Subaru’s New Spectrograph Will Read 100,000 Stars to Test Dark Matter and Galactic Mergers

Over the next few years, some 100,000 stars in the Local Group will have their light split into rainbows. For many of those stars, each spectrum will encode a star’s speed, its chemical fingerprint, and, for main-sequence turn-off stars, clues to when it formed. Astronomers plan to use those spectra as fossils to reconstruct how the Milky Way, Andromeda, and their dwarf companions grew over billions of years.

The survey is called Galactic Archaeology. It uses the ‘Ōnohi‘ula Prime Focus Spectrograph, or PFS, on the 8.2-meter Subaru Telescope in Hawai‘i. Its Hawaiian name means “perceiving the realm of our origins.” PFS can place 2,394 optical fibers across a 1.24-square-degree field, then record spectra covering 380 to 1,260 nanometers in a single exposure. Science fibers are aimed at individual stars, turning one telescope pointing into thousands of simultaneous chemical and velocity measurements.

The Subaru Strategic Program for PFS will run for 360 nights over six years, having begun in 2025. About one-third of that time, 130 nights, is set aside for Galactic Archaeology. The survey is described in a new arXiv preprint. It is a plan and forecast, not final survey results.

Does Dark Matter Form a Peak or a Plateau?

Dwarf galaxies are among the best places to test how dark matter behaves on small scales. The standard cold dark matter model predicts a cusp: the density of dark matter should rise steeply toward a galaxy’s center. But many observations suggest a core: a flatter central density. One possible reconciliation is that bursts of star formation and supernova feedback can push dark matter outward, erasing a cusp.

For its dark-matter test, PFS will target six dwarf spheroidal galaxies: Draco, Ursa Minor, Sextans, Sculptor, Fornax, and Boötes I. It will also observe the dwarf irregular NGC 6822 as a contrast. The survey aims to measure velocities for about 18,000 member stars across the dwarf sample; it expects metallicities for roughly 17,900 and [α/Fe] for about 8,900. For the six dwarf spheroidals, the samples reach beyond each galaxy’s nominal tidal radius. It will use several dynamical models, including axisymmetric Jeans analysis and fourth-order velocity moments, to break the long-standing degeneracy between dark matter density and the anisotropy of stellar motions. It will also measure the abundance ratio of alpha elements, such as magnesium and silicon, to iron—a chemical clock that records bursty star formation. The paper describes expected constraints; whether the data favor cusps, cores, or a mix remains to be seen.

Andromeda as a Foil to the Milky Way

The Milky Way’s last major merger, Gaia-Sausage-Enceladus, happened 8 to 10 billion years ago. Since then, its accretion history has been relatively quiet—an unusual path for a galaxy of its mass in the standard cosmological model. Andromeda, or M31, may have had a more violent past. Comparing the two is a way to test whether the Milky Way is special.

PFS will observe about 30,000 stars across 45 square degrees of M31’s halo and outer disk. It will measure line-of-sight velocities and metallicities for those stars, and alpha-to-iron ratios for roughly 15,400 of them. These data can reveal whether M31’s disk has the same two chemical sequences as the Milky Way’s thin and thick disks, and whether its halo was built mainly by a major merger or many minor ones. The survey will also target about 2,000 stars around M33 to test whether that smaller spiral has a double-structured stellar halo.

Listening to the Milky Way’s Outer Disk

The outer disk and halo of the Milky Way are stretched and warped by ongoing gravitational tugs. The Sagittarius dwarf galaxy and the Large Magellanic Cloud are both exerting ongoing perturbations; the Large Magellanic Cloud in particular is expected to leave a wake in the halo. Because dynamical times are long in these outer regions, the disturbances persist, making the outskirts a living record of recent and ancient accretion.

PFS will target faint main-sequence turn-off stars, which are unbiased tracers of the underlying population. It will measure their velocities, metallicities, and—using stellar evolution models—ages out to about 30 kiloparsecs from the Galactic center. That range can map debris from Gaia-Sausage-Enceladus and the Sagittarius stream; other PFS halo fields are designed to probe the reflex motion caused by the Large Magellanic Cloud in the distant halo. The goal is a picture in position, chemistry, and age of how the Galaxy responds when smaller galaxies fall in.

What Could Blunt the Fossil Record

The plan is ambitious, and the paper is careful about uncertainty. The 130 nights assume a clear-sky fraction of about 70 percent, leaving roughly 91 clear nights. Many target stars are faint, so some spectra will have low signal-to-noise ratios. Membership contamination from foreground Milky Way stars must be modeled. Binary stars can inflate velocity measurements, so some dwarf-spheroidal fields will be observed twice. Dynamical models still face degeneracies between mass and velocity anisotropy.

Most importantly, the cusp-versus-core problem may not have a single answer. Baryonic feedback, fuzzy dark matter, self-interacting dark matter, and other ideas can all produce cores. The PFS survey is designed to measure the star formation histories and density profiles together, which should help distinguish between them. The paper presents expected constraints; the interpretation of any cores will still depend on models. The M31 comparison also depends on simulations of galaxy formation, which carry their own assumptions.

A Fossil Record Made of Starlight

If the survey unfolds as planned, by 2030 it will have gathered spectra for nearly 100,000 stars across the Local Group. Each spectrum is a line in a long story: how fast a star moves, what it is made of, and when it formed. Read together, those lines can reconstruct the assembly of galaxies and the behavior of the invisible matter that surrounds them. The ‘Ōnohi‘ula instrument is named for perceiving the realm of our origins. Its success will be measured in fossils of light.