BACKGROUND OF THE INVENTION1. Field of the Invention[0001]
This invention relates to a crystal growth method of oxide, cerium oxides, promethium oxides, multi-layered structure of oxides, manufacturing method of field effect transistor, field effect transistors, manufacturing method of ferroelectrics non-volatile memory and ferroelectric non-volatile memory, which are particularly suitable for use to oxide electronics developed on silicon substrates.[0002]
2. Description of the Related Art[0003]
Silicon oxide (SiO[0004]2) films made by thermal oxidation of silicon (Si) have been exclusively used as gate insulating films of MOS-FET (metal-oxide-semiconductor FET) because of their high electric insulating ability, low interface state density, easiness to process, thermal stability, and other advantages. These SiO2films made by thermal oxidation for use as gate insulating films, however, have low specific dielectric constants (∈r˜3.8) and must be formed very thin on Si substrates. Along with the progress toward thinner gate insulating films, short channels and other requirements to meet the demand for integration, various problems arose such as dielectric break-down of gate insulating films and pinch-off of channels caused by influences from source-drain voltages (short channel effect), and gate insulating films will soon come to limit in terms of their materials. Under the circumstances, the need for new gate insulating films having high dielectric constants is being advocated as a technical subject of MOS-FET of the sub 0.1 micron generation, in addition to a further progress of lithographic technologies, needless to say (for example, (1) MTL VLSI Seminar (Massachusetts Institute of Technology)). If a gate insulating film can be made by using a material of a high dielectric constant, it will need not be so thin. Therefore, gate leakage will be prevented, and short-channel effects will be prevented as well.
On the other hand, researches on ferroelectric non-volatile memories (FeRAM) have come to be active (for example, (2) Appl. Phys. Lett., 48(1986)1439, (3) IEDM Tech. Dig., (1987)850, (4) IEEE J. Solid State Circuits, 23(1988)1171, (5) 1988 IEEE Int. Solid-State Circuits Conf. (ISSCC88), (6) Digest of Technical Papers, THAM 10.6 (1988)130, and (7) Oyo Butsuri, 62(1993)1212). Among these ferroelectric nonvolatile memories, what is considered to be closest to practical use is a ferroelectric non-volatile memory of a quasi-DRAM structure (using two-transistors and two-capacitors type memory cells, or one-transistor and one-capacitor type memory cells). This structure is advantageous in making it easier to prevent interference with the Si process because CMOS process and ferroelectric capacitor process can be separated by using an inter-layer insulating film. However, the structure of this ferroelectric non-volatile memory does not meet the use of a scaling law of a Si device. Therefore, as microminiaturization progresses, it is necessary to employ a more complex structure or use a material with a larger value of residual polarization in order to ensure a certain amount of charge storage in the capacitor. On the other hand, there are many research institutes tackling with the study of ferroelectric non-volatile memories using MFS-FETs (metal-ferroelectrics-semiconductor)-FET type memory cells, including MFMIS (metal-ferroelectrics-metal-insulator-semiconductor)-FET type memory cells, FCG (ferroelectric capacitor gate) type memory cells, and so forth), which constitute two major subjects together with those of a quasi-DRAM structure mentioned above. The latter type ferroelectric non-volatile memories match with the scaling low, and merely need a very small value of residual polarization (about ˜0.1 μC/cm[0005]2). Additionally, since they need only one transistor for storage and hence contribute to a decrease of the cell size, they are advantageous for high integration. Furthermore, since they are of a nondestructive readout type, they are more advantageous also against fatigue, which might be an essential problem of ferroelectric materials, than destructive readout type memory cells with two transistors and two capacitors, or one transistor and one capacitor, and are also available for high-speed operations. Because these excellent properties are expected, MFS-FET ferroelectric non-volatile memories are now recognized as ultimate memories ((8) Appl. Surf. Sci. 113/114(1997)656).
