---
DOAJ_listed: '1'
OA_place: publisher
OA_type: gold
PlanS_conform: '1'
_id: '23115'
abstract:
- lang: eng
  text: "We investigate the association between galaxies and neutral O I absorption
    systems at z ∼ 6, which trace metalenriched gas during the epoch of reionization.
    We identify 40 galaxies across six quasar fields, residing in 15\r\noverdensities
    within 300 kpc of the background sight lines. Five O I absorption systems are
    associated with five of\r\nthese overdensities, yielding a covering fraction of\r\n0.27
    0.10\r\n+0.13\r\nwithin 300 kpc. The absorption occurs beyond typical\r\nvirial
    radii, indicating that the gas traces extended overdensity environments rather
    than individual galaxy halos,\r\nunlike the z ∼ 0 circumgalactic medium (CGM),
    which is largely bound to halos. These galaxy-associated\r\nabsorbers account
    for ∼35% of all O I systems seen in blind quasar surveys, implying the remainder
    arise in\r\nlower-mass galaxies below our detection threshold or in dense neutral
    intergalactic medium pockets. The CGM\r\naround these galaxies contains ≳ 2 ×
    106 M⊙ of oxygen, comparable to the interstellar medium oxygen mass of\r\nthe
    galaxies themselves, suggesting that the surrounding environment holds as much
    metal mass as the galaxies.\r\nAll five galaxy-associated systems show significantly
    higher\r\nlog NCII NOI ( / )\r\nratios than absorbers lacking galaxy\r\nassociations.
    Furthermore, relative abundance ratios ([Si/O], [C/O]) reveal that four of the
    five exhibit\r\nenrichment patterns consistent with Population III nucleosynthesis
    at the outskirts of galaxy overdensities. These\r\nrare systems offer a unique
    window into the role of first-generation stars in shaping the early metal enrichment
    of\r\ngalaxies and their environments."
acknowledgement: "IOP Science home\r\nAccessibility Help\r\n\r\nJournals\r\nBooks\r\nPublishing
  Support\r\nLogin\r\nThe Astrophysical Journal\r\nThe American Astronomical Society,
  find out more.\r\n\r\nThe following article isOpen access\r\nEIGER. VIII. First
  Stars Signatures in the Connection between O I Absorption and Galaxies in the Epoch
  of Reionization\r\nJack Higginson, Rongmon Bordoloi, Robert A. Simcoe, Jorryt Matthee,
  Daichi Kashino, Ruari Mackenzie, Ivan Kramarenko, Simon J. Lilly, Anna-Christina
  Eilers, Rohan P. NaiduShow full author list\r\n\r\nPublished 2026 February 24 •
  © 2026. The Author(s). Published by the American Astronomical Society.\r\nThe Astrophysical
  Journal, Volume 999, Number 1\r\nCitation Jack Higginson et al 2026 ApJ 999 49\r\nDOI
  10.3847/1538-4357/ae3e81\r\n\r\nPDFOpens in a new tab.\r\nePub\r\nAuthors\r\nFigures\r\nTables\r\nReferences\r\nArticle
  data\r\nPDFOpens in a new tab.\r\nePub\r\nArticle metrics\r\n626 Total downloads\r\n\r\n22
  total citations on Dimensions.\r\nShare this article\r\nArticle information\r\nAbstract\r\nWe
  investigate the association between galaxies and neutral O i absorption systems
  at z ∼ 6, which trace metal-enriched gas during the epoch of reionization. We identify
  40 galaxies across six quasar fields, residing in 15 overdensities within 300 kpc
  of the background sight lines. Five O i absorption systems are associated with five
  of these overdensities, yielding a covering fraction of \r\n within 300 kpc. The
  absorption occurs beyond typical virial radii, indicating that the gas traces extended
  overdensity environments rather than individual galaxy halos, unlike the z ∼ 0 circumgalactic
  medium (CGM), which is largely bound to halos. These galaxy-associated absorbers
  account for ∼35% of all O i systems seen in blind quasar surveys, implying the remainder
  arise in lower-mass galaxies below our detection threshold or in dense neutral intergalactic
  medium pockets. The CGM around these galaxies contains  ≳  2 × 106 M⊙ of oxygen,
  comparable to the interstellar medium oxygen mass of the galaxies themselves, suggesting
  that the surrounding environment holds as much metal mass as the galaxies. All five
  galaxy-associated systems show significantly higher \r\n ratios than absorbers lacking
  galaxy associations. Furthermore, relative abundance ratios ([Si/O], [C/O]) reveal
  that four of the five exhibit enrichment patterns consistent with Population III
