Quantitative metamaterial property extraction

TL;DR

This paper introduces a new, precise extraction model for metamaterials that accurately represents S parameter data across broad frequencies using Lorentzian resonance regions, outperforming previous models in Bayesian correctness likelihood.

Contribution

The paper presents a novel extraction model for metamaterials that provides a causal, quantitative fit to S parameters over a wide frequency range, incorporating Lorentzian resonance dispersion.

Findings

The new model achieves higher correctness likelihood in Bayesian analysis.

It accurately captures resonance dispersion as Lorentzian and causal.

Outperforms previous models in fitting S parameter data.

Abstract

We examine an extraction model for metamaterials, not previously reported, that gives precise, quantitative and causal representation of S parameter data over a broad frequency range, up to frequencies where the free space wavelength is only a modest factor larger than the unit cell dimension. The model is comprised of superposed, slab shaped response regions of finite thickness, one for each observed resonance. The resonance dispersion is Lorentzian and thus strictly causal. This new model is compared with previous models for correctness likelihood, including an appropriate Occam's factor for each fit parameter. We find that this new model is by far the most likely to be correct in a Bayesian analysis of model fits to S parameter simulation data for several classic metamaterial unit cells.