Energy efficient sub-terahertz electrical interconnect
Author(s)
Holloway, Jack Wade,1980-
Download1252061479-MIT.pdf (64.21Mb)
Other Contributors
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science.
Advisor
Ruonan.
Terms of use
Metadata
Show full item recordAbstract
With the end of Moore's Law and Dennard scaling in silicon platforms, coupled with the increase in computational demand across applications, the semiconductor industry has seen a move towards high-density compute leveraging multiple dies in package. These types of products have been partially enabled by short-reach, energy-efficient, high-speed interconnect in package. Big data and AI/ML applications have pushed the development of longer-reach, high-capacity, and energy efficient interconnect enabling connectivity between racks across large data centers. This work investigates and demonstrates a new interconnect technology that fills a meter-class interconnect gap in these applications. By leveraging the wide transmission bandwidth and low-losses associated with dielectric waveguides in the sub-THz regime (100 GHz - 1 THz), large baseband data rates are aggregated across multiple channels, multiplexed on to a single electrical channel, efficiently coupled into a dielectric waveguide, and transmitted between chips. In this work, enabling component technologies are developed and demonstrated, including planar broadband couplers and high-performance sub-THz multiplexers operating in the 220-330 GHz WR-3.4 band -- both technologies designed to ease implementation and packaging costs. Lastly, an end-to-end link is realized in a 130nm Silicon Germanium BiCMOS process and is demonstrated utilizing a small cross-section polymer dielectric waveguide. The link achieves 105 Gbps in a 250 ̄ 400 [mu]m² waveguide cross section, demonstrating a state of the art 330 Gbps/mm figure of merit and better than 5 pJ/bit energy efficiency.
Description
Thesis: Ph. D., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, September, February, 2021 Cataloged from the official PDF of thesis. Includes bibliographical references (pages 175-184).
Date issued
2021Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer SciencePublisher
Massachusetts Institute of Technology
Keywords
Electrical Engineering and Computer Science.