It is problems with their manufacturing process that prevent practical use of these MFS-FET ferroelectric non-volatile memories. It is extremely difficult to grow ferroelectric materials directly on Si substrates. Therefore, growth of buffer layers of insulating materials on Si substrates is recognized as one of most important technologies. In case of a MFIS (metal-ferroelectrics-insulator-semiconductor)-FET ferroelectric non-volatile memories which is one of MFS-FET ferroelectric non-volatile memories, gate voltage is distributed to an insulating layer as well, and this causes the drawback that the write voltage is high. To prevent it, the insulating layer must be one with a high dielectric constant. On the other hand, material properties required as a ferroelectric material used here are a low dielectric constant, appropriate value of residual polarization (typically around ˜0.1 μC/cm[0006]2, although depending upon the device design), and most seriously, good squareness ratio. Additionally, for the purpose of realizing a better interface, it is the important condition that these materials can be grown at low temperatures. Thus, choice and development of materials is required from the standpoint different from that of the one-transistor and one-capacitor type. There are a lot of research reports on MFIS-FET ferroelectric non-volatile memories. However, because of insufficient surface properties, there is almost no reports about practically usable ones including the requirement for retention (charge retaining property). On the other hand, a MFMIS structure enabling the use of an existing SiO2film by thermal oxidation as the gate insulating film is also under consideration ((9) Jpn. J. Appl. Phys., 33(1994)5207), and this is considered to be relatively close to the stage of practical use. There is also proposed an approach stepping forward from that by separately making a ferroelectric capacitor and connecting it to a polycrystalline Si gate by wiring ((10) Japanese Patent Laid-Open Publication No. hei 8250608 and (11) Japanese Patent Laid-Open Publication No. hei 9-205181). This method facilitates device isolation between a ferroelectric material and a Si transistor, and at the same time, because the design choice in areal ratio between the capacitor and the gate, sufficient polarization can be obtained with a low write voltage by reducing the relative area of the capacitor. However, it is difficult to reliably obtain a necessary squareness ratio with polycrystalline ferroelectric materials, and this method will also encounter the limit of the material of the SiO2film by thermal oxidation in progress of microminiaturization. Eventually, also the key technology for realizing MFS-FET ferroelectric non-volatile memories is just the growth of an insulating film with a high dielectric constant on a Si substrate. Moreover, in order to realize a steep interface and a low interface state density equivalent to SiO2films by thermal oxidation, it is advantageous to epitaxially grow a material good in lattice matching, and there is such an extremely high technical hurdle that a channel can be made in the Si<110>orientation having the largest mobility of Si as the substrate and it should be on a Si(001) substrate being used exclusively as the MOS-FET substrate.
On the other hand, it is greatly significant to introduce oxide materials other than SiO[0007]2into the semiconductor industry. High-temperature superconductive materials discovered in 1986 ((12) Z. Phys. B., 64, 189-193(1986)), needless to say, and oxide materials especially having perovskite or related structures have very important physical properties for semiconductor devices, such as ferroelectricity, high dielectric constant, superconductivity, colossal magnetoresistance, and so forth ((13) Mater. Sci. Eng., B41(1996)166, and (14) J. Ceram. Soc. Japan, Int. Ed., 103(1995)1088). For example, among ferroelectric materials of capacitors for ferroelectric non-volatile memories mentioned above, zirconium titanate (PZT) having a large value of spontaneous polarization and a low process temperature (for example, (15) J. Appl. Phys. 70, 382-388(1991)) and bismuth strontium tantalate (Bi2SrTa2O1((16) Nature, 374(1995)627, (17) Appl. Phys. Lett., 66(1995)221, (18) Mater. Sci. Eng., B32(1995)75, (19) Mater. Sci. Eng., B32(1995)83, (20) Appl. Phys. Lett., 67(1995)572, (21) J. Appl. Phys., 78(1995)5073, (22) Appl. Phys. Lett., 68(1996)566, (23) Appl. Phys. Lett., 68(1996)690, and (24) International Laid-Open Publication WO93/12542) are the twin greatest materials. Furthermore, including the discovery of colossal magnetoresistance materials (CMR materials) in the group of Mn oxides, which are variable in resistivity over some digits under application of a magnetic field ((25) Phys. Rev. Lett. 74(1995)5108), great interest has come to be attracted to how high potential capacities these oxide materials have ((26) Mater. Sci. Eng., B41(1996)166, and (27) J. Ceram. Soc. Japan, Int. Ed., 103(1995)1088), and technologies for making thin oxide films have been developed remarkably in these ten years or so.
