively increased. The reason for this can be that the polycrystalline silicon carbide film 11c is increased in proportion of (200), (220), and (311) crystal planes having relatively high reactivity and is decreased in average crystallite diameter.

(62) Moreover, the polycrystalline silicon carbide film 11c is preferably such that the proportion of (111) crystal planes having an orientation angle of 67.5 or more in the (111) crystal planes observed by X-ray diffraction is less than 80%. In this case, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 can be still more effectively increased. The reason for this can be that the degree of exposure of crystal planes having lower stability than the (111) crystal plane in the crystals having (111) crystal planes exposed is increased, which makes it possible to increase the reactivity of the crystals having (111) crystal planes exposed.

(63) The polycrystalline silicon carbide film 11c forming the surface layer of the feed substrate 11 is such that the absolute amount of shift of the LO peak observed by Raman spectroscopic analysis with an excitation wavelength of 532 nm from 972 cm.sup.1 is less than 4 cm.sup.1. Therefore, in this embodiment, the monocrystalline silicon carbide epitaxial growth film 20 can be formed at a high rate of growth.

(64) The above embodiment has described an example in which each of the feed substrate 11 and the seed substrate 12 is formed of a support member 11b or 12b and a polycrystalline silicon carbide film 11c or 12c. However, the present invention is not limited to the above structure. For example, as shown in FIG. 4, the feed substrate 11 may be formed of a silicon carbide substrate containing silicon carbide. Furthermore, as shown in FIG. 5, the seed substrate 12 may be formed of a polycrystalline silicon substrate containing a polycrystalline silicon carbide with a 3C crystal polymorph.

(65) The silicon carbide substrate can be produced, for example, by coating a graphite base material with a polycrystalline silicon carbide by CDV and then mechanically or chemically removing graphite. Alternatively, the silicon carbide substrate can be produced by reacting a graphite material with silicate gas to convert the graphite material into silicon carbide. Still alternatively, the silicon carbide substrate can be produced by adding a sintering aid to silicon carbide powder and sintering the silicon carbide powder at a high temperature of 1600 C. or above.

(66) The present invention will be further described below with reference to specific examples, but is not at all limited by the following specific examples.

Production Example 1

(67) A graphite material (15 mm by 15 mm by 2 mm) made of high-purity isotropic graphite having a bulk density of 1.85 g/cm.sup.3 and an ash content of 5 ppm or less was used as a base material. The base material was put into a CVD reactor and a 30 m thick polycrystalline silicon carbide coating was formed on the base material by CVD to produce Sample 1. Silicon tetrachloride and propane gas were used as source gases. The deposition was performed at an ordinary pressure and 1200 C. The deposition rate was at 30 m/h.

Production Example 2

(68) A 50 m thick polycrystalline silicon carbide coating was formed on the surface of a graphite material in the same manner as in Production Example 1 above except that the reaction temperature was at 1400 C. and the deposition rate was at 60 m/h, resulting in the production of Sample 2.

Production Example 3

(69) A 50 m thick polycrystalline silicon carbide coating was formed on the surface of a graphite material in the same manner as in Production Example 1 above except that the reaction temperature was at 1250 C., the deposition rate was at 10 m/h, and CH.sub.3SiCl.sub.3 was used instead of silicon tetrachloride, resulting in the production of Sample 3.

Production Example 4

(70) A 50 m thick polycrystalline silicon carbide coating was formed on the surface of a graphite material in the same manner as in Production Example 1 above except that dichlorosilane (SiH.sub.2Cl.sub.2) and acetylene were used instead of silicon tetrachloride and propane gas, the reaction temperature was at 1300 C., and the deposition rate was at 10 m/h, resulting in the production of Sample 4. In Sample 4, the thickness of the polycrystalline silicon carbide coating was approximately 1 mm.

(71) (X-Ray Diffraction Measurement)

(72) The surface layers of the above produced Samples 1 to 4 were subjected to X-ray diffraction. The X-ray diffraction was performed using Ultima manufactured by Rigaku Corporation. The measurement results are shown in FIG. 6.

(73) As shown in FIG. 6, in Samples 1 and 2, not only a diffraction peak (2=35.6 corresponding to the (111) crystal plane but also diffraction peaks corresponding to crystal planes other than the (111) crystal plane and having an intensity of 10% or more of the intensity of the diffraction peak corresponding to the (111) crystal plane were observed. Specifically, in Samples 1 and 2, besides a diffraction peak (2=35.6 corresponding to the (111) crystal plane, a diffraction peak) (2=41.4 corresponding to the (200) crystal plane, a diffraction peak)(2=60.0 corresponding to the (220) crystal plane, and a diffraction peak)(2=71.7 corresponding to the (311) crystal plane, all of which have an intensity of 10% or more of the intensity of the diffraction peak corresponding to the (111) crystal plane, were observed.

