On Thursday, March 8, 2018, Kevin Morot, PhD student in Optics and Radiofrequencies at theInstitute of Microelectronics Electromagnetism and Photonics and the Microwave and Characterization Laboratory (IMEP-LAHC), will defend his thesis "RF integration and modeling of 3D interconnects for photonic interposer". The defense will take place at 10:15 am, in the amphitheater of the Bâtiment Pôle Montagne on the Bourget-du-Lac campus.
thesis summary
For many years, microelectronics engineers have relied on Moore's Law, which predicts that integrated circuits will become denser and denser at a steady pace, to deliver faster, cheaper and more energy-efficient systems. Now, however, this law is running out of steam, and system manufacturers are forced to turn to disruptive solutions to keep up the performance race. 3D circuit integration has been developed to meet this challenge. In data centers, the nerve centers of communication networks, signal propagation is spread over optical and electrical networks. An essential component for conversion between these two transmission modes is the electro-optical transmitter. As transmission rates increase, the optical signal must be routed closer and closer to the electronic components to reduce cost and energy consumption. To reach the 10 Terabits per second transmission capacity targeted for 2020, the maximum distance covered by the electrical signal between the transmitter and the processor is estimated at 10 mm. This calls for a new approach to integrating the transmitter directly into the digital circuit package.
The photonic interposer is proposed to achieve this co-integration by benefiting from three-dimensional integration. The first stage of the work presented in this dissertation aims to assess the limits of the
of the photonic interposer for the targeted applications. To this end, we demonstrate that "middle" integration of the via through the silicon (i.e. TSV) is preferable to its "last" version. Design rules are then established to ensure that interconnections can be adapted. A methodology dedicated to extracting the maximum utilization length of chains satisfying bandwidth criteria is put in place. It is based on 3D electromagnetic simulation of each interconnection and cascading of the extracted models. Preferred routing paths are deduced, and the results obtained reveal strong routing constraints arising from the limitations of current technology.
Developing an interposer capable of meeting these specifications involves optimizing the performance of 3D interconnect chains. A wide-band, parameterizable model is developed for each 3D interconnect. These models are established on the basis of electromagnetic simulation plans, whose definition domains fall within the current technological context and include longer-term technological perspectives deemed promising. The models are validated by comparison with the results of peak characterizations up to 110 GHz. A first conclusion of the work identifies TSVs and RDL lines as the most penalizing elements of the 3D network. The parameters that most affect their attenuation and delay are identified: substrate conductivity and, to a lesser extent, oxide and conductor thicknesses. Finally, the parameterizable models are cascaded to optimize the maximum operating length and energy
energy consumption of 3D photonic interposer chains in two technological contexts: one constrained by current processes and the other including breakthrough solutions.
Technological, design and routing recommendations are drawn from this study. The benefits of thinning the substrate, thickening the RDL lines and using high-resistivity substrates are quantified. The footprint of the interposer front panel is reduced by routing fast signals through the TSV and RDL up to 3 mm for a standard substrate. In the case of highly resistive substrates, transmission across the entire surface of the interposer is made possible without signal regeneration.