  nucleosynthesis at the outskirts of galaxy overdensities. These rare systems offer
  a unique window into the role of first-generation stars in shaping the early metal
  enrichment of galaxies and their environments.\r\n\r\nExport citation and abstract\r\nBibTeX\r\nRIS\r\n\r\nPrevious
  article in issue\r\nNext article in issue\r\nRelated links\r\n\r\nOriginal content
  from this work may be used under the terms of the Creative Commons Attribution 4.0
  licence. Any further distribution of this work must maintain attribution to the
  author(s) and the title of the work, journal citation and DOI.\r\n\r\n1. Introduction\r\nThe
  first generations of stars formed shortly after the Epoch of Recombination within
  dark matter overdensities, producing the first significant quantities of metals
  and enriching the intergalactic medium (IGM) while reionizing hydrogen and helium
  (e.g., R. S. Klessen & S. C. O. Glover 2023). These metals and gas eventually coalesced
  into the first galaxies, which over cosmic time evolved into Milky Way–like systems
  today. Studying these young galaxies and their environments provides critical insight
  into galaxy formation and evolution, constraining models of stellar feedback, gas
  accretion, and chemical enrichment (C.-A. Faucher-Giguère & S. P. Oh 2023; C. Peroux
  & D. Nelson 2024).\r\n\r\nAt low redshift, decades of observations using deep ground-based
  spectroscopy and Hubble Space Telescope (HST) data of background quasars and foreground
  galaxies have built a rich understanding of the circumgalactic medium (CGM) and
  its interplay with galaxies (e.g., J. Tumlinson et al. 2017). The CGM plays a crucial
  role in gas recycling, star formation regulation, and quenching; understanding its
  structure and composition informs galaxy growth models (Y. Peng et al. 2015; E.
  Zinger et al. 2020).\r\n\r\nHigh-redshift studies, leveraging ground- and space-based
  near-infrared (NIR) spectroscopy of quasars, reveal that the early Universe was
  already chemically enriched by the epoch of reionization (e.g., T. J. Cooper et
  al. 2019; L. Christensen et al. 2023; R. L. Davies et al. 2023a). Among tracers
  of neutral and low-ionization gas, O i has emerged as a key probe: surveys report
  an increase in O i number density at z  ≳  5–6 (G. D. Becker et al. 2019; L. Christensen
  et al. 2023; A. M. Sebastian et al. 2024; A. Sodini et al. 2024), while higher-ionization
  species like C iv and Si iv decline, consistent with a softer and evolving UV background
  at the end of reionization (T. J. Cooper et al. 2019; V. D’Odorico et al. 2022;
  R. L. Davies et al. 2023a, 2023b).\r\n\r\nHistorically, connecting absorption systems
  with their host galaxies was challenging due to the difficulty of obtaining deep,
  high-redshift galaxy spectroscopy. This has changed with the advent of JWST, providing
  unprecedented views of early galaxies and their impact on surrounding gas through
  surveys such as EIGER (D. Kashino et al. 2023) and ASPIRE (F. Wang et al. 2023).
  First systematic studies of cool, low-ionization CGM traced by Mg ii around z ∼
  6 galaxies suggest that high-redshift CGM differs significantly from its low-redshift
  counterpart: these galaxies typically have low masses, and strong stellar feedback
  can eject metals far beyond the halo’s gravitational potential (R. Bordoloi et al.
  2024). Such pioneering studies now allow systematic investigations of the CGM in
  the early Universe.\r\n\r\nThe UV background during reionization was inhomogeneous
  and rapidly evolving (e.g., S. R. Furlanetto & S. P. Oh 2005; B. D. Oppenheimer
  et al. 2009; K. Finlator et al. 2013; G. D. Becker et al. 2015; S. E. I. Bosman
  et al. 2022), initially dominated by Population III stars (K. Nakajima & R. Maiolino
  2022; K. Nakajima et al. 2025), then by early star-forming galaxies (L. Jiang et
  al. 2022; J. Matthee et al. 2022), with active galactic nuclei (AGN) contributing
  significantly by z ∼ 5 (A. D’Aloisio et al. 2017; G. Kulkarni et al. 2019; C.-A.
  Faucher-Giguère 2020). Modeling indicates that pockets of Population III star formation
  may persist throughout the Epoch of Reionization (L. Tornatore et al. 2007; R. H.
  Mebane et al. 2018; J. L. Johnson & A. Aykutalp 2019; A. Venditti et al. 2023; O.
