Solid phase epitaxy of 3C-SiC on Si(001)

[0001] This application claims priority to and the benefits of U.S. Patent Application No. 62/067,021 filed on Oct. 22, 2014, the entirety of which is herein incorporated by reference.

[0002] Silicon carbide (SiC) possesses unique electronic and physical properties such as wide band gap, high breakdown field, high thermal conductivity, and low thermal expansion coefficient. These unique electronic and physical properties make it the material of choice for power electronics that withstand higher breakdown voltage and operate at higher temperatures, far exceeding the performance of their Si counterparts.

[0003] SiC exists in a variety of polymorphic crystalline structures called polytypes, e.g., 3C—SiC, 6H—SiC, 4H—SiC, etc., where “C” indicates cubic phases, and “H” hexagonal. The different polytypes can be described by the different ordering, or stacking sequence, of the Si—C bilayers along the c-axis, and can have a wide range of different electronic and physical properties. Out of the more than 200 polytypes, the 3C (cubic zinc blende), 4H and 6H hexagonal structures have the lowest formation energies.

[0004] The 3C—SiC is the only cubic phase, and is advantageous for metal-oxide-semiconductor (MOS) device architecture due to its higher channel mobility because near interface traps are not active in the SiO₂/3C—SiC system. In addition, its cubic structure in principle enables it to be realized as a thin film on Si(001) and thereby readily integrated with current Si MOS technology, a key criterion for future technological application. Conversely, the hexagonal structure of the 4H and 6H phases are incompatible with growth on Si(001). Integration as a thin film on Si also has significant economic implications, because SiC bulk crystals are very costly to produce, and at present only the 4H— and 6H—SiC bulk wafers are available commercially—no bulk 3C—SiC single crystal substrate wafers are available, and thus it must be grown on a different substrate.

[0005] Therefore there has been significant interest to produce low-cost, large area 3C—SiC films on a low cost substrate which serves as a surrogate substrate for subsequent epitaxial growth of 3C—SiC single crystal thin film device heterostructures.

[0006] In recent years, considerable efforts have been spent on the growth of SiC on Si substrates, typically by molecular beam epitaxy (MBE) and chemical vapor deposition (CVD) techniques. However, due to the large lattice mismatch (20%), the resulting SiC films are not of high enough quality required for high-power electronic devices. The use of a buffer layer has been shown to seed the growth of higher quality SiC films. Previous buffer layers have been created through the chemical reaction of the Si substrate and carbon-containing gases such as C₂H₄ and CO at elevated temperatures.

[0007] When the buffer is porous, as is in the case of CO due to the formation of volatile Si—O species, the significant relaxation of the elastic stresses between the SiC film and Si substrate leads to films with far less defects.

[0008] However, this process with CO gas has only been done during high pressure CVD.

[0009] Here, an alternate method of producing a porous SiC buffer layer on a Si substrate is demonstrated, through solid state reaction with a pre-deposited amorphous C film at about 950° C. in ultrahigh vacuum (UHV).

[0010] This new method has several advantages over current processes. First, the resulting 3C—SiC buffer layer is much thinner (1-2 nm), with a thickness directly controlled by the thickness of the C film. Second, it works at a much lower temperature (about 950° C.) than those with gaseous species (1100-1400° C.), reducing the thermal budget and the excess energy available for formation of undesirable defects. Third, it is compatible with both CVD and MBE, the two major techniques used to grow device quality 3C—SiC heterostructures.

[0011] This invention demonstrates a new route for the synthesis of a 3C—SiC buffer layer on Si(001) for the purpose of seeding subsequent epitaxial growth of 3C—SiC on Si(001) for the integration of 3C—SiC with Si electronic technology. The synthesis occurs by depositing and annealing an amorphous carbon film on Si(001), resulting in solid state epitaxy of a few nanometer thick crystalline 3C—SiC layer on Si(001).

[0017] This invention demonstrates a new route for the synthesis of a 3C—SiC buffer layer on Si(001) for the purpose of seeding subsequent epitaxial growth of 3C—SiC on Si(001) for the integration of 3C—SiC with Si electronic technology. The synthesis occurs by depositing and annealing an amorphous carbon film on Si(001), resulting in solid state epitaxy of a few nanometer thick crystalline 3C—SiC layer on Si(001).

[0018] This is a method of producing a porous SiC buffer layer on a Si substrate.

[0019] This method concerns a solid state reaction with a pre-deposited amorphous C film at about 950° C. in ultrahigh vacuum (UHV).

[0020] The amorphous carbon layer is deposited on a hydrogen passivated Si(001) substrate by magnetron sputtering of a C target at room temperature at a rate of 0.8 nm/min. The layer thickness can be precisely controlled by deposition time.

[0021] The sputtering conditions were 15 sccm Ar flow, 3 mT total pressure, and 100 W DC plasma power.

[0022] The deposited film exhibits a smooth surface morphology with a root-mean-square (RMS) roughness of 0.3 nm as determined by atomic force microscopy, as illustrated in FIG. 1.

