Ultrahigh-Q Torsional Nanomechanics through Bayesian Optimization

TL;DR

This paper uses Bayesian optimization to design nanoribbons with ultrahigh quality factors, achieving exceptional dissipation dilution and sensitivity for torque sensing and quantum experiments at room temperature.

Contribution

It introduces a Bayesian optimization approach to maximize dissipation dilution in nanoribbons, leading to record-high Q factors and enhanced sensing capabilities.

Findings

Q factors exceeding 100 million at room temperature

Q-frequency product surpassing 10^13 Hz

Thermal torque sensitivity around 10^-20 N·m/√Hz

Abstract

Recently it was discovered that torsion modes of strained nanoribbons exhibit dissipation dilution, giving a route to enhanced torque sensing and quantum optomechanics experiments. As with all strained nanomechanical resonators, an important limitation is bending loss due to mode curvature at the clamps. Here we use Bayesian optimization to design nanoribbons with optimal dissipation dilution of the fundamental torsion mode. Applied to centimeter-scale Si$_3$N$_4$ nanoribbons, we realize $Q$ factors exceeding 100 million and $Q$-frequency products exceeding $10^{13}$ Hz at room temperature. The thermal torque sensitivity of the reported devices is at the level of $10^{-20}\;\text{N}\,\text{m}/\sqrt{\text{Hz}}$ and the zero point angular displacement spectral density is at the level of $10^{-10}\;\text{rad}/\sqrt{\text{Hz}}$; they are moreover simple to fabricate, have high thermal conductivity, and can be heavily mass-loaded without diminishing their $Q$, making them attractive for diverse fundamental and applied weak force sensing tasks.