In situ Thermal Trimming of Waveguides in a Standard Active Silicon Photonics Platform

TL;DR

This paper introduces suspended heater structures in silicon photonics that enable in situ thermal trimming of waveguides, allowing precise control of their effective refractive index with moderate power, and demonstrates potential for device tuning and aging analysis.

Contribution

It presents a novel in situ thermal trimming method using suspended heaters in a standard silicon photonics platform, enabling bi-directional index tuning and device bias point trimming.

Findings

Achieved effective index changes of -5.18e-3 (SiN) and -7.9e-4 (Si) at 1550 nm.

Demonstrated bi-directional index trimming up to 0.02 in SiN waveguides.

Estimated long-term aging behavior of the heaters over five years.

Abstract

We present suspended heater structures fabricated in a standard C- and O-band silicon (Si) photonics platform that can achieve sufficiently high local temperatures to induce effective refractive index trimming of Si and silicon nitride (SiN) waveguides with 30 - 90 mW of applied electrical power. Following thermal trimming at moderate powers ($\leq$ 40 mW), maximum changes in the averaged waveguide effective refractive index of $-5.18 \times 10^{-3}$ and $-7.9 \times 10^{-4}$ are demonstrated in SiN and Si waveguides, respectively, at a wavelength of 1550 nm. At higher powers, SiN waveguides exhibit positive averaged effective index changes up to $\approx$0.02, demonstrating bi-directional index trimming. As an example application, we demonstrate bias point trimming of a carrier injection Mach-Zehnder switch. Through investigations of the origin of the thermal trimming effect, we hypothesize that changes in the silica (SiO$_2$) waveguide cladding may be a primary underlying mechanism at temperatures $\gtrsim$300$^{\circ}$C, with significant trimming of SiN waveguide cores occurring at larger temperatures $\gtrsim$510$^{\circ}$C. As the trimming experiments represent a form of accelerated thermal aging, we estimate the aging behavior of the suspended heaters by fitting and extrapolating the measured datasets to 100 - 200$^{\circ}$C operating temperatures over five years.