Native Silicon Oxide Properties Determined by Doping.

The physico-chemical properties of native oxide layers, spontaneously forming on crystalline Si wafers in air, can be strictly correlated to the dopant type and doping level. In particular, our investigations focused on oxide layers formed upon air exposure in a clean room after Si wafer production, with dopant concentration levels from ≈1013 to ≈1019 cm-3. In order to determine these correlations, we studied the surface, the oxide bulk, and its interface with Si. The surface was investigated using the contact angle, thermal desorption, and atomic force microscopy measurements which provided information on surface energy, cleanliness, and morphology, respectively. Thickness was measured with ellipsometry and chemical composition with X-ray photoemission spectroscopy. Electrostatic charges within the oxide layer and at the Si interface were studied with Kelvin probe microscopy. Some properties such as thickness, showed an abrupt change, while others, including silanol concentration and Si intermediate-oxidation states, presented maxima at a critical doping concentration of ≈2.1 × 1015 cm-3. Additionally, two electrostatic contributions were found to originate from silanols present on the surface and the net charge distributed within the oxide layer. Lastly, surface roughness was also found to depend upon dopant concentration, showing a minimum at the same critical dopant concentration. These findings were reproduced for oxide layers regrown in a clean room after chemical etching of the native ones.

4H-SiC surface morphology after Al ion implantation and annealing with C-cap.

The root mean square (rms) surface roughness extracted from atomic force microscopy is widely employed to complement the characterisation of ion implantation processes in 4H-SiC. It is known that the protection of a carbon film eliminates or mitigates roughening of the SiC surface during postimplantation annealing. This study, based on a rich original data collection of Al+ ion implanted 4H-SiC samples, allows for a quantitative description of the surface morphology as a function of the annealing temperature and time and of the Al implanted concentration. With increasing thermal budget, the evolution from flat, to blurred with ripples, granular, and finally jagged surface, results in a monotonous increase in the root mean square roughness. Additional information is given by the trends of the roughness exponent and of the correlation length, extracted from the height-height correlation function, which account for the surface evolution below 1700°C and for the effect of the Al implanted concentration on the ripple size, respectively. A combination of low roughness parameter and high correlation length identify the transition from ripples to jagged morphology. LAY DESCRIPTION: Selective area doping is a key step in the fabrication of hexagonal Silicon Carbide (4H-SiC) power electronic devices. It is achieved by ion implantation followed by a high temperature postimplantation annealing to restore the lattice and electrically activate the dopants. Aluminium, the preferred p-type dopant, is electrically activated at temperature ranging between 1500°C and 2000°C. The time required to complete the activation process is longer the lower the annealing temperature, spanning between some minutes and hundreds of hours. During annealing, 4H-SiC wafers are encapsulated by a temperature-resistant carbon layer (C-cap) in order to avoid step bunching and reduce surface roughening. Nevertheless, surface modifications can occur at high temperature. For this reason, the characterisations of 4H-SiC doping processes report not only the electrical activation of the dopants, but also the root mean square surface roughness obtained at the end of the process. However, rms values can be scattered because technological parameters such as the heating system and the way to deposit and remove the C-cap can affect the final result as well as the process parameters. Furthermore, the C-cap resistance to long annealing has been proven only by electrical measurements, but the surface morphology has never been observed. This work presents a quantitative characterisation of the surface morphology of Al implanted 4H-SiC as a function of the annealing temperature, time and of the Al implanted concentration, independent of the heating system and of the C-cap technology. The produced sample collection allowed to correlate characteristic surface features with the corresponding quantities extracted from image analysis that can be more sensitive to process parameters than the sole rms. These findings can be used to enrich process optimisation tools.

Size effect on high temperature variable range hopping in Al+ implanted 4H-SiC.

The hole transport properties of heavily doped 4H-SiC (Al) layers with Al implanted concentrations of 3 × 1020 and 5 × 1020 cm-3 and annealed in the temperature range 1950-2100 °C, have been analyzed to determine the main transport mechanisms. This study shows that the temperature dependence of the resistivity (conductivity) may be accounted for by a variable range hopping (VRH) transport into an impurity band. Depending on the concentration of the implanted impurities and the post-implantation annealing treatment, this VRH mechanism persists over different temperature ranges that may extend up to room temperature. In this framework, two different transport regimes are identified, having the characteristic of an isotropic 3D VRH and an anisotropic nearly 2D VRH. The latter conduction mechanism appears to take place in a rather thick layer (about 400 nm) that is too large to induce a confinement effect of the carrier hops. The possibility that an anisotropic transport may be induced by a structural modification of the implanted layer because of a high density of basal plane stacking faults (SF) in the implanted layers is considered. The interpretation of the conduction in the heaviest doped samples in terms of nearly 2D VRH is supported by the results of the transmission electron microscopy (TEM) investigation on one of the 5 × 1020 cm-3 Al implanted samples of this study. In this context, the average separation between basal plane SFs, measured along the c-axis, which is orthogonal to the carrier transport during electrical characterization, appears to be in keeping with the estimated value of the optimal hopping length of the VRH theory. Conversely, no SFs are detected by TEM in a sample with an Al concentration of 1 × 1019 cm-3 where a 3D nearest neighbor hopping (NNH) transport is observed.