A Milky Way-like barred spiral galaxy at a redshift of 3

This study investigates whether dynamically cold, barred stellar disks existed in the early Universe, contrary to expectations that high gas turbulence at high redshift suppresses bar formation. Using JWST/NIRCam observations from the CEERS program, the authors identify and characterize a Milky Way-like barred spiral galaxy, ceers-2112, at z ≈ 3. The work tests theoretical predictions that bars are rare beyond z ≈ 1.5 in Milky Way progenitors and evaluates implications for early disk settling, bar formation timescales, and the baryon versus dark matter dominance in galaxy centres at early epochs.

Prior work shows a high fraction of bars in massive local disk galaxies, including the Milky Way. Simulations (e.g., Illustris/TNG, NewHorizon) typically predict a strong decline or near absence of bars beyond z > 1.5, especially in lower-mass Milky Way progenitors. Observational studies report dynamically cold gas disks at z ≈ 5 and early massive rotating disks by z ~ 2–3, while JWST/CEERS has begun revealing bars at z > 1 in rest-frame near-infrared imaging. Theoretical frameworks within ΛCDM posit high gas accretion, turbulence, and strong feedback-driven outflows at z > 5, which can hinder early bar formation by keeping disks near-axisymmetric for longer. Some state-of-the-art simulations (e.g., TNG50) suggest that massive disks and possibly bars could exist by z ~ 4, but producing barred disks beyond z > 1.5 remains challenging, especially at lower masses.

Data: JWST/NIRCam CEERS Epoch 1 imaging (June 21–22, 2022) in seven filters (F115W, F150W, F200W, F277W, F356W, F410M, F444W), with 0.03 arcsec px−1 sampling, PSF FWHM 0.066–0.161 arcsec, and ~29 mag (5σ) depth. HST/ACS and WFC3 imaging (F606W, F814W, F125W, F140W, F160W) from CANDELS were recalibrated and drizzled to the JWST scale. Morphology workflow: (1) Constructed a stacked NIRCam image by PSF-convolving individual bands to F444W using empirical PSFs and pypher-derived kernels, then combining with ccdproc.combine. (2) Performed isophotal analyses (photutils.isophote) on individual bands and on the stack, examining ellipticity and position-angle profiles to identify a bar-dominated region (ε > 0.4, |ΔPA| < 15°) and outer disk where ellipticity/PA drop. (3) Deprojected the stacked image using disk inclination from outer isophotes (εdisk = 0.23 → i ≈ 41°) and carried out Fourier decomposition of azimuthal light with m = 1–6 components; a prominent m = 2 peak (|m2| > 0.4) with roughly constant phase within the bar supports a strong bar. Robustness checks varied PA and inclination (±5°) without changing the bar identification; bar/interbar contrast analysis yielded consistent bar length and strength. (4) Fitted a one-component Sérsic model (statmorph), yielding a low Sérsic index (n ≈ 0.65) and residuals highlighting a bar and spiral arms, indicating the need for multiple components. (5) Performed a 2D photometric decomposition (GASP2D) modeling a double-exponential disk plus a Ferrers bar, masking spiral arms to recover bar parameters; uncertainties estimated via Monte Carlo simulations on mock galaxies. Additional wavelength-dependent checks constructed short-wavelength (F115W–F200W) and long-wavelength (F277W–F444W) stacks to assess dust/stellar-population effects on morphology; the bar is more evident at longer wavelengths. Photometric redshift and SED modeling: Measured HST+JWST photometry with rainbow using small apertures for colors and larger apertures for normalization; estimated photometric redshift with EAZYpy (tweak_fsps_QSF_12_v3 templates), cross-validated with Dense Basis and Prospector. Derived spatially resolved stellar population properties with synthesizer assuming delayed-exponential SFHs, exploring ages up to the cosmic age at z = 3.03, a range of τ (100 Myr–5 Gyr), Bruzual & Charlot metallicities, Calzetti attenuation (AV 0–5 mag), and a Chabrier IMF, including nebular continuum and lines. Assessed systematics using integrated photometry fits with FAST (τ-model), Dense Basis, and Prospector (parametric and non-parametric continuity SFHs), with broad priors on mass, metallicity, attenuation, ages, and τ.

