Experimental Realization of Optimized Ternary Mirror Coatings

Technical Summary: Experimental Realization of Optimized Ternary Mirror Coatings

Problem Statement
Advanced optical applications, particularly gravitational-wave interferometers and ultra-stable lasers, face a persistent challenge in reducing Brownian thermal noise while maintaining low optical absorption and high reflectivity. Traditional binary dielectric stacks are limited by the trade-off between these parameters; identifying materials that simultaneously offer low noise, low absorption, and a broad range of refractive indices is difficult. While multi-material coatings have been proposed theoretically to bury absorptive, low-noise layers beneath noisy, low-absorption layers, previous experimental realizations have been limited to proof-of-concept demonstrations lacking a fully predictive design-to-fabrication pipeline.

Methodology
The authors developed a design framework based on a "Double-Stack of Doublets" (DSD) architecture, optimized via a multi-objective evolutionary algorithm (Borg MOEA). This approach seeks Pareto-optimal solutions that simultaneously minimize thermal noise, optical absorption, and transmittance without subjective weighting of objectives.

The design strategy involves two distinct sub-stacks:

1. Bottom Stack: Utilizes high refractive index contrast materials (SiNx/SiO2 or Ti:GeO2/SiO2) to achieve high reflectivity with fewer layers.
2. Top Stack: Composed of low-absorption materials (Ti:Ta2O5/SiO2 or Ta2O5-TiO2/SiO2) to shield the more absorptive bottom layers from the incident laser field.

Layer thicknesses are determined not by standard quarter-wave rules but by the optimization algorithm to balance competing objectives. The thermal noise is modeled using the Braginsky model, weighted by layer thickness and material loss angles. The study focuses on two specific material systems:

• SiNx-based: A proof-of-concept using silicon-rich silicon nitride.
• Ti:GeO2-based: An advanced system using titanium-doped germania, targeting lower optical losses.

Fabrication was performed at the Laboratoire des Matériaux Avancés (LMA) using ion-beam sputtering (IBS). The SiNx design required a split-deposition process between an R&D system and the main "Grand Coater," necessitating a split-annealing protocol to prevent crystallization in the top stack. The Ti:GeO2 design utilized a multi-step annealing process to achieve sub-ppm absorption.

Key Contributions and Results
The paper reports the first experimental realization of two distinct multi-material coatings designed through this fully predictive optimization framework.

• SiNx-based Coating:

  • Performance: The measured amplitude spectral density (ASD) reduction factor was 0.82 relative to current LIGO/Virgo End Test Mass (ETM) coatings, with a thermal noise of $10.6 \times 10^{-18} \text{ m}/\sqrt{\text{Hz}}$.
  • Agreement: The experimental results showed remarkable agreement with theoretical predictions, validating the design-to-fabrication pipeline.
  • Absorption: Measured absorption was approximately 1.5 ppm, meeting the design target.

• Ti:GeO2-based Coating:

  • Absorption: Successfully achieved sub-ppm absorption (0.7 ppm), surpassing the Carniglia limit for the binary system and meeting the ambitious 0.5 ppm target.
  • Thermal Noise: The measured thermal noise (ASD RF of 0.81) was higher than the theoretically predicted level of 0.71.
  • Discrepancy Analysis: A dedicated tolerance analysis confirmed that random thickness errors (±0.5 nm) and refractive index uncertainties could not account for the performance gap. The authors conclude the discrepancy likely stems from unforeseen material incompatibilities or process-induced stress not captured by single-layer characterizations, rather than stochastic fabrication errors.

Significance and Claims
The authors assert that this work establishes a robust methodology for producing complex, high-performance optical coatings tailored for precision experiments. The successful fabrication of the SiNx coating serves as a crucial proof-of-concept, demonstrating that even materials with relatively high absorption can be integrated into low-noise designs via the DSD architecture.

While the Ti:GeO2 system did not fully meet its thermal noise target, its achievement of sub-ppm absorption validates the potential of the advanced material combination. The paper emphasizes that while material selection sets the theoretical performance limit, mastering the interplay between materials during co-deposition and annealing is paramount to fully exploiting these limits. The work provides a reliable pathway for creating bespoke optical coatings with properties previously inaccessible, moving beyond empirical demonstrations to a predictive, optimization-driven paradigm.