Optical Quality-Assessment of High-Order One-Dimensional Silicon Photonic Crystals With a Reflectivity Notch at $\lambda \sim 1.55\ \mu\hbox{m}$

Abstract

In this paper, high-order silicon/air vertical 1-D photonic crystals (PhCs) featuring a deep reflectivity notch (Q of about 3000) within the third telecommunication transmission window are fabricated by electrochemical micromachining of silicon and optically characterized. Although, from scanning electron microscope (SEM) investigations, the microfabricated structures appear to be of high quality in terms of in-plane and out-of-plane uniformity as well as of low surface roughness, optical measurements highlight variability of the spectral position of reflectivity notches, which turns out to be affected by fine variations of structural features induced by fabrication. A nondestructive optical method is used to recover the distribution of typical fabrication features, which is not easily quantifiable with other techniques in these high aspect-ratio vertical structures, from the experimentally detected distribution of the optical features. Optical distribution is obtained by scanning the device with high spatial resolution and recording spectral reflectivity at normal incidence (in the wavelength range 1.0-1.7 μm) on different locations of the 1-D PhC surface (X-Y plane). Distribution of silicon thickness variation on the X-Y plane is then inferred by best fitting the experimental spectral reflectivity data with theoretical spectra that are calculated by taking into account nonidealities of both PhC structure and measuring setup. On-sample fabrication-induced variation of the average silicon thickness, as a function of the X-Y position, is estimated to be limited to ±80 nm (percentage error of ±1%) within a range of ±100 μm from a reference location. As a consequence of such thickness variation, a maximum shift of the reflectivity spectrum of 40 nm occurs as a function of the X-Y position in the range of ±100 μm. Sample-to-sample average silicon thickness variation among nominally equivalent structures is also tested in similar X-Y positions and found to be lower than ±1.5% (absolute variation of ±120 nm), within a range of ±15 μm, with respect to the average value on the investigated samples.