− Identified a strong stellar bar in ceers-2112 at zphot = 3.03 ± 0.05 using JWST/NIRCam CEERS imaging, supported by multiple independent diagnostics (isophotes, Fourier m = 2 peak with |m2| > 0.4, Sérsic residuals, and 2D disk+bar decomposition). − Bar parameters: Ferrers bar length rFerrers = 0.42 ± 0.03 arcsec (3.3 kpc); bar strength Sbar = 0.23 ± 0.01; multiple bar-length estimates from different methods are consistent (≈0.42–0.49 arcsec). − Disk inclination derived from outer isophotes: i ≈ 41° (εdisk = 0.23). − Galaxy structural and stellar-population properties: stellar mass M* = 3.9 × 10^10 M⊙; mass-weighted stellar age 620 ± 150 Myr; stellar mass-density along the bar log Σ ≈ 8.4 M⊙ kpc−2. − Formation timeline: the stellar disk likely assembled at z ≈ 5; the bar formed roughly 200 Myr later and assembled in ≈400 Myr. − ceers-2112 is consistent with being a Milky Way progenitor in both morphology and early mass-assembly history; it is among the closest in mass over the first 4 Gyr. − Implications: (1) baryons could have already dominated galaxy inner dynamics by z = 3; (2) dynamically cold stellar disks existed by z ≈ 4–5; (3) bar formation can proceed rapidly (~400 Myr) at early times despite prior expectations of suppression by gas turbulence.

The detection of a well-developed stellar bar in a z ≈ 3 disk galaxy directly addresses the question of whether dynamically cold disks and bar instabilities arise early. The clear m = 2 Fourier signature, stable bar phase, and 2D decompositions confirm a bar rather than misidentified spiral structure from a compact bulge. The inferred rapid formation of a dynamically cold disk by z ≈ 5 followed by bar growth within ~400 Myr suggests efficient gas consumption and disk settling at early epochs. These observations challenge cosmological simulations that underproduce barred disks beyond z > 1.5, especially at lower masses, and indicate that models must reproduce early baryon-dominated inner disks with significant ordered rotation. The results place constraints on feedback implementations and the baryonic cycle required to transition from turbulent, gas-rich states to bar-unstable disks. As a plausible Milky Way progenitor in both structure and mass assembly, ceers-2112 informs the timeline of bar-driven secular processes and potential contributions to central stellar migration in the first few gigayears. Comparisons with TNG50 demonstrate partial consistency (massive disks and some early bars) but underscore the need for improved realism to match the observed prevalence and properties of high-z bars.

JWST/CEERS imaging reveals ceers-2112 as a Milky Way-like barred spiral galaxy at z ≈ 3 with M* ≈ 3.9 × 10^10 M⊙, a strong bar of ~3.3 kpc length, and a mass-weighted age of ~0.6 Gyr. The galaxy’s disk likely assembled by z ≈ 5 and its bar formed rapidly thereafter (~400 Myr), implying that dynamically cold stellar disks and baryon-dominated inner regions already existed at early times. This discovery provides stringent empirical constraints for galaxy-formation models regarding the timing of disk settling, bar instability, and the interplay of gas fractions, feedback, and angular momentum evolution. Future work should obtain spectroscopic confirmation of the redshift and kinematics, map gas content and dynamics (e.g., with ALMA and JWST/NIRSpec), and expand samples to quantify the bar fraction and properties at z > 2. Simulations should explore feedback and baryon–dark matter coupling capable of producing early, bar-unstable disks across a range of masses.

− The morphology is derived from a composite light distribution spanning rest-UV to near-IR; dust attenuation and spatially varying young stellar populations may bias light relative to the true stellar mass distribution. − The disk is faint at shorter wavelengths, and spiral arms can complicate isophotal measurements, potentially affecting ellipticity and position-angle profiles. − GASP2D does not model spiral arms; masking is required and may influence bar parameter recovery. − Deprojection (inclination and PA) introduces uncertainties, though tests varying these by ±5° did not change the bar identification. − The redshift is photometric (no spectroscopic confirmation reported here). − Possible foreground contamination was mitigated via careful aperture selection and normalization but cannot be entirely excluded. − Wavelength-dependent S/N variations (lower in short-wavelength stacks) can affect bar detectability in individual bands.