If oxide materials having these very high functional physical properties can be developed on Si which is the basis of the semiconductor industry, these materials will get a high marketability. However, because of difficulties between these functional oxide materials and Si, such as mutual thermal diffusion and differences in thermal expansion coefficient, it is usually difficult to directly grow these functional oxide materials on Si.[0008]
As discussed above, almost all of these functional oxide materials have structure based on a perovskite structure. Many of them, such as yttrium-based superconductive materials having critical temperatures beyond the liquid nitrogen temperature and Bi[0009]2SrTa2O9mentioned above, have structures called layered perovskite having a very large anisotropy. In these layered perovskite structured oxides, superconductive current paths, polarization axes, etc. are limited to specific directions, and also in case of simple perovskite structured oxides, there are many in which polarization axes are limited to specific directions like that in PZT. Therefore, when they are used to make devices, it is important to specifically orient oxides, or more preferably, epitaxially grow them relative to bases, in order to draw out their maximum properties.
Ceria (cerium dioxide: CeO[0010]2) having a fluorite structure is one of candidates for materials of gate insulating films with high dielectric constants for the sub 0.1 micron generation to substitute for SiO2films by thermal oxidation because of its thermal stability, high specific dielectric constant (∈r˜26) and very good lattice matching with Si substrates (misfit: about 0.35%), and it is considered to be one of most ideal buffer layer materials for epitaxially growing perovskite-related oxides on Si substrates. Actually, researches are being made on epitaxial growth of ceria on Si substrates, but most of them are directed to CeO2(111)/Si(111) structures easy for atomic close-packed growth (for example, (28) Jpn. J. Appl. Phys. 34(1995), L688, and (29) Japanese Patent Laid-Open Publication No. hei 7-25698). However, with regard to Si(001) substrates which are most important for practical application, there is the story that, under the recognition that the epitaxial CeO2(001)/Si(001) structure was just the ideal structure, various proposal were made on multi-layered structures, devices, and so on ((30) Japanese Patent Laid-Open Publication No. hei 2-267104, (31) Japanese Patent Laid-Open Publication No. hei 6-97452, (32) Japanese Patent Laid-Open Publication No. hei 10-182292, and others), but it was not realized. Heretofore, it has been believed that CeO2(110) having an antiphase domain epitaxially grows, reflecting the dimer structure by the surface reconstructed structure of Si(001)-2×1, 1×2. Further, regarding the growth temperature, it is considered that a temperature around800° C. at which no Si02film is formed on the Si surface in a high vacuum ((33) J. Vac. Sci. Technol. A13(1995)772) is the lower limit of epitaxial temperature ((34) Jpn. J. Appl. Phys. 33(1994), 5219, (35) Appl. Phys. Lett. 56 (1990), 1332, (36) Appl. Phys. Lett. 59(1991), 3604, (37) Physica C 192 (1992)154, (38) Jpn. J. Appl. Phys. 36(1997), 5253, and (39) Japanese Patent Laid-Open Publication No. hei 9-64206). Furthermore, under the belief that the growth temperature must be lowered to obtain a steep interface, electron beam assisted epitaxial growth, for example, is under trial. However, the lower limit of epitaxial growth temperature heretofore reported is 710° C. ((40) 1998 spring symposium of Oyo Butsuri Gakkai, Presentation No. 28p-PA-1). Although there are a few reports on CeO2(001)/Si(001) structures, there is no experimental data ((41) Japanese Patent Laid-Open Publication No. hei 2-267104), no discussion is made about separation from substrate diffraction ((42) Solid State Comm. 108(1998)225), and none could prove epitaxial growth. Although there are proposals of solid solution (Ce, Zr)O2of zirconia (zirconium dioxide: ZrO2) and ceria ((43) Jpn. J. Appl. Phys. 35(1996), 5150), CeO2/(Ce, Zr)O2/Si multi-layered structure ((44) Jpn. J. Appl. Phys. 36(1997), 5253 and (45) 1998 spring symposium of Oyo Buturi Gakkai, Paper 29p-ZF-4), and CeO2/SrTiO3/Si structure ((46) Jpn. J. Appl. Phys. 30(1991)L1136), which use the solid solution as a buffer layer. However, it has not been reported at all that CeO2(001) could be epitaxially grown directly on a Si(001) substrate.
