Uncertainty Quantification and Variance-Based Sensitivity Analysis for Microstrip Line on IC-Substrate

In this paper, we present numerical uncertainty quantification using polynomial chaos expansion and variance-based global sensitivity analysis of a microstrip fabricated on IC-substrates. Microstrip design parameters such as line width, trace height, substrate thickness, relative dielectric constant, and loss tangent are assumed to have Gaussian probability distribution depending on IC-substrate manufacturing processes. Those input parameter variations are represented by orthogonal polynomials and their coefficients. The impact of the varied design input parameters on the electrical characteristics such as a characteristic impedance and propagation constant is quantitatively analyzed. For computational efficiency, non-intrusive spectral projection method with Smolyak sparse grid technique is employed to mitigate the numerical multi-dimensional integration. Variance-based sensitivity analysis using Sobol indices is also performed for the investigation of significant input parameters on each output parameter at several frequencies.