  Zier et al. 2025).\r\n\r\nThe large reservoir of neutral O i observed in quasar
  absorption lines at high redshift is typically self-shielded, providing robust estimates
  of chemical abundances. Studying the association of these absorbers with galaxies
  offers unique insight into the CGM of early galaxies, and into potential contributions
  from Population III stars to early chemical enrichment.\r\n\r\nIn this work, we
  study the cool, neutral CGM around galaxies at z ∼ 6, traced by O i absorption from
  the EIGER survey (D. Kashino et al. 2023). We focus on O i absorbers associated
  with [O iii]5008-emitting galaxies at 5.875 <z < 6.378 within 300 kpc of six background
  quasars. We quantify absorber incidence, column densities, and mass estimates, and
  analyze ionic abundances to probe chemical enrichment, including potential contributions
  from Population III stars.\r\n\r\nThis paper is organized as follows. Section 2
  describes the observations; Section 3 details our methods for galaxy selection,
  redshift estimation, galaxy properties, defining galaxy overdensities, absorber
  measurements, and completeness; Section 4 presents the results, including the association
  of O i absorption with galaxies, the incidence of galaxy-associated O i absorbers,
  their chemical abundances, and physical interpretations; Section 5 summarizes our
  conclusions. Throughout, we adopt a ΛCDM cosmology with H0 = 67.66 km s−1 Mpc−1
  (Planck Collaboration et al. 2020).\r\n\r\n2. Observations\r\nEIGER (Program ID
  1243; PI: S. J. Lilly) is a large JWST Wide Field Slitless Spectroscopy (WFSS) program
  totaling 126.5 hr, conducting deep WFSS observations in six extragalactic fields,
  each centered on an ultraluminous quasar at z ≥ 6. These quasars have also been
  observed with deep optical and NIR medium- and high-resolution spectroscopy with
  Magellan/FIRE, Very Large Telescope/X-shooter, and Keck High Resolution Echelle
  Spectrometer (HIRES)/MOSFIRE, providing over 100 hr of complementary NIR spectroscopy
  across the six sight lines. We refer the reader to A.-C. Eilers et al. (2023) and
  A.-C. Ďurovčíková et al. (2025) for data reduction and details of the ground-based
  quasar spectroscopy. In addition, all fields are covered by deep JWST/NIRCam and
  HST imaging. A detailed description of the survey design, methodology, and spectroscopic
  and photometric observation details is provided in D. Kashino et al. (2023), J.
  Matthee et al. (2023), and R. Bordoloi et al. (2024). In this work, we make use
  of the complete EIGER survey of z ≥ 5.9 galaxies spanning all six extragalactic
  fields (D. Kashino et al. 2026).\r\n\r\nWe note that all six quasars were chosen
  with prior knowledge of metal absorption systems along their lines of sight, which
  could in principle bias any attempt to associate galaxies with those absorbers.
  In contrast, a galaxy-centric approach starts with quasar–galaxy pairs without prior
  knowledge of absorption and then builds statistics systematically (see e.g., R.
  Bordoloi et al. 2011; J. K. Werk et al. 2013; R. Bordoloi et al. 2014). However,
  once a blind, well-characterized galaxy survey is in place, the two approaches converge.
  This is conceptually similar to a program that combines quasar spectroscopy performed
  without prior absorber selection with a complete galaxy survey. This work employs
  deep blind galaxy searches in each of the EIGER fields, yielding an emission-line
  complete sample of spectroscopic galaxies (J. Matthee et al. 2022; R. Bordoloi et
  al. 2024) and a complete census of absorption lines along these sight lines (Section
  3.5). By combining these two elements—a galaxy survey with quantified completeness
  and an absorption survey with well-defined sensitivity—we establish a robust framework
  for conducting a census of CGM absorption around galaxies. This combined approach
  has been successfully demonstrated at low redshifts (J. N. Burchett et al. 2019).\r\n\r\n3.
  Methods\r\nIn this section, we describe how galaxy redshifts and properties are
  measured and how absorption lines are identified and measured.\r\n\r\n3.1. Galaxy
  Redshifts\r\nSince our focus is on galaxies at z ∼ 6 near the quasar sight lines,
  we begin with the full EIGER catalog of [O iii] emitters (D. Kashino et al. 2026),
  which contains 948 galaxies detected via [O iii] emission within 5.33 ≤ z ≤ 6.97.