Researches are being made also on yttria (yttrium oxide: Y[0011]2O3) having a C-rare-earth structure (bixbyite) because it has material properties similar to ceria. However, it involves the same problems as ceria, and in usual, Y2O3(110) epitaxially grows on Si(001) substrates ((47) Appl. Phys. Lett. 71(1997), 903).
On the other hand, there is an example in which a perovskite oxide epitaxially grows on ceria (or yttria) (for example, (48) Appl. Phys. Lett. 68(1996)553). Therefore, if ceria can be controlled in orientation, a device making use of the characteristics of functional oxides will be realized.[0012]
Under these circumstances, technologies for epitaxially growing CeO[0013]2(001) on Si(001) substrates most important for practical application are greatly important.
OBJECTS AND SUMMARY OF THE INVENTIONIt is therefore an object of the invention to provide a method for crystal growth of oxides capable of epitaxially growing cerium oxide, yttrium oxide, and rare earth oxides having crystalline structures similar to them in the (001) orientation on (001)-oriented silicon substrates to realize epitaxial rare earth oxide(001)/silicon (001) structure; cerium oxides; promethium oxides; multi-layered structure of oxides; method for manufacturing a field effect transistor by using the method for crystal growth of oxides to make a gate insulating film; field effect transistor, method for manufacturing a ferroelectric non-volatile memory by using the method for crystal growth of oxides to make a ferroelectric capacitor; and ferroelectric nonvolatile memory.[0014]
The Inventors made researches toward resolution of the above-discussed problems involved in the conventional technologies. Their outline is explained below. In the explanation made below, growth of cerium oxide is taken as a typical example.[0015]
To attain the object of the invention, it is important that the surface of a silicon substrate has steps and terraces at least once. This can be realized by the method proposed by Ishizaka and Shiraki et al ((49) J. Electrochem. Soc., 133(1986)666), for example, or other similar methods. The growth apparatus is preferably one of molecular beam epitaxy (MBE) apparatus, laser ablation apparatus, reactive vacuum evaporation apparatus, and so forth, excellent in surface controllability and permitting to keep a clean surface in ultra high vacuum and permitting observation of the surface by reflection high energy electron diffraction (RHEED). Basically, however, any apparatus is acceptable provided it can control the pressure, temperature, and so on. When growing cerium dioxide (CeO[0016]2), cerium dioxide itself is used as the source material of cerium in most cases. However, since oxygen with a low vapor pressure selectively volatilizes, it is desirable to use cerium metal in order to control a low oxygen partial pressure. But oxide source materials are considered usable if the high vacuum can be maintained by additionally using a getter pump, for example for evacuation of the growth chamber, for example. Cerium metal has a high melting point and a low vapor pressure, and it is desirable to vaporize it by using a high-temperature Knudsen cell, electron beam vapor deposition, excimer laser, or the like. Additionally, for controlling interaction of these gases or metal elements, etc. with the highly active surface of the silicon substrate, it is important to employ a low-temperature process. There is no description directed to surface treatment in reports given heretofore, and there is one reporting that epitaxial CeO2could be made on a silicon substrate by using cerium metal and activating it by RF plasma under a substrate temperature of 450° C. through 600° C. and a oxygen partial pressure of (4-6 )×10−4Torr (assumed to be 10−2Torr near the substrate) ((50) Japanese Patent Laid-Open Publication No. hei2-267104). However, there is no description on epitaxial growth orientations. In the ranges of substrate temperature and oxygen partial pressure taught therein, a SiO2film is made on the silicon substrate surface, and growth of epitaxial CeO2(001) does not follow. If RF plasma is applied for activation, reaction of this mode will be promoted further.
