Field solver technologies for variation-aware interconnect parasitic extraction
Author(s)
El-Moselhy, Tarek Ali
DownloadFull printable version (23.41Mb)
Other Contributors
Massachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science.
Advisor
Luca Daniel.
Terms of use
Metadata
Show full item recordAbstract
Advances in integrated circuit manufacturing technologies have enabled high density onchip integration by constantly scaling down the device and interconnect feature size. As a consequence of the ongoing technology scaling (from 45nm to 32nm, 22nm and beyond), geometrical variabilities induced by the uncertainties in the manufacturing processes are becoming more significant. Indeed, the dimensions and shapes of the manufactured devices and interconnect structures may vary by up to 40% from their design intent. The effect of such variabilities on the electrical characteristics of both devices and interconnects must be accurately evaluated and accounted for during the design phase. In the last few years, there have been several attempts to develop variation-aware extraction algorithms, i.e. algorithms that evaluate the effect of geometrical variabilities on the electrical characteristics of devices and interconnects. However, most algorithms remain computationally very expensive. In this thesis the focus is on variation-aware interconnect parasitic extraction. In the first part of the thesis several discretization-based variation-aware solver techniques are developed. The first technique is a stochastic model reduction algorithm (SMOR) The SMOR guarantees that the statistical moments computed from the reduced model are the same as those of the full model. The SMOR works best for problems in which the desired electrical property is contained in an easily defined subspace. (cont.) The second technique is the combined Neumann Hermite expansion (CNHE). The CNHE combines the advantages of both the standard Neumann expansion and the standard stochastic Galerkin method to produce a very efficient extraction algorithm. The CNHE works best in problems for which the desired electrical property (e.g. impedance) is accurately expanded in terms of a low order multivariate Hermite expansion. The third technique is the stochastic dominant singular vectors method (SDSV). The SDSV uses stochastic optimization in order to sequentially determine an optimal reduced subspace, in which the solution can be accurately represented. The SDSV works best for large dimensional problems, since its complexity is almost independent of the size of the parameter space. In the second part of the thesis, several novel discretization-free variation aware extraction techniques for both resistance and capacitance extraction are developed. First we present a variation-aware floating random walk (FRW) to extract the capacitance/resistance in the presence of non-topological (edge-defined) variations. The complexity of such algorithm is almost independent of the number of varying parameters. Then we introduce the Hierarchical FRW to extract the capacitance/resistance of a very large number of topologically different structures, which are all constructed from the same set of building blocks. The complexity of such algorithm is almost independent of the total number of structures. All the proposed techniques are applied to a variety of examples, showing orders of magnitude reduction in the computational time compared to the standard approaches. In addition, we solve very large dimensional examples that are intractable when using standard approaches.
Description
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2010. Cataloged from PDF version of thesis. Includes bibliographical references (p. 207-213).
Date issued
2010Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer SciencePublisher
Massachusetts Institute of Technology
Keywords
Electrical Engineering and Computer Science.