  From this catalog, we select galaxies that (1) lie at least 10,000 km s−1 in the
  foreground of each quasar , (2) are within a projected impact parameter of 300 pkpc,
  and (3) have O i λ1302 coverage in the quasar NIR spectra outside of the Gunn–Peterson
  trough. This yields a sample of 40 galaxies. Of these, 33 fall within the central
  of the EIGER spectroscopic footprint (R. Bordoloi et al. 2024), observed with both
  JWST/NIRCam reverse grism modules (A and B). Because the two modules disperse light
  in opposite directions, they provide independent moderate-resolution (R ∼ 1500)
  spectra of the same target, enabling us to assess redshift measurement uncertainties.\r\n\r\nFollowing
  the procedure in R. Bordoloi et al. (2024), we visually inspect each NIRCam 2D spectrum
  using the custom Python application programming interface (API) zgui(R. Bordoloi
  et al. 2025), extract 1D spectra from each module, and fit Gaussians to the emission
  line to determine redshifts. Redshifts from the two modules are consistent with
  a mean velocity offset of Δv = 9.3 km s−1 with a standard deviation of 24 km s−1
  (Figure 1). We therefore adopt an average redshift uncertainty of ≈24 km s−1 for
  the [O iii] emitters, sufficient for associating metal absorption systems with their
  host galaxies. Hereafter, we focus on this sample of 40 galaxies.\r\n\r\nZoom InZoom
  OutReset image size\r\nFigure 1. Redshift uncertainty distribution of the [O iii]
  emitter EIGER galaxies. The histogram shows the difference in redshift between emission-line
  measurements from NIRCam Grism Module A and Module B spectra, expressed in km s−1.
  Individual redshifts were derived from Gaussian fits to the [O iii] and Hβ lines.
  The two modules yield consistent results with a mean offset and standard deviation
  of Δz = 0.0028 ± 0.0073 (Δv = 9.3 ± 24 km s−1).\r\n\r\nDownload figure:\r\n\r\nStandard
  image\r\nHigh-resolution image\r\n3.2. Galaxy Stellar Mass, Star formation Rate,
  Halo Mass\r\nTo estimate galaxy stellar masses and star formation rates (SFRs),
  we follow the procedure of J. Matthee et al. (2022). We first measure the [O iii]
  and Hβ emission-line fluxes of each galaxy, and combine these with JWST/NIRCam broadband
  photometry (F115W, F200W, and F356W) to perform spectral energy distribution fitting
  at the galaxy redshift using the Prospector code (B. D. Johnson et al. 2021). We
  assume a G. Chabrier (2003) initial mass function (IMF), adopt MIST isochrones (J.
  Choi et al. 2016; A. Dotter 2016), a nonparametric star formation history model
  (R. P. Naidu et al. 2024), and the dust attenuation curve of D. Calzetti et al.
  (2000). Galaxy SFRs are derived by averaging over the most recent 50 Myr of star
  formation history from the Prospector posteriors.\r\n\r\nFigure 2 (left) shows the
  impact parameter distribution of the 40 EIGER galaxies relative to the background
  quasars (black star symbols). The galaxies are randomly distributed with respect
  to the quasar sight lines. The right panel shows their stellar masses and SFRs,
  with symbols color coded by galaxy redshift. Following R. Bordoloi et al. (2024),
  we estimate halo masses and virial radii (Rvir) using abundance matching from P.
  Behroozi et al. (2019). Uncertainties in Rvir are propagated from the 16th and 84th
  percentile stellar mass errors, giving a mean uncertainty of ΔRvir = 1.8 kpc. The
  galaxies span stellar masses of and SFRs of , with all systems being star-forming
  by selection.\r\n\r\nZoom InZoom OutReset image size\r\nFigure 2. Properties of
  40 spectroscopically confirmed galaxies at 5.9 ≤ z ≤ 6.4 within 300 kpc of background
  quasars in the EIGER survey. Each square denotes a galaxy and is color coded by
  redshift. Left: angular positions and impact parameters relative to the background
  quasar (black star), with dashed concentric rings marking 100, 200, and 300 kpc
  at z = 6. Right: stellar mass vs. SFR, averaged over the last 50 Myr. Error bars
  show the 16th–84th percentile uncertainties in stellar mass and SFR.\r\n\r\nDownload
  figure:\r\n\r\nStandard image\r\nHigh-resolution image\r\nOne galaxy (z = 5.910384,
  galaxyID = 15362, qsoID = J159) has an exceptionally high stellar mass of , beyond
  the calibrated range of the P. Behroozi et al. (2019) relation. This object exhibits
  a compact, pointlike morphology and likely hosts an obscured AGN, placing it among
  the recently identified population of little red dots (LRDs; e.g., J. Matthee et
  al. 2024). We do not detect broad Hβ for this object. However, this is not unexpected,
  as LRDs can have Hα/Hβ values as high as ∼10 (M. Brooks et al. (2025; A. Torralba
  et al. 2026, in preparation). As robust AGN modeling at these redshifts is not yet
