The state-of-the-art in each of the major materials (Diamond, SiC and III-nitrides) was discussed with the advantages, disadvantages, and requirements of each enumerated.

In grain boundaries in compound semiconductors such as SiC, the interface stoichiometry and the wrong bonds between like atoms are of much importance. Firstly, a general definition of the interface stoichiometry in such grain boundaries has been discussed. Secondly, the atomic and electronic structures of the {211} Σ=3 boundary in SiC have been examined by using the self-consistent tight-binding method, based on the atomic models with bonding networks similar to those in the models of the same boundary in Si or Ge. The wrong bonds have significant effects through the large electrostatic repulsion and the generation of localized states as well as those in the {122} Σ=9 boundary in SiC. And the different bond lengths of the wrong bonds very much affect the local bond distortions at the interfaces, which determines the relative stability among the present models.

Surface electronic structures of β-SiC reconstructed (001) surfaces and the Al/β-SiC(001) interface have been investigated by employing a tight-binding method. Distinct surface electronic characteristics corresponding to different surface reconstructions are discussed based on the interpretation of surface density of states. The calculations of the Al/β-SiC (001) interface indicate that aluminum deposition on β-SiC(001) surface may induce the substrate to return to the ideal unreconstructed surface and that Al-C interaction is stronger than Al-Si interaction. Al deposition on C-rich surfaces may form a better bonded interface than that on the Si-rich surfaces. Our findings are in good agreement with available experimental and theoretical results.

The molecular dynamics method employing an empirical potential energy function to describe the Si-C interaction has been used to determine the minimum energy sites for Si and C adatoms on C-terminated SiC (001) substrates. It is found that whereas a single C adatom lies on the carbon dimer bond, this site only becomes energetically favourable for silicon adatoms when they interact to form a dimer pair.

Using an accurate tight-binding method, we examine the structure and dynamics at a C(lll) step. We find the H vibration frequencies at a step are distinguishable from the flat surface. Removal of H from the surface creates an unstable bulk terminated diamond structure which relaxes to a graphitic layer and pulls away from the bulk. The explanation for these results can be given in terms of the relationship of the band structure to the relaxed geometries. The addition of H stabilizes this surface and reverts it back to sp3 bonding. The addition of CH3 near a step is also discussed. These results are important towards the understanding of possible diamond growth mechanisms that occur at a step.

A mass-selected low kinetic energy (1–50 eV) ion source is used to expose the diamond (111) surface to ionized atomic hydrogen (H+) at controlled impact kinetic energy. We report the result of 20 eV and 50 eV kinetic energy exposures as measured by ultraviolet photoelectron spectroscopy (UPS). UPS is found to be a useful probe of hydrogen adsorption on diamond.

Secondary electron (SE) yield was enhanced by a factor of ∼30 and surface conductance increased up to 10 orders of magnitude when O-terminated or non-terminated natural diamond (100) surfaces were exposed to atomic H. The SE yield from atomic H-exposed surfaces was spatially dependent on near-surface microcrystalline perfection enabling defect-contrast imaging in the conventional SE mode of the scanning electron microscope (SEM). Ultraviolet photoelectron spectroscopy (UPS) on atomic H-exposed surfaces revealed an intense low energy peak attributed to photoexcitation of secondary electrons into unoccupied hydrogen-induced states near the conduction band edge and their subsequent escape into vacuum. The low energy photoemission peak, enhanced SE yield and enhanced surface conductivity were completely removed via high temperature annealing or exposure to atomic O creating the denuded or O-terminated surfaces, respectively.

The surface Fermi level position of undoped epitaxial diamond layers is estimated from contact potential difference between Au reference and diamond measured by Kelvin probe method. The surface Fermi level position of the as-grown layer is located at the energy of 0.75 eV above the valence band edge. O2 plasma treatment leads to an upward shift of the surface Fermi level position to an energy of 1.89 eV from the valence band edge. The surface Fermi level is located at an energy of 0.97 eV above the valence band edge after H2 plasma treatment. Reversible change in the surface Fermi level position is found between O2 and H2plasma treatments. A change in the band bending is observed at the surface of polycrystalline diamond films treated with various ways by X-ray photoclcctron spectroscopy (XPS) analysis. A variation in the current-voltage characteristics of epitaxial and polycrystalline diamonds treated with O2 and H2 plasmas can be qualitatively explained in terms of a change in the band bending due to the shift of the surface Fermi level position.

A novel substrate preparation procedure which can be employed to remove the original surface from as-received C(001) natural diamond substrates has been developed. A description of the various substrate processing steps which includes, low-energy ion implantation of C and O, high-temperature annealing, electrochemical etching and surface plasmas treatments is presented. Also demonstrated is the growth of topographically excellent homoepitaxial films by rf-plasma-enhanced chemical vapor deposition using water/ethanol mixtures on C(001) substrates.