The Inventors made researches on conditions for epitaxial growth of CeO[0017]2(001) on silicon (001) substrates from various standpoints, including a decrease of the growth temperature to control rates of generating cerium silicide, silicon oxide, for example. In conclusion, it is essentially important to prepare the surface of the silicon (001) substrate in form of a dimer structure of 2×1 and 1×2 surface structure before the growth, and it is also important how the source material is supplied during the growth. Regarding the latter subject, method of supply the source material, it is important, more specifically, to first start the supply of oxidic gas like oxygen onto the silicon (001) substrate surface and subsequently start the supply of the source material of Ce. Its reason has not been clarified yet. However, it is presumed that the surface of the silicon (001) substrate is covered by molecules or atoms of oxidic gas like oxygen before the supply of the source material of Ce is started, and this promotes (001) epitaxial growth of CeO2.
In addition to the above-mentioned matters, it is important to appropriately select the ratio of the supply amount of oxidic gas relative to the supply amount of the source material of Ce under a certain growth temperature, or the partial pressure of the oxidic gas or its supply amount when the supply amount of the source material of Ce is constant. FIG. 1 shows relationship between growth temperature T and ratio O/Ce of the supply amount of oxidic gas relative to the supply amount of the source material of Ce when the latter is constant. For convenience, shown on the ordinate of FIG. 1 are values of O[0018]2flow rate [sccm] instead of O/Ce. It is known from FIG. 1 that there is a certain restriction in the region where CeO2(001) can epitaxially grow. It is also known that the upper limit of the growth temperature is around 300° C. As the growth temperature decreases, the range of O/Ce or O2partial pressure increases. When it becomes lower than 100° C., a significant range can be obtained, and this is preferable for practical application.
The Inventors reached the present invention after making investigation from various viewpoints, in addition to the above-explained researches and knowledge.[0019]
According to the first aspect of the invention, there is provided a crystalline growth method of an oxide comprising the steps of:[0020]
processing a surface of a (001)-oriented silicon substrate into a dimer structure by 2×1, 1×2 surface reconstruction; and[0021]
epitaxially growing a rare earth oxide of a cubic system or tetragonal system in the (001) orientation on the silicon substrate.[0022]
According to the second aspect of the invention, there is provided a crystalline growth method of an oxide comprising the steps of:[0023]
processing a surface of a (001)-oriented silicon substrate into a dimer structure by 2×1, 1×2 surface reconstruction; and[0024]
epitaxially growing a rare earth oxide of a cubic system or tetragonal system in the (001) orientation on the silicon substrate by using a source material containing at least one kind of rare earth element in an atmosphere containing an oxidic gas.[0025]
In the second aspect of the invention, the rare earth oxide is epitaxially grown typically at a growth temperature lower than 300° C. and preferably a growth temperature not higher than 100° C.[0026]
According to the third aspect of the invention, there is provided a crystal growth method of an oxide comprising the steps of:[0027]
vaporizing a silicon oxide film from the surface of the (001)-oriented silicon substrate by heating it in vacuum with a pressure not higher than[0028]1×10−6Torr; and
epitaxially growing a rare earth oxide of a cubic system or tetragonal system in the (001) orientation on the silicon substrate from which the silicon oxide film is vaporized.[0029]
According to the fourth aspect of the invention, there is provided a cerium oxide having a bixbyite structure.[0030]
According to the fifth aspect of the invention, there is provided a promethium oxide having a bixbyite structure.[0031]
According to the sixth aspect of the invention, there is provided a multi-layered structure of oxides comprising:[0032]
a (001)-oriented silicon substrate;[0033]
a first CeO[0034]2film grown on the silicon substrate at a first growth temperature; and
a second CeO[0035]2film epitaxially grown on the first CeO2film at a second growth temperature higher than the first growth temperature.
In the sixth aspect of the invention, the second CeO[0036]2film is typically (001)-oriented. the first growth temperature is in the range from the room temperature to approximately 300° C., for example. A SiOxfilm may lie along the interface between the silicon substrate and the first CeO2film.
According to the seventh aspect of the invention, there is provided a multi-layered structure of oxides comprising:[0037]
a (001)-oriented silicon substrate;[0038]
a SiO[0039]xfilm on the silicon substrate;
an amorphous CeO[0040]yfilm on the SiOxfilm; and
a (001)-oriented CeO[0041]2film epitaxially arranged with respect to the silicon substrate on the amorphous CeOyfilm.
