ATLAS. II. Extremely High Incidence of Balmer Line Absorption with Predominant Blueshifts in LRDs: Statistical Insights through Comparison with Type 1 AGNs

Hiroto Yanagisawa, Masami Ouchi, Tomokazu Kiyota, Makoto Ando, Yuichi Harikane, Yuta Kageura, Minami Nakane, Yoshiaki Ono, Yui Takeda

First listed 2026-07-30 | Last updated 2026-07-28

Abstract

We present the statistical properties of H$α$ and H$β$ line absorption in little red dots (LRDs) at $z\simeq2.5$--7.2 using archival JWST/NIRSpec spectra from the DAWN JWST Archive and complementary NIRSpec/IFU observations. Among 40 LRDs with broad H$α$ and [O~{\sc iii}] obtained with medium- or high-resolution gratings, 14 objects exhibit H$α$ absorption. We find that the incidence of Balmer line absorption is $\sim35$\% ($=14/40$), significantly higher than that in SDSS low-$z$ type 1 AGNs ($\sim0.04$\%), demonstrating that Balmer line absorption occurs approximately 850 times more frequently in LRDs than in type 1 AGNs. We combine our 14 detections with 32 additional LRD Balmer absorbers from the literature, yielding a census of 46 absorbers. Their velocities span $Δv_\mathrm{abs}(\mathrm{Hα})\sim-430$ to $+140\ {\rm km\,s^{-1}}$, markedly narrower than the $-800$ to $+1600\ {\rm km\,s^{-1}}$ range of Balmer absorption in SDSS type~1 AGNs, for which our simulations confirm that the velocity difference is too large to be explained by detection incompleteness. The lower absolute absorber velocities in LRDs may partly reflect the shallower gravitational potential at their characteristic BLR radii. We also find that 38 of the 46 absorbers (83\%) are blueshifted, with only eight redshifted, indicating that most of Balmer absorbers are moving outward. An analytic model with radiation pressure suggests that most absorbers with $N_{\rm H}\gtrsim10^{24}\ {\rm cm^{-2}}$ remains gravitationally bound. The smaller number of redshifted (i.e., infalling) absorbers may indicate that outbound absorbers lose density: some return to the BLR, whereas others undergo stronger radiative acceleration and escape.

Short digest

Using archival JWST/NIRSpec medium- and high-resolution spectra plus complementary IFU data, ATLAS II searches 40 z≈2.5–7.2 little red dots with broad Hα and [O III] for Balmer absorption. Fourteen show Hα absorption, an incidence of about 35%—roughly 850 times the ≈0.04% rate in low-redshift SDSS type 1 AGNs—and a 46-absorber literature-plus-new census has substantially narrower Hα velocity offsets than the SDSS comparison. Fully 38 of 46 absorbers are blueshifted, pointing to predominantly outward-moving dense gas, while the escape/radiation-pressure analysis argues that high-column absorbers are often still bound and may either fall back toward the BLR or accelerate outward as they lose density.

Key figures to inspect

• Figure 1. This figure defines the color-selected LRD parent sample and identifies which broad-Hα sources host detected Hα absorption, establishing the selection context behind the reported 14/40 incidence.
• Figure 5. This is the central statistical result: it directly contrasts the Balmer-absorption fraction in the LRD sample with the exceptionally low SDSS type 1 AGN fraction and shows how the result relates to other JWST LRD censuses.
• Figure 6. The compiled Hα velocity-offset distribution shows both the narrow LRD range and the strong predominance of blueshifted components in the 46-absorber census, which underpin the paper's outflow interpretation.
• Figure 7. The forward-modeling test demonstrates that observational incompleteness cannot plausibly transform the broad SDSS type 1 AGN absorption-velocity distribution into the narrow distribution observed for LRDs.
• Figure 11. This escape diagram connects the measured blueshifted absorbers to the radiation-pressure model, illustrating the conclusion that dense, high-column outflowing gas can remain gravitationally bound while lower-density material may escape.