GOALI: A Truly CMOS Compliant On-Chip Optical Interconnect System for VLSI Chips

本文提出的研究计划旨在展示一个纯片上、多ghz、光学互连系统，该系统采用商业硅代工厂的深亚微米CMOS技术实现。关键的管理理念是严格限制光学互连组件技术，只要可能，在传统的最先进的集成电路中发现的材料和工艺。为了解决电气互连问题，CMOS材料技术已经扩展到包括低k电介质，新冶金和铜封装薄膜。我们建议使用这些工具来实现光互连。具体来说，光学互连将基于由氮化硅和低k介电体组成的波导，这些介质已经作为CMOS互连介电层和由标准CMOS电阻多晶硅薄膜制成的肖特基势垒接触多晶硅光电二极管而存在。使用多晶硅光电探测器而不是晶体光电探测器使我们能够将光电探测器从高性能数字VLSI芯片中典型的噪声衬底环境移到噪声较小且控制良好的后端层。本实验室对基于非晶硅的MSM光电二极管进行了初步研究。mosfet本身性能的提高将用于实现多ghz接收器。唯一不能在CMOS中实现的组件是高速，高功率的光源，它将使用单边缘发射激光二极管。所有与CMOS相关的晶圆制造将由我们的工业合作伙伴安捷伦技术公司提供。此外，安捷伦科技在ASIC和光学IC设计方面拥有丰富的专业知识，也将与我们合作评估概念并改进设计。这种独特的关系是确保所提出技术的实用性的关键。因此，本研究的智力优势在于，该方法专门利用现代硅CMOS技术中的材料和加工步骤来嵌入片上光学互连结构。据我们所知，这是第一次尝试开发、设计和演示一个完整的集成系统，它将为硅技术的进一步发展提供有价值的数据。拟议研究的影响是三重的。首先，它为沿着摩尔定律的技术进步提供了必要的动力，远远超出了ITRS预测的路径。拟议的研究旨在解决方案，可以成为未来5-10年的使能技术之一。其次，它为学术和产业合作与参与提供了一个平衡的论坛。这种密切的合作确保了潜在技术的实用性。此外，除了常规出版渠道外，研究成果还可以通过我们的行业合作伙伴更快、更有说服力地传播。最后，落基山脉地区在半导体和光学工程方面拥有高度集中的研发设施。这两个学科之间的交叉授粉对于孵化新技术至关重要。建议的研究将允许我们创建跨学科的课程和研讨会。我们还建议举办第一次完全符合CMOS的片上光互连国际研讨会，聚集该领域的研究人员进一步交流思想和讨论问题。利用我们的研究成果，我们将在其他相关会议上提供指导，如ISCAS和ICCC。此外，我们将邀请一名高中科学教师参与项目，在当地一所高中创建一个“科学博览会”项目。该项目将有多达4名高中生参与波导材料的调试和测量。学生们将利用该项目的成果参加科罗拉多科学博览会的竞赛。

The research program proposed here is designed to demonstrate a purely on-chip, multi-GHz, optical interconnect system implemented in deep submicron CMOS technology from commercial silicon foundries. The critical governing philosophy is to strictly limit the optical interconnect component technology wherever possible to materials and processes found in conventional state- of-the-art integrated circuits. In an effort to address electrical interconnect issues, CMOS materials technology has expanded to include low-k dielectrics, new metallurgy, and copper encapsulation films. We propose to use these very tools to implement optical interconnects. Specifically, the optical interconnect will be based on waveguides composed of silicon nitride and low-k dielectrics already present as CMOS interconnect dielectric layers and Schottky barrier contacted polysilicon photodiodes made from standard CMOS resistor polysilicon films. The use of polysilicon photodetectors rather than crystalline photodetectors allows us to move the photodetectors out of the noisy substrate environment that is typical in high performance digital VLSI chips to less noisy and more well controlled back-end layers. Preliminary work has been performed in our laboratory on MSM photodiodes based on amorphous silicon. The increased performance of the MOSFETs themselves will be leveraged to implement multi-GHz receivers. The only component that cannot be implemented in CMOS is a high speed, high power optical source for which a single edge emitting laser diode will be used. All CMOS related wafer manufacturing will be provided by our industrial partner, Agilent Technologies. In addition, Agilent Technologies, with their extensive expertise in ASIC and optical IC designs, will also collaborate with us to evaluate concepts and improve designs. This unique relationship is a key to ensure the practicality of the proposed technology. Therefore, the intellectual merit of the proposed research lies in the method that makes exclusive use of materials and processing steps in modern silicon CMOS technologies to embed on-chip optical interconnect structures. This is, to our best knowledge, the first attempt to develop, design, and demonstrate a complete integrative system that will provide valuable data for further advances in silicon technology.The impact of the proposed research is threefold. Firstly, it provides necessary impetus for technology advances along the Moore's Law well beyond the path that is projected by ITRS. The proposed research is aimed at solutions that can become one of the enabling technologies for the next 5-10 years. Secondly, it provides a well-balanced forum for academic and industrial collaboration and participation. Such close collaboration ensures practicality of potential technologies. Furthermore, research results can be disseminated faster and more convincingly through our industrial partners in addition to regular publication channels. Lastly, the Rocky Mountain region has a high concentration of R&D facilities in semiconductor and optical engineering. Cross-pollinations between these two disciplines are crucial for incubating new technologies.The proposed research will allow us to create cross-disciplinary curriculum and seminar. We also propose to organize a first international workshop on fully CMOS compliant on-chip optical interconnects to assemble researchers in this field to further exchange ideas and discuss issues. Using our research results, we will offer tutorials at other related conferences, such as ISCAS and ICCC. Furthermore, we will bring a high-school science teacher into the project to create a "Science Fair" project in a local high school. The project will be involved with up to 4 high school students participating in debugging and measurement wave guide materials. The students will compete in Colorado Science Fair competition using the results from the project.