In the sixth and seventh aspects of the invention, x of the SiO[0042]xfilm is normally in the range of 1≦x≦2, and y of the amorphous CeOyfilm is normally in the range of 1.5≦y≦2.
Oxides multi-layered structures according to the six aspect (with and without the SiO[0043]xfilm between the silicon substrate and the first CeO2film) and the seventh aspect of the invention can be illustrated as shown in FIGS. 2, 3 and4, respectively.
According to the eighth aspect of the invention, there is provided a manufacturing method of a field effect transistor comprising the steps of:[0044]
processing a surface of a (001)-oriented silicon substrate into a dimer structure by 2×1, 1×2 surface reconstruction; and[0045]
forming a gate insulating film by epitaxially growing a rare earth oxide of a cubic system or tetragonal system in the (001) orientation on the silicon substrate.[0046]
In the eighth aspect of the invention, the gate insulating film is formed by epitaxially growing the rare earth oxide on the silicon substrate typically at a growth temperature lower than 300° C. and preferably at a growth temperature lower than 100° C. It is also possible that the surface of the silicon substrate is processed into the dimer structure by heating the silicon substrate in vacuum with a pressure not higher than 1×10[0047]−6Torr and thereby vaporizing a silicon oxide film from the surface, and the gate insulating film is formed by epitaxially growing the rare earth oxide on the silicon substrate.
According to the ninth aspect of the invention, there is provided a manufacturing method of a field effect transistor comprising the steps of:[0048]
processing a surface of a (001)-oriented silicon substrate into a dimer structure by 2×1, 1×2 surface reconstruction; and[0049]
forming a gate insulating film by epitaxially growing a rare earth oxide of a cubic system or tetragonal system in the (001) orientation on the silicon substrate in an atmosphere containing an oxidic gas at a growth temperature lower than 300° C. by using a source material containing at least one kind of rare earth element.[0050]
According to the tenth aspect of the invention, there is provided a manufacturing method of a field effect transistor comprising the steps of:[0051]
vaporizing a silicon oxide film from a surface of a (001)-oriented silicon substrate by heating the silicon substrate in vacuum with a pressure not higher than 1×10[0052]−6Torr;
forming a gate insulating film by epitaxially growing a rare earth oxide of a cubic system or tetragonal system in the (001) orientation on the silicon substrate from which the silicon oxide film is vaporized.[0053]
According to the eleventh aspect of the intention, there is provided a field effect transistor comprising:[0054]
a (001)-oriented silicon substrate;[0055]
a gate insulating film made of a (001)-oriented rare earth oxide of a cubic system or tetragonal system which is epitaxially grown on the silicon substrate; and[0056]
a ferroelectric film epitaxially grown on the gate insulating film.[0057]
According to the twelfth aspect of the invention, there is provided a manufacturing method of a ferroelectric non-volatile memory, comprising the steps of:[0058]
processing a surface of a (001)-oriented silicon substrate into a dimer structure by 2×1, 1×2 surface reconstruction;[0059]
epitaxially growing a rare earth oxide of a cubic system or tetragonal system in the (001) orientation on the silicon substrate; and[0060]
epitaxially growing a ferroelectric film on the rare earth oxide.[0061]
According to the thirteenth aspect of the invention, there is provided a ferroelectric nonvolatile memory characterized in the use of a field effect transistor which includes:[0062]
a (001)-oriented silicon substrate;[0063]
a gate insulating film made of a (001)-oriented rare earth oxide of a cubic system or tetragonal system which is epitaxially grown on the silicon substrate; and[0064]
a ferroelectric film epitaxially grown on the gate insulating film.[0065]
According to the fourteenth aspect of the invention, there is provided a ferroelectric nonvolatile memory comprising:[0066]
a (001)-oriented silicon substrate;[0067]
a (001)-oriented rare earth oxide of a cubic system or tetragonal system which is epitaxially grown on the surface of a first region of the silicon substrate;[0068]
a capacitor using a ferroelectric film which is epitaxially grown on the rare earth oxide; and[0069]
MIS-FET formed in a second region of the silicon substrate,[0070]
the capacitor and a gate electrode of the MIS-FET gate being connected to each other by wiring.[0071]
According to the fifteenth aspect of the invention, there is provided a ferroelectric nonvolatile memory comprising:[0072]
a single-crystal insulating substrate;[0073]
