Scaling of gate dielectrics for advanced CMOS applications

It is widely believed that a gate oxide thinner than 1.5 nm is required for future CMOS devices within 10 years, if the traditional scaling trend is to be continued. Apart from manufaturability and reliability issues, a major problem with such a thin gate oxide is the high gate leakage current due to quantum mechanical tunneling of electrons and holes. Despite such potential problems, many researchers have continued to push the limit of thermal SiO2. Meanwhile, many laboratories are actively searching for alternative gate dielectrics to replace thermal SiO2 when it becomes necessary. The candidates that show promise include silicon nitride, silicon oxynitride, TiO2, Ta2O5, ZrO2, HfO2, and SrTiO3. This paper will review the recent research advances in all of the aforementioned gate dielectrics, and discuss the challenges that lie ahead.