  established, we estimated its halo mass by linearly extrapolating the P. Behroozi
  et al. (2019) relation from the stable stellar mass regime. We explicitly note where
  this assumption may affect our results in the following sections. Additionally,
  this galaxy has not been plotted in the right panel of 2.\r\n\r\n3.3. Galaxy Overdensities\r\nA
  significant fraction of z ≥ 6 galaxies in the EIGER survey are not isolated but
  are found in overdense environments, either as mergers or members of “protogroups”
  (see, e.g., D. Kashino et al. 2026). The EIGER [O iii] galaxy catalog already merges
  individual [O iii] emission clumps within 2″ into a single system, even when they
  exhibit complex morphologies or merger signatures. To identify larger-scale overdensities,
  we therefore search for galaxies that lie at similar redshifts but are widely separated
  on the sky.\r\n\r\nWithin 300 kpc from the QSO sight lines, we identify overdensities
  around the 40 selected EIGER galaxies using a friends-of-friends algorithm (R. Bordoloi
  et al. 2025). Galaxies are grouped as an overdensity if they lie within 1 pMpc in
  projected separation and within ±1000 km s−1 in redshift space. A galaxy is considered
  part of an overdensity only if all members are “fully” connected, i.e., each galaxy
  in the overdensity satisfies these conditions with all other members. Around each
  detected O i absorption system, we further search for overdensities within ±1500
  km s−1 of the absorber redshift, and merge these associations to ensure all clustered
  galaxies are treated as a single overdensity. This conservative approach ensures
  that we capture the full galaxy environment of O i absorbers. The 40 galaxies studied
  in this work are part of 15 distinct overdensities, with each grouping containing
  1–5 members. The number of galaxies in each grouping is reported in Table 1. The
  typical velocity dispersion of these overdensities is 449.2 km s−1.\r\n\r\nTable
  1. O i Absorption Lines and Associated Galaxy Properties\r\n\r\nLabel\tQuasar\tzgala\tR\t〈v〉b\t#c\tzO
  I\t\t\t\t\t\r\n \t \t \t(kpc)\t(km s−1)\tGalaxies\t \t(cm−2)\t(cm−2)\t(cm−2)\t(cm−2)\t(cm−2)\r\n(1)\t(2)\t(3)\t(4)\t(5)\t(6)\t(7)\t(8)\t(9)\t(10)\t(11)\t(12)\r\na\tJ0100\t5.941\t119.6\t157\t5\t5.945\t\t\t\t\t\r\nb\tJ1148\t6.008\t157.5\t130\t3\t6.011\t\td\t\t\t\r\nc\tJ159\t5.934\t25.2\t−920\t5\t5.913\t\t\t\t⋯\t⋯\r\nd\tJ159\t6.049\t40.8\t230\t2\t6.054\t\t<13.12\t\t⋯\t⋯\r\ne\tJ1030\t6.075\t147.9\t−131\t2\t6.076\te\t\t\t⋯\t⋯\r\nNotes.
  The galaxies listed are the primary members of groups with detections at the lowest
  impact parameter from the host quasar. aRedshift of the primary member of the overdensity.
  bVelocity offset of O i relative to the primary member galaxy of the overdensity.
  cNumber of galaxy members in the overdensity. dFor this system, Si II λ1304 was
  not present, so the value given here is derived from fitting Si II λ1526. eThis
  system is blended; see discussion in Section 3.4.\r\n\r\nDownload table as: \r\nASCII\r\nTypeset
  image\r\n\r\nThroughout this work, we define the galaxy with the smallest impact
  parameter to the background quasar sight line within each galaxy overdensity as
  the primary member. This convention is adopted solely to simplify the presentation
  and does not imply a unique physical association between the intervening absorption
  system and this galaxy. In fact, the concept of a unique absorber–galaxy association
  at high redshift is significantly less robust than in the low-redshift Universe
  (e.g., R. Bordoloi et al. 2024). As discussed in Section 4.1, none of the O i absorption
  systems lie within the virial radius of any individual halo. Even at low redshift,
  the galaxy with the smallest impact parameter is not necessarily the one with the
  smallest velocity offset from the absorber, and therefore does not constitute evidence
  for a unique absorber–galaxy pair (A. Hamanowicz et al. 2020). This behavior is
  also observed for galaxies residing in groups (R. Bordoloi et al. 2011). Our analysis,
  therefore, focuses on the broader environment, which we argue is most appropriately
  characterized by galaxy overdensities, and our choice of a primary galaxy does not
  affect our main conclusions.\r\n\r\n3.4. Absorption-line Detection and Measurements\r\nWe
  follow the procedures outlined in R. Bordoloi et al. (2024) to identify and measure
  the strength of O i absorption systems around EIGER galaxies, and provide a brief
  summary here. We first use a custom Python-based 1D spectrum viewer to identify
  and tabulate all intervening absorption lines in the EIGER quasar spectra. A semiautomated
  CGM absorption-line analysis pipeline is then used to perform the measurements.
