Non-uniform Birefringence in Highly-reflective Substrate-Transferred GaAs/Al$_{0.92}$Ga$_{0.08}$As Coatings at 1064 nm

TL;DR

This work investigates non-uniform birefringence in substrate-transferred GaAs/Al$_{0.92}$Ga$_{0.08}$As coatings used as high-reflectivity mirrors at 1064 nm. A single-reflection birefringence-mapping method is developed to quantify spatial variations across coatings from 18 to 194 mm in diameter, yielding an average birefringence of $ψ = 1.09 ± 0.18$ mrad and non-uniformity typically near $0.1$ mrad, with larger edge effects. The epi-on-growth-wafer coating shows uniform birefringence, indicating the substrate-transfer process as the primary source of non-uniformity, which also correlates with bond-wave–induced strain and sub-nm surface height variations. When applied to a LIGO-like cavity, the non-uniform birefringence contributes negligible optical loss (well below 1 ppm per reflection, about two orders of magnitude below the A+ budget), supporting the viability of AlGaAs coatings while highlighting areas for process improvement to minimize residual non-uniformity.

Abstract

Using a custom-built scanning system, we generated maps of birefringence on reflection at $λ=1064$~nm from single-crystal GaAs/Al$_{0.92}$Ga$_{0.08}$As Bragg reflectors (henceforth ``AlGaAs coatings''). Ten coatings were bonded to fused silica substrates and one remained on the epitaxial growth wafer. The average phase difference on reflection between beams polarized along the fast and slow axes of the coating was found to be $ψ= 1.09 \pm 0.18$~mrad, consistent with values observed in high-finesse optical reference cavities using similar AlGaAs coatings. Scans of substrate-transferred coatings with diameters between 18 and 194 millimeters showed birefringence non-uniformity at a median level of $0.1$~mrad. A similar epitaxial multilayer that was not substrate transferred, but remained on the growth wafer, had by far the least birefringence non-uniformity of all mirrors tested at $0.02$~mrad. On the other hand, the average birefringence of the epi-on-wafer coating and substrate-transferred coatings was indistinguishable. Excluding non-uniformity found at the location of crystal and bonding defects, we conclude that the observed non-uniformity was imparted during the substrate transfer process, likely during bonding. Quantifying the impact on the scatter loss in a LIGO-like interferometer, we find that birefringence non-uniformity at the levels seen here is unlikely to have a significant impact on performance. Nonetheless, future efforts will focus on improved process control to minimize and ultimately eliminate the observed non-uniformity.

Figures (10)

• Figure 1: Simplified diagram of the setup for measuring birefringence on reflection. (Not shown: Power monitoring pick-off, collimating and focusing optics, steering mirrors, active beam-steering, and rastering $xy$-stage supporting the sample. The beam interrogates the sample at normal incidence. The orientation of the sample is indicated by the "flat" (chord) next to the aluminum reference mirror. This was varied, but was typically $\pm45^\circ$, or $\pm135^\circ$ w.r.t. the analyzer s-pol, as shown in the top view here.
• Figure 2: Average birefringence of a 93 mm diameter AlGaAs mirror on 10 mm thick silica (S/N: E1800006-02). The magnitude of the birefringence obtained from this data is $1.13\pm0.10$ mrad.
• Figure 3: A 100 mm$\times$10 mm sample (S/N: E1800006-03) mounted on the rastering stage.
• Figure 4: Birefringence measured in various cavity experiments compared to the average birefringence measured in our samples. The average birefringence and the RMS deviation from the mean is shown at the top of the figure.
• Figure 5: Birefringence variation in a 93 mm AlGaAs coating on 100 mm diameter by 10 mm thick fused silica substrate, S/N: E1800006-02. Fast axis of the average birefringence is at 45$^\circ$ to the analyzer.