High-k Gate Dielectric Reliability

After CMOS technology comes into 45nm, High K gate dielectric will become the promising candidate of SiO2 and SiNxOy to release the unacceptable and huge gate leakage current due to quantum direct tunneling through ultra thin barrier. Recently, HfO2 received much focus due to its suitable dielectric constant, good thermal stability to silicon and low defect density. In this paper, the reliability issues of ultra-thin (EOT<1nm) HfN/HfO2 gate stack were extensively and in detail discussed and studied. The main obtained results and conclusion are as follows: 1) Stress-induced Leakage current of the ultra-thin HfN/HfO2 gate stack with low trap density is negligible before gate dielectric breakdown; 2) For the studied ultra-thin HfN/HfO2 gate stack, its TDDB characteristics are determined by intrinsic traps not process-induced traps. 3) It's firstly observed and proved that there is electric field dependent breakdown mechanism of ultra-thin HfN/HfO2 gate stacks under gate injection. Under high voltage stress, HfO2 bulk layer breakdown dominates the whole gate dielectric breakdown, while under low voltage stress, interfacial layer between HfO2 bulk layer and silicon substrate does; 4) The charge trapping characteristics are obtained using carrier separation experiments. It is found that under high voltage stress, hole trapping in HfO2 bulk layer is dominant; while under low voltage stress, electron trapping in interfacial layer does dominate; 5) Based on the electric field dependent TDDB and charge trapping characteristics, and through analyzing the energy band diagram of HfO2 dielectric, a new breakdown model of Metal/High K gate stack was proposed. Generally, it is firstly reported that the TDDB characteristics of ultra-thin HfO2 gate dielectrics depends on polarity as well as electric field strength of stresses, so conventional lifetime prediction isn't adoptable any more to ultra-thin high K gate dielectrics. The achievements in this research, however, will give a meaningful help to develop a correct lifetime prediction method.