  For each galaxy, we extract a ±1500 km s−1 spectral window centered on its systemic
  redshift, for all transitions of interest. In this work, we analyze slices around
  O i λ1302, Si ii λ1260, Si ii λ1304, Si ii λ1527, C ii λ1334, Si iv λ1393, Si iv
  λ1402, C iv λ1548, and C iv λ1550. Each slice is locally continuum-normalized using
  an automated Legendre polynomial fit, with the polynomial order chosen via the Bayesian
  Information Criterion. The continuum fits and absorption features are visually inspected,
  and both rest-frame equivalent widths (Wr) and apparent optical depth (AOD) column
  densities are measured. We define a detection as any absorption system with Wr ≥
  3σw, where σw is the 1σ uncertainty in Wr. We note that this uncertainty contains
  both the statistical uncertainty due to the signal-to-noise ratio of the data and
  the continuum-fitting uncertainty. Nondetections are reported as 2σw upper limits
  measured within ±100 km s−1 of the systemic redshift of the galaxy. For column density,
  we report the 2σ uncertainty on AOD column density as an upper limit on nondetections.
  When a transition is detected in multiple instruments, each spectrum is analyzed
  independently, and the inverse-variance-weighted mean is adopted as the final Wr.
  The details of this automated approach and full documentation of the APIs used can
  be found in R. Bordoloi et al. (2025).\r\n\r\nFor all detections, we perform Voigt
  profile fitting using the rbvfit package, which employs Markov Chain Monte Carlo
  (MCMC) sampling to simultaneously fit multiple absorption systems across different
  instruments (R. Bordoloi & J. Higginson 2025). This approach yields robust column
  densities, resolves the kinematic structure, and enables accurate modeling of blended
  lines. Additionally, since we obtain a marginalized posterior distribution for the
  fitted parameters, the results yield accurate column density estimates even if the
  Doppler b parameter or velocity centroids are not well constrained in moderate-resolution
  spectra (see discussion in R. Bordoloi et al. 2024). In our case, most of the O
  i Si ii C ii transitions are all covered by higher-resolution HIRES spectroscopy
  and thus their properties are well constrained. rbvfit supports two MCMC samplers,
  emcee (D. Foreman-Mackey et al. 2013) and zeus (M. Karamanis et al. 2021; M. Karamanis
  & F. Beutler 2020), both of which are used in this work depending on the complexity
  of the fit. We typically adopt 50 walkers with initial perturbations of either 10−3
  or 10−6, and default chain lengths of 8000–20,000 steps, ensuring the total number
  of steps exceeds 50 times the autocorrelation time. Additional steps are run as
  needed to achieve convergence. In each case, a flat prior is assumed with reasonable
  physical bounds.\r\n\r\nFigure 3 shows an example of Voigt profile fits to O i λ1302
  and Si ii λ1304 lines simultaneously modeled across FIRE, X-shooter, and HIRES.
  We fit a two-component Voigt profile with a mean velocity offset from the host galaxy
  of voffset =  157 km s−1 . This joint fitting maximizes information from all instruments
  while preserving the kinematic detail provided by higher-resolution data. Galaxy
  overdensities associated with this system are marked with green vertical ticks.
  This system has a total O i column density 13.46. All other O i fits are presented
  in Figure 8. All absorption-line measurements are presented in Table 1.\r\n\r\nZoom
  InZoom OutReset image size\r\nFigure 3. Voigt profile fits to O i and Si ii absorption
  profiles at z = 5.945 toward the quasar SDSS J0100+2802. Shown are the flux (black),
  error spectrum (red), best-fit Voigt model (orange/black), and individual Voigt
  components (thin blue), using data from FIRE, X-shooter, and HIRES. Vertical green
  lines mark galaxy members of the associated overdensity. Spectra from all instruments
  and all transitions are simultaneously fit using the rbvfit Bayesian absorption-line
  fitter.\r\n\r\nDownload figure:\r\n\r\nStandard image\r\nHigh-resolution image\r\nWe
  note that an O i and Si ii absorption system is detected along the J1030 quasar
  sight line at z ∼ 6.075, blended with an intervening C iv absorber at z ∼ 4.948
  (Figure 8, top left panels; see also C. G. Díaz et al. 2021). The z ∼ 6.076 system
  is confirmed by the independent detection of Si ii, Mg ii, C ii, and Fe ii absorption
  in FIRE and X-shooter data, and we simultaneously fit the C iv, O i, and Si ii systems
  across three instruments. Although the proximity of the C iv doublet in wavelength
  space introduces an inherent degeneracy with the O i and Si ii pair, high-resolution
  HIRES data reveal sufficient kinematic substructure to warrant an additional component
  at a mean velocity offset of voffset = –85.4 km s−1. We interpret this component
  as O i, supported by the aforementioned coincident detections. To constrain the
  system, we use the secondary lines to fix the velocity centroid of the O i component
  and perform simultaneous Voigt profile fitting across all three instruments, including
  both C iv transitions and single-component O i and Si ii models. The column density
  reported in this work corresponds to this singular O i component, and the system
  is marked with an asterisk in Table 1.\r\n\r\n3.5. O i Absorption Detection Completeness\r\nWe
  quantify the completeness of O i absorption detection around each galaxy as follows.