a (001)-oriented rare earth oxide of a cubic system of tetragonal system which is epitaxially grown on the surface of a first region of the single-crystal insulating substrate;[0074]
a capacitor using a ferroelectric film which is epitaxially grown on the rare earth oxide; and[0075]
MIS-FET formed in a silicon film which is epitaxially grown on the surface of a second region of the single-crystal insulating substrate,[0076]
the capacitor and a gate electrode of the MIS-FET being connected to each other by wiring.[0077]
In the present invention, the (001)-oriented silicon substrate intends to involve a silicon substrate offset from the (001) orientation within a range which can be regarded to be substantially equivalent to the (001) orientation.[0078]
In the present invention, the rare earth oxide is any one of oxides of rare earth elements like cerium (Ce), yttrium (Y), and so forth, and may be an oxide of one kind of rare earth elements, or two or more kinds of rare earth elements. When the rare earth oxide is expressed as ReO[0079]z(where Re is one or more rare earth elements), 0<z≦3 is normally satisfied. Although the rare earth oxide is a cubic system or a tetragonal system, the cubic or tetragonal system includes those slightly distorted within a range which can be regarded as the cubic or tetragonal system substantially. The rare earth oxide typically takes a fluorite structure (CeO2structure) or a C-rare-earth structure (Y2O3structure; bixbyite structure).
In the present invention, for the purpose of epitaxially growing the rare earth oxide in the (001) orientation more reliably, a source material containing at least one kind of rare earth element is supplied preferably after the oxidic gas is supplied onto the surface of the silicon substrate in the process of epitaxial growth of the rare earth oxide. The source material containing at least one kind of rare earth element may be either one made up of at least one kind of rare earth element or one made up of a rare earth oxide, for example. In this case, the rare earth element is a metal element.[0080]
In the present invention, a further step of annealing may be provided to anneal the rare earth oxide in vacuum of a pressure not higher than 1×10[0081]−6Torr at a temperature not lower than the growth temperature of the rare earth oxide after epitaxially growing the rare earth oxide. As a result of the annealing in the vacuum, part of oxygen is removed from the rare earth oxide. Especially when the rare earth oxide is CeOz, by annealing it in vacuum of a pressure not higher than 1×10−6Torr at a temperature not lower than the growth temperature of PmOzafter epitaxially growing CeOzand thereby removing part of O from CeOz, CeOz-dhaving an oxygen-defective fluorite structure or a C-rare-earth structure (bixbyite structure) can be produced. In CeOz-d, z-d is normally satisfies 1.5≦z-d≦2, and typically satisfies1.5<z-d<1.8. Also when the rare earth oxide is PmOz, CeOz-dhaving a bixbyite structure can be produced in the same manner. Additionally, a step is provided to homoepitaxially grow an additional rare earth oxide on the former rare earth oxide at a growth temperature higher than the growth temperature of the former rare earth oxide after its epitaxial growth. Further, another step may be provided to epitaxially grow a functional oxide on the rare earth oxide after its epitaxial growth.
In the present invention, depending upon the growth conditions, a silicon oxide film or defective layer not thicker than[0082]5 nm may be formed along the interface between the silicon substrate and the rare earth oxide after growth of the rare earth oxide.
In the present invention, the functional oxide typically has a perovskite structure or a layered perovskite structure. Essentially, any can be the functional oxide, such as, for example, ferroelectric material, superconductive material, pyroelectric material and piezoelectric material.[0083]
According to the invention having the above-summarized structure, by processing the surface of a (001)-oriented silicon substrate into a dimer structure by 2×1, 1×2 surface reconstruction and preferably supplying a source material containing at least one kind of rare earth element after starting the supply of an oxidic gas onto the surface of the silicon substrate in the process of epitaxial growth of a rare earth oxide, the rare earth oxide can be epitaxially grown excellently in the (001) orientation on the (001)-oriented silicon substrate.[0084]
The above, and other, objects, features and advantage of the present invention will become readily apparent from the following detailed description thereof which is to be read in connection with the accompanying drawings.[0085]