The reconstruction of the C(100) surface and its reaction with CH2 are studied by a cluster model at several theoretical levels [1–13]. For the reconstruction ofthe C(100) surface, the calculated surface dimer bond length is found to be very sensitive to the level of theoretical treatment and the spin state. A single-determinant SCF treatment gives a closed-shell singlet state, higher in energy than the triplet state, and with a dimer length of 1.401 Å, 0.279 Å shorter than the triplet. The correct ground state is a singlet, but a multi-determinant wavefunction is required for its description. At the CI level, the surface dimer bond length in the ground state is found to be 1.508 Å and the energy decrease on dimer formation with respect to the ideal C(100)-lxl surface is 2.28 eV per dimer. For the reaction of CH2 with C(100), no barrier is found for the chemisorption of CH2on the surface and the reaction is highly exothermic. The surface is converted from C(100)-2×1 to C( 100)-1×1 upon CH2 chemisorption.

Studies of diamond heteroepitaxy on silicon indicate that C-C surface species act as nucleation precursors. We have investigated the conversion of the Si(100) 2×1 surface to SiC using C2H4 to obtain an understanding of how C-C species may be formed and to determine the effect of an O-adlayer on enhancing or selecting the reaction channel which leads to these species. Under appropriate conditions, the interaction between C2H4 and the clean silicon surface yields both SiC and C-C species. The presence of an O-adlayer significantly reduces the activity of silicon and enhances the formation of sp2 and sp3 C-C species. These results provide key insights into diamond nucleation conditions in conventional growth processes.

Two wet chemical cleaning processes (a conventional chromic acid clean and an electrochemical etch) and a H-plasma exposure have been employed to clean natural type lib semiconducting diamond C(001) wafers. The effects of these processes on the diamond surface have been assessed and compared. As evidenced by Auger electron spectroscopy (AES), an oxygen free surface could be obtained following annealing to 900°C for the electrochemical process compared to 1050°C for the chromic acid etch. In addition, the technique of Atomic Force Microscopy (AFM) demonstrated the presence of oriented pits on the surface of samples electrochemically etched for long times at high currents. Furthermore, heteroepitaxial Cu films have been grown on the diamond substrates cleaned by a process as described above. By means of Ultraviolet Photoemission Spectroscopy (UPS) a Schottky barrier height of ΦB≊ 1.0 eV was measured. Furthermore, the presence a negative electron affinity (NEA) has been determined.

We report photoemission measurements from the (111)2×1 diamond-titanium monoxide (TiO) interface. Submonolayer deposition of TiO on the (111)2×1 diamond surface modifies the electron affinity of the surface from positive to negative. No change in the band bending was observed as a result of the TiO deposition. Total electron yield measurements from the diamond-TiO interface were also performed. The yield spectra, as expected for a negative electron affinity (NEA) surface, have emission thresholds that are in good agreement with the absorption coefficient of diamond. To further understand the emission properties of NEA (111) diamond surfaces we also compare the electron yield photo-excitation spectra of the diamond-TiO interface with the yield spectra of hydrogenated (111)1×1:H diamond surfaces.

This study presents the results of surface investigation of the heteroepitaxial AIN/SiC interface. The analytical tools employed included UPS, XPS, Auger spectroscopy, and LEED. The surface electronic states were characterized by uv photoemission obtained at surface normal. Conclusions drawn from this study are that the AIN/SiC structure results in a negative electron affinity surface which is extremely sensitive to defect density. The surface Fermi level is found to be near the middle of the AlN gap, and a possible band alignment between the AlN and SiC is presented.

Application of surface science methods to single crystal diamond surfaces requires the preparation of clean, well-ordered surfaces and accurate measurement of substrate temperature. Cleaning of diamond (100) in H2SO4/HNO3/HClO4 produced several infrared absorption features between 1025 and 1275 cm-1, as observed by infrared multiple-internal-reflection spectroscopy. These modes are assigned to surface hydroxyl and bridge-bonded oxygen. Heating an oxidized surface to ca. 1130 °C caused disappearance of a surface hydroxyl mode centered at 1080 cm-1. We show by atomic force microscopy that an as-polished diamond (100) sample is covered by grooves and ridges several nm in height, implying a modest density of atomic steps. The surface of a diamond that underwent etching via numerous adsorption/desorption experiments in ultrahigh vacuum and was acid cleaned several times was essentially unchanged, indicating a minimal perturbation of the surface topography. The capability of Fizeau interferometry for accurate measurement of single-crystal diamond temperatures is demonstrated.