(74) On the other hand, in Samples 3 and 4, a first-order diffraction peak)(2=35.6 corresponding to the (111) crystal plane and a diffraction peak)(2=75.5 corresponding to the (222) crystal plane, which is a higher-order diffraction peak of the above first-order diffraction peak, were observed, but no other first-order diffraction peak having an intensity of 10% or more of the intensity of the first-order diffraction peak corresponding to the (111) crystal plane was observed.

(75) Arranged in Table 2 below are the relative intensities of the first-order diffraction peaks corresponding to various crystal planes in Samples 1 to 4, with 100 representing the intensity of the first-order diffraction peak corresponding to the (111) crystal plane.

(76) TABLE-US-00002 TABLE 2 Corresponding Relative Intensity of First-Order Diffraction Peak Crystal Plane Sample 1 Sample 2 Sample 3 Sample 4 (111) 100 100 100 100 (200) 13 40 0 0 (220) 12 51 0 0 (311) 12 47 1 0

(77) (Calculation of Average Crystallite Diameter)

(78) The average crystallite diameters of Samples 1 to 4 were calculated based on the results of the above X-ray diffraction measurement, using the Hall's equation. In the calculation, data of diffraction peaks relating to the (111) crystal plane, the (200) crystal plane, the (220) crystal plane, and the (311) crystal plane were used. The results are shown in Table 3 below.

(79) TABLE-US-00003 TABLE 3 Average Crystallite Diameter () Sample 1 350 Sample 2 455 Sample 3 1136 Sample 4 1012

(80) As seen from the results shown in Table 3 above, the average crystallite diameters of Samples 1 and 2 were less than 700 A, more specifically, less than 500 A and the average crystallite diameters of Samples 3 and 4 were more than 700 A, more specifically, more than 1000 A.

(81) (Evaluation of Orientation of (111) Crystal Plane)

(82) Next, Samples 1 to 4 were measured in terms of angle at which the diffraction peak corresponding to the (111) plane appeared while each of them was rotated as shown in FIG. 7. The results are shown in FIGS. 8 to 11. In the graphs shown in FIGS. 8 to 11, the abscissa represents the orientation angle () shown in FIG. 7. The ordinate represents the intensity.

(83) Furthermore, Table 4 below shows the proportion of the intensity integral in a region of 67.5 and higher orientation angles () to the intensity integral in an entire region of 15 to 90 orientation angles () ((intensity integral in region of 67.5 and higher orientation angles ())/(intensity integral in entire region of 15 to 90 orientation angles ()). Note that the ((intensity integral in region of 67.5 and higher orientation angles ())/(intensity integral in entire region of 15 to 90 orientation angles ()) corresponds to the proportion of (111) crystal plane having an orientation angle of 67.5 or more in the (111) crystal planes observed by X-ray diffraction.

(84) TABLE-US-00004 TABLE 4 Proportion of (111) Crystal Planes Having 67.5 C. or Higher Orientation Angle Sample 1 38.6% Sample 2 48.5% Sample 3 96.5% Sample 4 96.9%

(85) As shown in FIGS. 8 and 9 and Table 4 above, in Samples 1 and 2, a wide intensity distribution existed also in a region of orientation angles () less than 67.5 and the proportion of (111) crystal planes having an orientation angle () of 67.5 or more was less than 80%. On the other hand, in Samples 3 and 4, no wide intensity distribution existed in a region of orientation angles () less than 67.5 and the proportion of (111) crystal planes having an orientation angle () of 67.5 or more was more than 80%.

(86) (Raman Spectroscopic Analysis)

(87) The surface layers of the above produced Samples 1 to 4 were subjected to Raman spectroscopic analysis. An excitation wavelength of 532 nm was used for the Raman spectroscopic analysis. The measurement results are shown in FIG. 12.

(88) Next, from the measurement results shown in FIG. 12, the amounts () of shift of LO peaks from 972 cm.sup.1 in Samples 1 to 4 and full widths at half-maximum (FWHM) of the LO peaks were determined. The results are shown in FIG. 13.

(89) As shown in FIG. 13, Samples 3 and 4 each had an absolute value of of more than 4 cm.sup.1 and an FWHM of more than 7 cm.sup.1. On the other hand, Samples 1 and 2 each had an FWHM of more than 7 cm.sup.1 like Samples 3 and 4, but their absolute values of were less than 4 cm.sup.1.

