Roman Galactic Center Science Program

Proposed formation pathways for the NSC include cluster inspiral driven by dynamical friction (Antonini et al. 2012; Tsatsi et al. 2017), in-situ star formation fueled by episodic gas inflow (Alexander & Pfuhl 2014), and hybrid scenarios in which early inspirals are followed by continued gas accretion (van Donkelaar et al. 2024). The NSD is thought to arise from sustained star formation in the central molecular zone (Schultheis et al. 2021; Schödel et al. 2023). Distinct pathways should leave distinct kinematic signatures, including long-lived streams from infalling clusters (Arca Sedda et al. 2020).

Existing kinematic data are limited to the inner ~30 pc for proper motions plus sparse radial velocities (Feldmeier-Krause et al. 2025), and NSD models have relied on heavily extrapolated coverage (Sormani et al. 2022). Spectroscopy of the inner parsecs already shows the NSC is chemically inhomogeneous, with a metal-poor population that appears to rotate differently from the dominant metal-rich stars (Do et al. 2020), a possible infalling-cluster remnant.

The program will fit self-consistent phase-space (distribution-function) models of the NSC, NSD and inner bar simultaneously, using red clump stars as tracers, to recover the 3D density, mass distribution and rotation, and to measure the gravitational potential in the inner 25–50 pc — currently the dominant systematic in dynamical studies of the region. Roman's expected 0.4 mas/yr precision resolves the ~50 km/s NSD velocity dispersion at high significance, with ~100× more stars and ~50× more area than previous NSD modeling.

Dynamical models predict the Arches and Quintuplet (projected distances ~30 pc from Sgr A*) dissolve within ~20 Myr under the local tidal field (Kim et al. 2000; Portegies Zwart et al. 2002) and should already exhibit tails extending >10 pc, dominated by stars below ~2 M_⊙ (Habibi et al. 2014; Park, Goodwin & Kim 2018). Existing proper-motion studies reach only the bound population within r < 3 pc (Hosek et al. 2019; Rui et al. 2019) or lack the precision to isolate low-mass members.

The GBTDS field covers >10 pc along the clusters' projected orbits, whose bulk motions are known to high precision (Hosek et al. 2022), and reaches σ_μ ≤ 0.4 mas/yr down to ~0.5 M_⊙ for a young Arches-like population. Tail members will be identified as overdensities in proper motion, position and color space using mixture-model techniques already validated on the bound clusters, then characterized through their density profiles and mass functions.

The stripped population is also the largest uncertainty in the clusters' initial mass functions. The Arches IMF is measured to be top-heavy relative to the local IMF for M ≥ 2 M_⊙ (Hosek et al. 2019), but existing analyses neglect stars lost to the tails, which matters most at the low masses now becoming accessible with JWST.

Relative to the solar neighborhood, the central molecular zone has ~100× the stellar surface density, ~20× the gas surface density, gas temperatures of 50–100 K, fields of 10–1000 µG and turbulent pressures P/k ~ 10⁹ K cm⁻³ (Henshaw et al. 2023; Rathborne et al. 2014). Competing models predict different environmental dependence of the IMF's high-mass slope and characteristic mass (e.g., Bonnell et al. 2006; Larson 2005; Chabrier & Dumond 2024). The Arches IMF is top-heavy above 2 M_⊙ (Hosek et al. 2019); the program's tidal-tail measurements (Q2) are what allow this to be extended to lower masses.

Roughly 80 young massive stars in the Galactic center are not associated with a known cluster (Mauerhan et al. 2010; Dong et al. 2011; Clark et al. 2021). If most stars form in clusters (Lada & Lada 2003), these should be members of tidally dissolved clusters and sit inside co-moving groups; Martínez-Arranz et al. (2024) found candidate groups for 4 of 59 using HST proper motions (Libralato et al. 2021). Alternatively, isolated formation would point to a mode specific to massive stars or to this environment (Krumholz et al. 2010; Zinnecker & Yorke 2007).

About 50 of the known isolated massive stars fall within the GBTDS Galactic center field. Roman's astrometry reaches ~6.5 mag deeper than the existing HST analysis, enough to detect or exclude co-moving groups of low-mass stars around each one, and to test whether their trajectories trace back to the Young Nuclear, Arches or Quintuplet clusters.

Sgr A* is variable in the near-infrared on all timescales sampled, from minutes to years, and its variability is well described by a red-noise process with a characteristic break timescale of a few hours (Witzel et al. 2018; see Genzel et al. 2010 for a review). The longest continuous near-infrared light curve to date is ~24 hr with Spitzer. Ground-based monitoring is limited by nightly and seasonal gaps, and by weather.

The GBTDS Galactic center field will observe Sgr A* at 12.1-min cadence throughout each ~70-day high-cadence season, yielding uninterrupted light curves that extend the power spectrum by two orders of magnitude in frequency, test for long-timescale trends or periodicities, and characterize the flare rate and duty cycle across seasons. Roman's angular resolution (~0.1″) is coarser than adaptive-optics imaging, so Sgr A* photometry will require careful treatment of the confusing stellar background, an approach the team has developed with HST WFC3-IR.