[0023] The samples were then mounted on a Ta holder and transferred into an ultrahigh vacuum (UHV) system with a high-temperature heating stage. To form the SiC layer, samples were annealed at temperatures up to 950° C. for 30 minutes, with a ramp rate of ˜1° C./sec and pressure below 2×10⁻⁹ Torr. The temperature was monitored by a thermocouple embedded in the heating stage.

[0024] The formation of SiC is monitored by in situ x-ray photoelectron spectroscopy (XPS). Shown in FIG. 2 are XPS spectra for the C 1s and Si 2p signals for the as-deposited amorphous carbon film, and upon annealing at 850 and 950° C. The as-deposited C film exhibits a single carbon peak at 284 eV and doublet Si 2p peak at 99.5 eV with a broad hump at ˜102 eV. The feature at ˜102 eV is indicative of the formation of suboxides, which typically forms during or after the carbon deposition. For annealing below 850° C., no changes in the C 1s or Si 2p peaks were observed, indicating no reactions between the C and Si substrate.

[0025] Upon annealing at 850° C., we observe the presence of a shoulder on the C 1s peak at ˜282.5 eV and the onset of a peak at 101.5 eV in the Si 2p spectra. These emerging features with shifts in binding energy are consistent with the formation of Si—C bonds. The O 1s peak (not shown) is nearly absent indicating that the suboxide has either decomposed or desorbed at this temperature.

[0026] Upon annealing at 950° C., a single C peak is observed at 282.5 eV, indicating a complete conversion of C—C bonds to C—Si bonds. Similarly, the Si 2p peak is now composed of two peaks; a doublet at 99.5 eV characteristic of the substrate, and a peak at 101.5 eV indicative of the Si—C bond. These results indicate that initial SiC formation occurs at 850° C., and a complete SiC layer is formed at 950° C.

[0027] FIG. 3 shows a high-resolution transmission electron microscopy (TEM) image of a sample annealed at 950° C. A ˜5 nm thick SiC is seen on top of the Si(001) with a high density of fine faults along {111} planes. The fast Fourier transform (FFT) close to the [110] zone axis generated from the box area containing a portion of Si and SiC is shown as an inset. The d-spacing of {111} planes of SiC, calibrated with respect to Si(111), is ˜2.51 Å, consistent with that of 3C—SiC. The angle between two sets of SiC {111} planes is ˜70.5°, confirming that it is fcc type. It exhibits a cube-on-cube orientation relationship that can be written as (1-11)_Si∥(1-11)_Sic and [110]_Si∥[110]_Sic. Note that the streaks seen along the [1-11] direction in the FFT pattern are due to the fine faults, which help relieves the strain due to the large lattice mismatch between SiC and Si.

[0028] Pores tens of nanometers in diameter and ˜20 nm deep are also observed below the SiC epilayer in the Si substrate for samples annealed at 950° C., as illustrated in FIG. 4A. AFM observations also show pits spaced hundreds of nanometers apart on the surface of the SiC, where areas between pores exhibit a smooth morphology, consistent with the crystallization of the SiC. These observations of the pores are similar to those due to reactions with CO gas, which has been shown to help relieve strain and improve epilayer quality, and are formed here as a result of the outward diffusion of Si atoms to react with the C that are on the surface to form SiC.

[0029] The structure and electronic properties of the SiC/Si interface is calculated by density functional theory (DFT). Many interfaces were examined with varying stoichiometries. The interface with the lowest free energy is shown in FIG. 5, which contains less than a monolayer of C at the interface. The large lattice mismatch results in dislocations at the interface that are spaced approximately every 4 Si lattice spacings and every 5 SiC lattice spacings. The Si valence band is found to be at 1.27 eV above the SiC valence band. Thus the conduction bands are nearly identical across the interface. One mid-gap state, an acceptor band 230 meV above the Si valence band, appears and is well localized at the interface. The planar averaged charge density is also shown.

[0030] This all-solid thin film route offers a new approach for the solid phase epitaxy of a single crystalline 3C—SiC buffer layer to seed the growth of high quality 3C—SiC epilayer for the integration with Si MOS technology.

[0031] The use of a pre-deposited amorphous C layer as the carbon source in place of gaseous species such as CO offers significant advantages that the SiC buffer layer is single cubic phase, works at a much lower temperature (950° C.) than those with gaseous species (1100-1300° C.), is compatible with MBE and with CVD.

[0032] This new process also produces a porous SiC buffer layer which will relieve strain and improve crystallinity of the SiC epilayer. In addition, the SiC buffer layer can be much more precisely controlled down to the nanometer scale, by controlling the thickness of the amorphous carbon layer.

[0033] Many modifications and variations of the present invention are possible in light of the above teachings. It is therefore to be understood that the claimed invention may be practiced otherwise than as specifically described. Any reference to claim elements in the singular, e.g., using the articles “a,” “an,” “the,” or “said” is not construed as limiting the element to the singular.