Full-Band Particle Based Simulation for Three Dimensional Device Structures

随着未来半导体器件技术向0.1 Gm及以下的方向发展，对半导体器件模拟工具提出了新的要求，包括所采用的物理模型和提高物理精度的计算要求。与此同时，随着器件技术缩小到这一关键特征尺寸以下，半导体制造成本呈指数级增长，这要求在制造之前增加对模拟的依赖。随着器件的不断缩小，将出现的重要问题包括全三维几何效应、新材料系统、随机掺杂效应、离散电子充电效应，以及最终的最小维度的量子力学效应。在此，建议为一项为期三年的研究计划提供资金，目标是开发必要的器件模拟工具，以应对半导体工业协会(SIA)规划的路线图及更远的器件扩展的挑战，并特别关注模块化、健壮性和可靠性的要求。这些设备工具将使用全频段元胞自动机和蒙特卡罗粒子技术，这些技术是在以前的NSF资助下开发的，用于有效地求解半经典玻尔兹曼输运方程及其他方程。这些技术将与基于多重网格和双共轭梯度稳定方法的稳健场解算器相结合，包括任意二维和三维设备几何形状的非均匀网格。离散的杂质效应和载流子间的相互作用将通过网格/粒子力耦合模型包含在该模拟级别中，以评估基于随机杂质分布的器件工作特性的波动。将与工业合作伙伴合作，与最先进的设备技术进行比较和校准。由于三维半经典建模的计算需求，基于多处理器系统和工作站集群的高性能计算环境的研究将支持所提出的研究。交通模拟工具和耦合场解算器的分布式算法将被开发并应用于高端计算，正如之前的基金所开发的那样。我们将与工业界和实验团体合作，专注于将模拟工具与三个特定技术领域进行比较和校准，尽管该项目的范围远远超出这些领域。其中一个领域涉及EEPROM器件技术，其中3D效果很普遍，而且升高的电场也需要考虑全频段。另一项努力将集中在0.1gm栅长以下的大规模硅MOS器件上，其中在器件缩放方面出现了各种新的问题。最后，我们将考虑较新的材料系统，如SOI和Si/SiGe技术，这些系统对空穴传输知之甚少，需要改进包括全波段效应在内的传输模型。

9976484GoodnickScaling of future semiconductor device technologies towards 0.1gm and below is placing new demands on semiconductor device simulation tools in terms of the physical models employed, and the computational demands of increased physical accuracy. At the same time, the exponential increase of semiconductor manufacture costs as device technology shrinks below this critical feature size mandates an increased dependence on simulation prior to manufacture. Important issues which will occur as devices continue to shrink include full three-dimensional geometry effects, new material systems, random dopant effects, discrete electron charging effects, and ultimately quantum mechanical effects at the smallest dimensions.Herein is proposed funding for a three year program of research with the goal of developing the necessary device simulation tools to meet the challenges of device scaling along the projected Semiconductor Industry Association (SIA) roadmap and beyond with special attention given to the requirements of modularity, robustness and reliability. These device tools will employ full-band Cellular Automata and Monte Carlo particle-based techniques developed under previous NSF funding for efficient solution of the semi-classical Boltzmann transport equation and beyond. These techniques will be combined with robust field solvers based on multi-grid and Bi-conjugate gradient stabilized methods including non-uniform grids for arbitrary two- and three-dimensional device geometries. Discrete impurity effects and intercarrier interactions will be included in this simulation level through a coupled mesh/particle force model to assess the fluctuation in device operating characteristics based on random impurity distributions. Collaboration with industrial partners will be undertaken for comparison and calibration with state of the art device technologies. Due to the computational demands of 3D semi-classical modeling, the proposed research will be supported by investigation of high-performance computing environments based on multi-processor systems, and clusters of workstations. Distributed algorithms for both the transport simulation tools and the coupled field solvers will be developed and applied for high-end computing as developed under previous funding.We will focus on comparison and calibration of the simulation tools with three particular technology areas in collaboration with industry and experimental groups, although the scope of the project goes far beyond these. One area involves EEPROM device technology, where 3D effects are prevalent, and where elevated electric fields necessitate full-band consideration as well. Another effort will focus on scaled Si MOS devices below 0.1 gm gate length, where a variety of new problems arise in terms of device scaling. Finally, we will consider newer material systems such as SOI and Si/SiGe technology, where little is known for example about hole transport, and improved transport models including full-band effects' are necessary.***