Please use this identifier to cite or link to this item:
|Title:||Optimizing the integration and energy efficiency of through silicon via-based 3D interconnects|
|Abstract:||The aggressive scaling of CMOS process technology has been driving the rapid growth of the semiconductor industry for more than three decades. In recent years, the performance gains enabled by CMOS scaling have been increasingly challenged by highlyparasitic on-chip interconnects as wire parasitics do not scale at the same pace. Emerging 3D integration technologies based on vertical through-silicon vias (TSVs) promise a solution to the interconnect performance bottleneck, along with reduced fabrication cost and heterogeneous integration. As TSVs are a relatively recent interconnect technology, innovative test structures are required to evaluate and optimise the process, as well as extract parameters for the generation of design rules and models. From the circuit designer’s perspective, critical TSV characteristics are its parasitic capacitance, and thermomechanical stress distribution. This work proposes new test structures for extracting these characteristics. The structures were fabricated on a 65nm 3D process and used for the evaluation of that technology. Furthermore, as TSVs are implemented in large, densely interconnected 3D-system-on-chips (SoCs), the TSV parasitic capacitance may become an important source of energy dissipation. Typical low-power techniques based on voltage scaling can be used, though this represents a technical challenge in modern technology nodes. In this work, a novel TSV interconnection scheme is proposed based on reversible computing, which shows frequencydependent energy dissipation. The scheme is analysed using theoretical modelling, while a demonstrator IC was designed based on the developed theory and fabricated on a 130nm 3D process.|
|Appears in Collections:||School of Electrical, Electronic and Computer Engineering|
Files in This Item:
|Asimakopoulos 11.pdf||Thesis||16.18 MB||Adobe PDF||View/Open|
|dspacelicence.pdf||Licence||43.82 kB||Adobe PDF||View/Open|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.