CAREER: Design Methodologies for Three-Dimensional Integrated Circuits

摘要：40多年来，半导体工业在缩小集成电路上的晶体管方面取得了成功，现在我们的尺寸已经达到了几十纳米，而且市场每年都期待着新的、更快的芯片。尽管量子力学的收缩极限尚未达到，但制造设备的成本正变得令人望而却步，行业专家正在努力寻找方法来证明这一成本是合理的。一个有希望的方法是垂直堆叠芯片，从而满足市场需求，并重新利用现有设备来销售更多的芯片。这种堆叠过程，通常被称为三维集成电路（3D ic），已经在原型电路中被证明可以成功地增加芯片上的晶体管数量，但尚未被证明可以显著提高速度。理论上，3D集成电路可以将电路的速度提高2到5倍，具体取决于应用。这项工作的目标是通过实验发现和记录设计技术，这些技术将从三个方面提高3D ic的速度：改进内存设计，改进热流分析，以及寻找特别适合3D ic的应用。除了科学发现之外，该项目还通过开发最新集成电路技术的免费设计套件来改善工程教育。自1998年以来开发的所有设计工具包都有严格的知识产权控制，限制了教育工作者使用和扩展它们的能力。这个子项目被称为FreePDK项目，将建立在最新、最成功的免费设计工具包的基础上，通过全面的文献搜索和与工业界的合作，为教育和小企业提供急需的推动力。最后，与当地社区中心和国家教育基金会的外展伙伴关系将促进贫困学生对低成本电子产品的兴趣。制作水晶收音机和无线控制机器人的讲习班课程将在全国范围内分发，每年夏天由40名初高中学生参加测试。

CAREER: Design Methodologies for Three-Dimensional Integrated CircuitsPI: W. Rhett Davis, INST: North Carolina State University, Raleigh NC 27695-7911Proposal no: 0643700ABSTRACT:The semiconductor industry has succeeded in shrinking transistors on integrated circuits for more than 40 years, giving us dimensions now in the tens of nanometers and a market that expects new, faster chips every year. Although quantum-mechanical limits of shrinking have not yet been reached, the cost of manufacturing equipment is becoming prohibitively large, and industry experts are struggling to find ways to justify this cost. One promising approach is to stack chips vertically, thereby meeting the market's demands and re-using existing equipment to sell more chips. This process of stacking, often called three-dimensional integrated circuits (3D ICs), has been shown in prototype circuits to successfully increase the number of transistors on a chip, but it has not yet been shown to give significant improvements in speed. Theoretically, 3D ICs can increase the speed of circuits by a factor of two to five, depending on the application. The goal of this work is to experimentally discover and document the design techniques that will enable these speed improvements for 3D ICs in three ways: improved memory design, improved analysis of heat flow, and a search for applications that are especially well suited for 3D ICs. In addition to scientific discoveries, this project seeks to improve engineering education by developing a free design-kit for the latest integrated-circuit technology. All design-kits developed since 1998 have strict intellectual property controls, limiting educators' ability to use and expand them. This sub-project, called the FreePDK project, will build on the latest, most successful free design-kit with a thorough literature search and collaboration with industry, providing a much-needed boost to education and small businesses.Finally, an outreach partnership with a local community center and national education foundation will promote interest in low-cost electronics among disadvantaged students. A nationally distributed workshop curriculum to build crystal radios and a wirelessly-controlled robot will be developed tested each summer with forty middle-school and high-school students.