An overview of enabling materials technologies required for fabrication of electronic devices on diamond is presented. Emphasis is placed on electronic doping of diamond by boron ion implantation. Van der Pauw resistivity and Hall Effect measurements were used to determine the net carrier concentration, carrier mobility and resistivity of natural and synthetic diamonds implanted under various conditions. The measured results for a range of implantation conditions and post-annealing temperatures are discussed in the context cf a model developed by J.F. Prins1. The requirements placed on ohmic contacts to diamond, and a process for fabricating ohmic contacts, is discussed briefly. Finally, current-voltage characteristics of a simple MISFET fabricated on ion implanted natural diamond are presented and analyzed. 1J.F. Prins, Physical Review B, 38 (1988) 5576.

Metal-oxide-semiconductor field-effect transistors (FETs) have been fabricated using B-doped diamond thin films deposited on polycrystalline, (100) highly-oriented, and single crystal diamond insulating substrates. Diamond films were grown using a microwave plasma chemical vapor deposition technique. Various electrical and materials characterization techniques were employed to confirm that the films exhibited properties suitable for FET fabrication. Devices with gate lengths and widths of 2 μm and 314 μm respectively, were processed using standard photolithography. Silicon dioxide was used as the gate dielectric. Current-voltage characteristics of these devices have been measured during variable temperature cycling in air. Devices fabricated on the randomly oriented polycrystalline diamond substrates have been operated to 285°C. Field-effect transistors fabricated using the highly-oriented diamond substrates have been characterized to 400°C. Single crystal diamond devices exhibited saturation and pinch-off of the channel current at temperatures up to 500°C. These devices have been biased in amplifier circuit configurations that have been characterized from 20 Hz to 1 MHz. Single crystal FETs exhibited voltage gain over an extended temperature range. Transconductances as large as 1.7 mS/mm have been observed. The electronic properties, fabrication technologies, and performance of devices fabricated on the three diamond substrate materials will be discussed and compared.

Diamond is suitable for use as an ionizing particle detector for high rate, high radiation, and/or chemically harsh environments. A sampling calorimeter, a detector measuring the total energy of an incident particle, consisting of 20 alternating layers of diamond and tungsten has been constructed and tested. The diamond for the detector layers was grown by chemical vapor deposition with an averaged thickness of 500 μm. The active area of each layer was 3×3 cm2 with ohmic contacts on opposite faces forming a metal-insulator-metal structure. The calorimeter was tested with electrons of energies up to 5.0 GeV. The response of the diamond/tungsten calorimeter was found to be linear as a function of incident energy. A direct comparison of diamond/tungsten and silicon/tungsten calorimeters was made.

The planarization of rough polycrystalline diamond films synthesized by DC arc discharge plasma jet CVD was attempted using KrF excimer laser pulses. The effects of laser incidence angle and reaction gases (ozone and oxygen) on etching rate were studied. The temperature change of diamond and graphite with different laser fluences was calculated by computer simulation to explain the etching behavior of diamond films. The calculated threshold energy density for etching of pure crystalline diamond was about 1.7 J/cm2. However, the threshold energy density was affected by the angle of laser incidence. Preferential etching of a particular crystallographic plane was observed through scanning electron microscopy. The etching rate of diamond with ozone was lower than that with oxygen. Also, the etching rate of diamond films at normal laser incidence was lower than that of films tilted at 45° for laser fluences above 2.3 J/cm2. When the angle of incidence was 80° to the diamond surface normal, the peak-to-valley surface roughness was significantly reduced, from 30 μm to 0.5μm.

The breakdown electric field of 4H-SiC as a function of doping was measured using pn junction rectifiers, with maximum voltages of 1130 V being achieved. 4H-SiC vertical power MOSFET structures have shown specific on-resistances of 33 mΩ-cm2 for devices capable of blocking 150 V. A current density of 100 A/cm2 was achieved at a drain voltage of 3.3 V. Thyristors fabricated in SiC have also shown blocking voltages of 160 V and 100 A/cm2 at 3.0 V. High temperature operation was measured, with the power MOSFETs operating to 300°C, and the thyristors operating to 500°C.

Submicron 6H- and 4H-SiC MESFETs have shown good I-V characteristics to Vd= 40 V, with an Idss of 200–300 mA/mm. The maximum operating frequencies (fmax) achieved for 6H-SiC MESFETs is 13.8 GHz, with small-signal power gains of 9.8 dB and 2.9 dB at 5 GHz and 10 GHz, respectively. 4H-SiC MESFETs have demonstrated an RF output power density of 2.8 W/mm at 1.8 GHz. This is the highest power density ever reported for SiC and is 2–3 times higher than reported for comparable GaAs devices.