  For each galaxy overdensity, we compute the detection sensitivity on a pixel-by-pixel
  basis following the general framework of, e.g., K. M. Lanzetta et al. (1987), M.
  L. Bernet et al. (2010), and N. L. Mathes et al. (2017). In particular, we use a
  continuum-normalized spectral slice of ±1500 km s−1 centered on the expected O i
  absorption at the redshift of each galaxy overdensity.\r\n\r\nWe define the completeness
  in O i absorption detection in a single instrument as\r\n\r\nwhere H is the Heaviside
  step function, and is the equivalent-width cutoff at which completeness is defined.
  We adopt Å to facilitate comparison with the literature (see Section 4.2). Nσ is
  the detection threshold, which we set to Nσ = 3 by convention. The equivalent-width
  error, σEW(z), is given by\r\n\r\nwhere Δλj is the wavelength spacing between adjacent
  pixels (set by the spectrograph), and npix is the number of pixels over which the
  equivalent width is computed. We adopt the typical width of an O i absorber as npix
  for our analysis, yielding npix, X-shooter = 10, npix, HIRES = 50, and npix, FIRE
  = 11, respectively. This is a conservative approach to calculating completeness,
  as the typical width of the absorbers is larger than the resolving power of the
  instruments. We neglect continuum-fitting errors, so σEW(z) reflects only statistical
  uncertainties from the spectra. Any chip gaps or sky lines are masked. σI and Ic
  are the flux uncertainty and continuum levels of the spectrum.\r\n\r\nFor galaxy
  overdensities observed with multiple spectrographs, we define the combined completeness
  as requiring detection in at least one instrument across the relevant wavelength
  range. For example, with three or more instruments available,\r\n\r\nwhere the logical
  OR (∨) runs over all instruments i ∈ X-shooter, FIRE, HIRES. We then compute the
  mean completeness for each overdensity, and obtain the overall mean completeness,
  , by averaging over all overdensities weighted by their wavelength coverage. This
  yields a mean completeness of 91.4% for O i absorption around galaxy overdensities
  in the EIGER survey for 0.05 Å.\r\n\r\n4. Results\r\nIn this section, we present
  the variation of O i absorption around z ∼ 6 galaxies, the incidence of O i absorption
  per comoving path length that is due to O i absorption around galaxies, and the
  chemical abundance of these systems.\r\n\r\n4.1. Variation of O i Absorption around
  Galaxies\r\nSince all 40 galaxies in our sample reside within 15 identified overdensities
  (Section 3.3), we investigate whether O i absorption is preferentially detected
  in these environments. For clarity, we show the primary galaxy in each overdensity
  (as defined in Section 3.3) when presenting the radial distribution of O i absorption.
  Out of the 15 overdensities, 5 show associated O i absorption.\r\n\r\nFigure 4 shows
  the 2D spatial distribution of galaxies around these five detected systems, with
  the size depicting the virial radii of the galaxies. The galaxies are randomly distributed
  around the absorbers, and we do not find correlations with stellar mass, star formation,
  or other galaxy properties. In all five cases where we have O i detection, the galaxy
  member with the minimum impact parameter is also the member with the lowest R/Rvir.\r\n\r\nZoom
  InZoom OutReset image size\r\nFigure 4. Galaxy overdensities around z ∼ 6 O i absorbers
  in the EIGER survey. Each panel shows the background quasar (red star) at the center,
  with foreground galaxies associated with O i absorption marked as circles. Circles
  are color coded by the velocity offset between the galaxy and the O i absorber,
  and circle sizes are scaled to their virial radii. Only galaxies within 300 kpc
  of the quasar are shown, highlighting the overdensities nearest the absorbers. Dashed
  circles denote projected impact parameters of 50, 100, 200, and 300 pkpc from the
  quasar.\r\n\r\nDownload figure:\r\n\r\nStandard image\r\nHigh-resolution image\r\nFigure
  5 presents 1D the radial profiles around the primary galaxies in these systems.
