• PDF / 441,010 Bytes
• 7 Pages / 612 x 792 pts (letter) Page_size
• 24 Downloads / 238 Views

DOWNLOAD

REPORT

---

C3.1.1

Process Modeling For Advanced Devices Mark E. Law1, Kevin S. Jones2, Ljubo Radic1, Robert Crosby2, Mark Clark2, Kevin Gable2, Carrie Ross2 1 Department of Electrical and Computer Engineering 2 Department of Materials Science and Engineering University of Florida, Gainesville, FL 32608 USA INTRODUCTION – YESTERDAY’S TECHNOLOGY MODELED TOMORROW As device lots become more and more expensive, process modeling is increasingly important. Process simulation and modeling is increasingly sophisticated but the accuracy remains a problem. There is generally a time lag between the introduction of a particular process and its accurate modeling – the problem of “yesterday’s technology modeled tomorrow”. For many problems, absolute accuracy isn’t required. Relative trends provide excellent information about the process in question. For this reason, process simulation is still a useful technique for guiding process development. Over the last twenty years, device scaling has been relatively straightforward. This is no longer the case as a wide variety of process options are being discussed. In junction technologies alone, new doping schemes beyond implantation are being considered including plasma immersion and grown junctions. Anneals are moving to the millisecond and shorter time frames or are being done at low temperature with a phase change. These are combined with new materials (SiGe) and compounds that complicate activation and diffusion. Engineers are beginning to intentionally incorporate strain into their device designs to optimize the mobility and transistor response. All of these represent difficult challenges for process modeling tools. These new processes will require the rethinking of assumptions and retooling of software packages. As the pace of innovation in materials and processes quickens, it will be increasingly hard for process modeling tools to remain abreast of the changes. However, we do not see this as a necessarily bad thing. First, the only way process modeling will catch up is if the process technologists stop innovating. This implies the end of Moore’s Law – certainly a bad thing for the semiconductor industry. Second, process modeling doesn’t have to be perfectly accurate to be useful. It has to provide insight during design, optimization guidelines during implementation to manufacturing, and debug during large-scale manufacturing. All of these activities require different levels of accuracy and model sophistication. CHANGES IN PROCESSES – LAW’S LAW Current MOSFET technology is moving toward strained devices to optimize the mobility. Ultra-shallow junctions are required to control short channel effects and the deep source drain needs to be as heavily doped as possible to minimize contact resistance. Although we are moving toward a diffusion-less, meta-stable anneal point defects can still play a critical role in determining the activation of the dopant. Surface materials are changing (both silicide and gate dielectric) with their potential to change point defect kinetics. Ultra-fast ramps rely on