Atomic scale modeling of silicate interface properties for high-k gate dielectric applications

Aggressive scaling has led to silicon dioxide (SiO₂) gate dielectrics as thin as 15 Å in state-of-the-art CMOS technologies. As a consequence, static leakage power due to direct tunneling through the gate oxide has been increasing at an exponential rate. As technology roadmaps call for sub-10 Å gate oxides within the next five years, a variety of alternative high-k materials are being investigated as possible replacements for SiO₂. The higher dielectric constants in these materials allow the use of physically thicker films, potentially reducing the tunneling current while maintaining the gate capacitance needed for scaled device operation.; Atomic scale modeling methods based on first principles density functional theory are applied to a promising class of alternative dielectrics known as silicates (e.g. ZrSi_xO_y). Since the quality of the Si interface will ultimately determine the feasibility of an alternative dielectric, this work has focused on the interface properties of silicates in the context of MOS scaling limits.; Model interface calculations have been performed to show that oxide-like bonding of zirconium is energetically favored over silicide-like bonding at the Si interface; extended suboxide states within the interfacial transition region significantly limit the scalability of the silicate system from the point of view of equivalent oxide thickness. First principles calculations have also been performed to quantify the expected lowering of the conduction band offset with increasing Zr concentration. The use of charge transfer dipoles at the interface has been proposed as a means of overcoming the inherent asymmetry of silicate band offsets. Because limitations are shown to arise at a Si/SiO ₂-like interface, calculations suggest that possible solutions to overcoming them can be built on extensive knowledge of the Si/SiO₂ system.