next up previous contents
Next: Acknowledgment Up: Dissertation Robert Wittmann Previous: Kurzfassung

Abstract


The MOS transistor performance depends on both the channel length and the carrier transport properties of the channel material. Novel materials with higher carrier mobilities beyond the achievable limits of process-induced strained silicon are required to facilitate continued commensurate device scaling. Nitrogen is incorporated into ultra-thin gate oxides to eliminate parasitic effects. In this thesis, doping profiles in high-mobility materials are investigated and the NBTI reliability is analyzed for a CMOS technology with nitrided gate oxide.

Ion implantation will continue to be the primary technology for introducing dopant atoms into semiconductor wafers to form devices and integrated circuits (ICs). The reason for this lies in the flexibility of this doping technique in selection of dopant species, spatial location within the device, and in the accuracy of the number of implanted dopant atoms. The continued reduction in junction depths and lateral dimensions of the MOS transistor has directly resulted in growth of ion implantation applications in CMOS process technology, e.g. halo doping profiles have become necessary to suppress the short-channel effect. The physical ion implantation process can be effectively modeled on computers. Accurate simulation capabilities allow to optimize the doping profiles and to reduce the development time for a new CMOS technology.

The investigation of doping profiles for advanced CMOS applications has been carried out with a Monte Carlo ion implantation simulator, developed in the scope of several PhD theses. The three-dimensional simulator MCIMPL-II is based on the physical BCA approach and uses the universal ZBL potential. An empirical model is used for the electronic stopping of ions and the generated point defects are calculated by a modified Kinchin-Pease model. As part of this work, the simulator has been improved and extended from crystalline silicon to advanced target materials on the basis of experimental results. Boron and arsenic were implanted into biaxially strained silicon, SiGe layers of different composition, and germanium within the energy range from about 1keV to 60keV. The successful calibration of the simulator for these materials is demonstrated by comparison of predicted doping profiles with SIMS measurements. The Monte Carlo simulation of ion trajectories is useful to analyze the impact of material properties and physical effects like damage accumulation and channeling on the profiles. The main results of the investigation are a shift to shallower profiles with increasing Ge content in SiGe alloys, the produced point defects are significantly reduced in germanium compared to silicon, and the stress-induced volume dilation in strained silicon has almost no influence on the profiles.

As transistors get smaller in each successive CMOS generation, ensuring their reliability becomes increasingly difficult. The negative bias temperature instability (NBTI) effect in p-MOSFETs based on nitrided gate oxides has emerged as the dominant degradation mechanism for advanced CMOS technologies. As another part of this work, an experimental and simulation study for NBTI induced device parameter degradation was performed for a 90nm technology.

Finally, the improved Monte Carlo ion implantation simulator is applied for the calculation of dopant distributions in topological complex three-dimensional structures, composed of strained or relaxed enhanced mobility materials. The selected examples demonstrate the capabilities of the simulation tool to facilitate the processing of advanced devices. Furthermore, the impact of storing random bit sequences on the NBTI lifetime of an SRAM cell is analyzed by numerical simulations. All presented applications were requested or inspired by industry.


next up previous contents
Next: Acknowledgment Up: Dissertation Robert Wittmann Previous: Kurzfassung

R. Wittmann: Miniaturization Problems in CMOS Technology: Investigation of Doping Profiles and Reliability