Highly rectifying silicon Schottky contacts using energetically deposited graphitic carbon

The formation of high quality Schottky diodes (i.e. high rectification ratio, low saturation current, low series resistance, and ideality factor close to 1) by Filtered Cathodic Vacuum Arc deposition of carbon (FCVA C) onto p-type silicon substrates has been demonstrated in this thesis. Electrical measurement results clearly indicate that the choice of FCVA’s deposition parameters (i.e. energy and temperature) have an influence on the quality of the Schottky junction. Moreover, these measurements have assisted in identifying both the effect of the varying microstructure of graphitic carbon films (sp2/sp3 ratio) caused by different deposition parameters and the presence of a thin, resistive interfacial layer between carbon and p-Si. Technology Computer Aided Design (TCAD) simulations based on a Metal-Insulator-Semiconductor (M-I-S) and on a Metal-Resistor-Semiconductor (M-R-S) diode structures were constructed to fit the experimental data. The M-I-S diode structure is a 2-D model where the interfacial layer was considered as an insulator. This model estimates the ‘metal’ work function and the thickness of the interfacial layer to approximate the experimental I-V results. The alternative M-R-S diode structure utilised 2-D and 3-D models to simulate the electrical behaviour of carbon/p-Si diodes. The primary feature in the M-R-S 2-D and 3-D models is the definition of a resistive interface (representing experimentally observed carbon/Si mixed interfacial layer) between the graphitic carbon thin film and Si. In addition to metal work function determination, Schottky barrier lowering can be incorporated in the M-R-S model to include image force effects reported in literature. Both models achieved excellent agreement with the measurement results in the forward I-V region. For the M-R-S model, the agreement with the measurement results in the reverse I-V region has improved significantly compared to simulation using the M-I-S model. The M-R-S model provided information that could not be obtained from measurements and suggested a path to improved devices.