(90) (Evaluation of Rate of Growth of Monocrystalline Silicon Carbide Liquid Phase Epitaxial Growth Film)

(91) Using Samples 1 to 4 as feed substrates, respective monocrystalline silicon carbide epitaxial growth films 20 were produced under the conditions shown below by the method for liquid phase epitaxial growth described in the above embodiment. Then, the thickness of each of the monocrystalline silicon carbide epitaxial growth films 20 was measured by observing the cross section of the monocrystalline silicon carbide epitaxial growth film 20 with an optical microscope. The rate of growth of each monocrystalline silicon carbide epitaxial growth film 20 was determined by dividing the measured thickness by the time for epitaxial growth of silicon carbide.

(92) The results are shown in FIGS. 14 and 15. In FIGS. 14 and 15, the ordinate represents the rate of growth of each monocrystalline silicon carbide epitaxial growth film 20 and the abscissa represents the reciprocal (1/L) of the thickness (L) of the silicon melt layer 13.

(93) As seen from the results shown in FIGS. 14 and 15, in the case of using Samples 1 and 2 in which when the polycrystalline silicon carbide film 11c forming the surface layer of the feed substrate 11 was subjected to X-ray diffraction, not only a diffraction peak corresponding to the (111) crystal plane but also diffraction peaks other than the diffraction peak corresponding to the (111) crystal plane were observed as diffraction peaks corresponding to a crystal polymorph of polycrystalline 3CSiC, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 was high. On the other hand, in the case of using Samples 3 and 4 in which when the polycrystalline silicon carbide film 11c forming the surface layer of the feed substrate 11 was subjected to X-ray diffraction, only diffraction peaks corresponding to the (111) crystal plane were observed as diffraction peaks corresponding to a crystal polymorph of polycrystalline 3CSiC and, besides the first-order diffraction peak corresponding to the (111) crystal plane, no first-order diffraction peak having an intensity of 10% or more of the intensity of the first-order diffraction peak corresponding to the (111) crystal plane was observed, the rate of growth of the monocrystalline silicon carbide epitaxial growth film 20 was low. This shows that Samples 3 and 4 are less likely to cause the elution into the silicon melt layer 13.

(94) (Measurement Conditions of Rate of Growth of Monocrystalline Silicon Carbide Epitaxial Growth Film 20)

(95) Seed substrate: a silicon carbide substrate with a 4H crystal polymorph,

(96) Pressure of Atmosphere: 10.sup.6 to 10.sup.4 Pa, and

(97) Temperature of Atmosphere: 1900 C.

EXAMPLE

(98) Using the above produced Sample 1 as the feed substrate 11 and the above produced Sample 3 as the seed substrate 12, an experiment of liquid phase epitaxial growth of a monocrystalline silicon carbide was conducted under the same conditions as in the above experiment for evaluation of rate of growth. Thereafter, a photograph of the surface of Sample 3 serving as the seed substrate 12 was taken with a scanning electron microscope (SEM). The SEM photograph of the surface of Sample 3 is shown in FIG. 16. The photograph shown in FIG. 16 reveals that by using as the seed substrate 12 Sample 3 in which upon X-ray diffraction of the polycrystalline silicon carbide film a first-order diffraction peak corresponding to the (111) crystal plane is observed as a first-order diffraction peak corresponding to a crystal polymorph of polycrystalline 3CSiC but no other first-order diffraction peak having a diffraction intensity of 10% or more of the diffraction intensity of the first-order diffraction peak corresponding to the (111) crystal plane is observed, a hexagonal monocrystalline silicon carbide epitaxial growth film can be obtained.

COMPARATIVE EXAMPLE

(99) Using the above produced Sample 1 as a feed substrate and the above produced Sample 2 as a seed substrate, an experiment of liquid phase epitaxial growth of a monocrystalline silicon carbide was conducted under the same conditions as in the above experiment for evaluation of rate of growth. Thereafter, a photograph of the surface of Sample 2 serving as the seed substrate was taken with a scanning electron microscope (SEM). The SEM photograph of the surface of Sample 2 is shown in FIG. 17. The photograph shown in FIG. 17 reveals that in the case of using as a seed substrate Sample 2 in which upon X-ray diffraction of a polycrystalline silicon carbide film a first-order diffraction peak corresponding to the (111) crystal plane is observed as a diffraction peak corresponding to a crystal polymorph of polycrystalline 3CSiC but no other first-order diffraction peak having a diffraction intensity of 10% or more of the diffraction intensity of the first-order diffraction peak corresponding to the (111) crystal plane is observed, epitaxial growth hardly progresses and no suitable hexagonal monocrystalline silicon carbide epitaxial growth film can be obtained.

REFERENCE SIGNS LIST

(100) 10 . . . vessel 11 . . . feed substrate 11a . . . principal surface 11b . . . support member 11c . . . polycrystalline silicon carbide film 12 . . . seed substrate 12a . . . principal surface 12b . . . support member 12b . . . polycrystalline silicon carbide film 13 . . . silicon melt layer 20 . . . monocrystalline silicon carbide epitaxial growth film