Quantum mechanical CASTEP software calculations were performed using nickel and rhenium atom deposition onto cleaved surfaces (active, hydrogenated, and oxygenated). These calculations were performed without metal atoms and with metal atoms at selected positions (origin, a- and b-axis) in the unit cell. Binding energies for each of the metal atoms (nickel and rhenium) were calculated. Additionally, calculated energy bands with associated density-of-states and partial density-of-states were examined regarding the population of s, p, and d bonding characteristics. Nickel atom deposition onto silicon-rich surfaces tended to bond to the silicon atoms as well as the underlying carbon atoms. However, rhenium atom deposition showed bonding only to the silicon atoms. This was observed experimentally and is reported herein. Experimentally, the rhenium deposition surface is extremely smooth and has only Ohmic characteristics and low resistance.

Rhenium (Re) thin-film contacts (100-nm thick) were deposited on carbon-rich, nominally stoichiometric, and silicon-rich 6H–SiC surfaces, which were moderately doped with nitrogen (1.28×1018cm−3). Morphology (Dektak), phase formation (x-ray diffraction), chemistry (Auger electron spectroscopy), and electrical properties (I–V) were characterized for the as-deposited and annealed (120min, 1000°C, vacuum <1×10−6Torr) contacts. As-deposited films were nonohmic. Films grown on carbon-rich surfaces were nonspecular, granular, and often delaminated during characterization. At room temperature in air, the Re films on stoichiometric SiC remained optically specular reflecting for 3h, but then became hazy from oxidation. The Re films on silicon-rich surfaces, stored in air at room temperature, resisted ex situ oxidation for approximately 24h. The annealed samples remained specular without visible signs of oxidation. The annealing resulted in a reduction in surface roughness for all the films regardless of substrate chemistry. The phase separation between carbon and rhenium was observed based on the formation of interfacial Re clusters and a ∼10-nm graphite surface layer after annealing. Auger data showed that Si layers (5–10nm) deposited to create Si-rich surfaces were partially consumed to form rhenium silicide during annealing, and the sharp Re∕Si∕SiC interface became more diffused with Re detected ∼50nm deeper into the structure. The annealing of Re films on moderately doped (1.28×1018cm−3) SiC resulted in ohmic contacts with an average specific contact resistance of 7.0×10−5Ωcm2 for stoichiometric and 1.6×10−5Ωcm2 for silicon-rich samples. The annealed contacts on carbon-rich surfaces remained rectifying.

Auger Electron spectroscopy was used to study electron-stimulated oxidation (ESO) of SiC. The rate of oxidation was investigated as a function of electron-beam exposure (on and off), primary electron-beam energy (3–6keV), electron-beam current (25–500nA) and total chamber pressure. The oxidation rate correlated with overall chamber pressure rather than the partial pressure of H2O, CO or CO2 alone. The rate decreased as the primary-beam voltage Ep was increased. The oxidation rate increased as the primary-beam current was increased at higher pressures (2.2×10−7Torr). Oxidation did not occur in the absence of the electron beam.

In metalization work, controlled semiconductor surfaces and interfaces are necessary. Trace impurities such as carbon and oxygen at the metal/semiconductor interface may improve or degrade the electrical characteristics of a desired contact. In this study, Auger Electron Spectroscopy (AES) was used to investigate electron stimulated oxidation (ESO) of SiC. The research focused on chamber conditions necessary to prevent oxide formation on SiC surfaces during thin film metal deposition. Oxidation rate was studied as a function of beam exposure ton and off), primary electron beam energy, E-p, (3-6 keV), electron beam current, I-p, (25-500 nA), and total chamber pressure. Oxidation rate had a stronger correlation to overall chamber pressure rather than the partial pressure of H2O, CO, or CO2 alone. Oxidation rate decreased slightly as E-p was increased. Rate of oxidation increased at higher pressures (2.2 x 10(-7) Torr) with respect to I-p but was nearly independent of I-p at lower pressures (7.1 x 10-9 Torr). No oxide growth was detectable to AES in the absence of exposure to the electron beam.

SiC has generally been plasma etched in polymer-forming chemistries such as CHF3/O-2 or CF4/O-2, often with addition of H-2 to achieve acceptable surface morphologies. We find that under high ion density conditions gases such as SF6, NF3, IBr, and Cl-2 produce smooth surfaces that are free of hydrogen passivation effects. Etch rates in excess of 1500 Angstrom/min are achieved in electron cyclotron resonance (ECR) NF3 or Cl-2/Ar discharges with low additional rf chuck powers (100-150 W); dc bias of -120 to -170 V. The rates are somewhat lower (factors of 2-4) with IBr and SF6 chemistries, Ion-induced damage is evident from Hall measurements for SIC exposed to rf powers >150 W (dc bias >-170 V) under ECR conditions and >250 W (dc bias >-275) under reactive ion etch conditions. Efforts to anneal damage at these higher powers reveals a major annealing stage is evident at similar to 700 degrees C, with an activation energy of similar to 3.4 eV, but there is significant damage remaining even after 1050 degrees C annealing. Hydrogen passivation is a problem only in p-type SiC, and is removed at similar temperatures to ion-induced damage by annealing at similar to 700 degrees C under N-2 ambients. This is strongly correlated with secondary ion mass spectrometry measurements on deuterated samples annealed at different temperatures. (C) 1997 American Vacuum Society.