Photonic beam combiner chip achieves sub-1% precision on sky at Mt. Wilson

A glass photonic integrated circuit measuring a few centimeters has recorded better than 1% statistical precision in raw squared fringe visibility on bright stars during on-sky tests at the CHARA Array on Mt. Wilson, California. The chip, tested during 2024 and 2025 inside the Chara ARay Integrated Optics Testbench (CHARIOT), combines light from separate telescopes in the astronomical K-band (~2 to 2.4 µm) — a first for this class of beam combiner, according to its developers.

"To our knowledge, it's the first on-sky demonstration of this type of beam combiner within the K-band," said Aurélien Benoit, a postdoctoral research associate at Heriot-Watt University in Scotland. The device, written with ultrafast laser pulses by Heriot-Watt photonics professor Robert Thomson and Benoit, replaced larger and more complex mirrors-and-lenses beam-combination optics in the instrument.

How the chip works

The PIC guides and manipulates light through buried optical waveguides inside bulk glass. "Bringing waveguides close together allows light to couple between them, which provides a way to split and combine optical signals," Benoit explained. Combining light from the same astronomical object collected by separate telescopes produces interference fringes; the fringe contrast encodes the object's angular structure, such as a star's apparent diameter. Because resolving power scales with telescope separation rather than aperture diameter, interferometry reaches detail far finer than any single telescope in the array can resolve.

The chip also diverts light into calibration outputs, which account for fluctuations in incoming brightness. The measured performance — sub-1% precision in raw squared visibility — applies to bright stars; Thomson and Benoit distinguished that figure carefully from absolute accuracy and from sensitivity on fainter targets.

Why the fabrication method matters

Three choices define the device: ultrafast laser inscription, an infrared-compatible glass, and fiber-connected packaging. "We use ultrafast laser inscription — tightly focused, extremely short laser pulses — to permanently modify the refractive index inside glass to create buried waveguides," Benoit said. "This approach gives us the freedom to route optical waveguides in three dimensions."

The substrate is Infrasil, a high-purity silica glass with very low hydroxyl content. Hydroxyl-related absorption introduces unwanted losses in the infrared, so the low-OH material preserves transmission across the K-band. The glass itself doubles as the mechanical structure holding the waveguides, and fiber connections make the device practical to install and exchange inside an astronomical instrument.

The work, published by A. N. Dinkelaker et al. in Monthly Notices of the Royal Astronomical Society (552, 2, stag1745), caps a research program that began with Thomson's U.K. Science and Technology Facilities Council Advanced Fellowship in 2010. "This is a milestone in research that spans more than 15 years, and it reflects a sustained collaboration between photonics researchers and astronomers," Thomson said. The team included Aline Dinkelaker of the Leibniz Institute for Astrophysics Potsdam plus researchers from Humboldt University Berlin, the University of Cologne, and Georgia State University.

What remains unsolved?

Sensitivity, calibration, and scaling are open problems. "We need to preserve as many photons as possible through the complete optical system, while controlling polarization, wavelength-dependent behavior, and background noise," Thomson said. Combining more telescope beams makes the waveguide layout and calibration correspondingly more demanding; the engineering goal, he noted, is to add capability without losing the stability and simplicity that make integration attractive in the first place.

Benoit emphasized that the chip alone does not deliver the result. "A promising chip is the starting point," he said. "Fiber interfaces, packaging, detectors, calibration, and data analysis all determine whether this promise produces useful performance." During the 2024–2025 runs, the PIC recorded signals from a wide range of stars, including targets fainter than the system it replaced could reach.

What comes next?

Immediate applications include measuring stellar diameters, investigating close binary systems, and detecting circumstellar dust emission. "Exozodiacal dust is particularly interesting," Thomson said. "It's dust around other stars, analogous to the dust responsible for zodiacal light within our solar system." Mapping it helps astronomers assess the foreground emission that future instruments will face when imaging faint planets.

Longer term, Thomson sees extension into the mid-infrared supporting exoplanet detection and characterization of warm dust. For observatories, the demonstrated value is operational as much as scientific: CHARIOT showed that new photonic components can be tested inside a working observatory, closing the loop between laboratory development and real astronomical measurement. The open question for adoption is whether the sub-1% precision achieved on bright stars can survive the sensitivity and calibration demands of faint targets and larger beam counts.