  The marker size indicates the number of member galaxies in each overdensity, while
  the color encodes the velocity offset from the corresponding O i absorption. Panels
  (a) and (c) show impact parameter (R), and panels (b) and (d) show the same distribution
  normalized by the halo virial radius (R/Rvir).\r\n\r\nZoom InZoom OutReset image
  size\r\nFigure 5. Radial absorption profiles of O i absorption at z ∼ 6. Each point
  marks the impact parameter to the nearest galaxy in a galaxy overdensity, with all
  detected O i absorption arising in such environments. Overdensities with O i absorption
  are shown as circles, while nondetections (2σ limits) are indicated by downward
  arrows. Points are color coded by the velocity offset between the galaxy redshift
  and the O i absorber, and their size scales with the number of members in the overdensity.
  The mean uncertainty on the virial radius is ΔRvir = 1.81 kpc. Top panels: rest-frame
  equivalent width vs. impact parameter (R, left) and normalized impact parameter
  (R/Rvir, right). Bottom panels: same as above, but for O i column densities.\r\n\r\nDownload
  figure:\r\n\r\nStandard image\r\nHigh-resolution image\r\nMost of the O i detections
  occur within R ≲ 175 kpc of the primary galaxy, corresponding to an incidence rate
  of , estimated using 68% Wilson score confidence intervals. Within 100 kpc, the
  incidence is , increasing to for R ≤ 50 kpc. Conversely, at R  >  150 kpc, an incidence
  of is measured. Within 300 kpc of the primary galaxy, the overall incidence is ,
  with values of for 50 ≤ R ≤ 300 kpc and for 100 ≤ R ≤ 300 kpc. These rates are substantially
  weaker than the incidence of strong Mg ii absorption at z ∼ 6 (R. Bordoloi et al.
  2024), indicating that neutral oxygen preferentially traces a different phase of
  the circumgalactic environment.\r\n\r\nPanels (b) and (d) reveal that O i absorbers
  are typically located at much larger R/Rvir than strong low-ionization tracers such
  as Mg ii. This distinction suggests that O i absorption at z ∼ 6 is not confined
  to the CGM of an individual galaxy but instead reflects the larger-scale gaseous
  environment of galaxy overdensities—what might be described as “cosmic ecosystems”
  rather than isolated halos.\r\n\r\nComparing the stellar mass distributions of overdensities
  with and without O i absorption, we find no statistically significant differences.
  The mean stellar mass of galaxies in overdensities with absorption is , compared
  to for those without.\r\n\r\nWe emphasize that at z ∼ 6, the large R/Rvir values
  of these halos imply that O i absorption typically arises at scales well beyond
  the typical CGM [R/Rvir ≲ 1] (J. Tumlinson et al. 2017) probed by low-ionization
  lines at lower redshift. Unlike the low-z CGM, where metal absorbers are closely
  tied to the halo of a single massive galaxy, the O i absorbers studied here appear
  to be linked to extended networks of low-mass, young galaxies embedded in shared
  overdensities. Over cosmic time, these small systems and their diffuse gaseous environments
  are expected to merge and evolve into more massive systems, eventually forming Milky
  Way–like galaxies by z = 0. Thus, neutral O i at reionization is best understood
  not as a tracer of isolated galactic halos, but as a probe of the extended environments
  that connect galaxies, gas, and structure growth in the early Universe.\r\n\r\nBecause
  O i absorption traces dense, neutral gas and requires only minimal ionization corrections,
  we can place a conservative lower limit on the total oxygen mass associated with
  the EIGER galaxies. These estimates are intended as simple, order-of-magnitude calculations
  to provide intuition about the amount of neutral oxygen relative to the observed
  galaxy properties. Following the formalism of R. Bordoloi et al. (2014), the oxygen
  mass contained in an annulus bounded by Rin and Rout is\r\n\r\nwhere Rin and Rout
  are the inner and outer radii of the annulus, fc is the covering fraction (the incidence
  of O i detections in that bin), 〈NOI〉 is the mean O i column density in the annulus,
  and mO = 16 mp is the mass of an oxygen atom, with mp being the mass of a proton.
  For a conservative lower limit, we adopt fion = 1, where fion is the ionization
  fraction (i.e., we assume that all of the oxygen mass is in the neutral phase).
  We justify this assumption as follows: O i is tightly coupled to H i through charge
  exchange, and in dense gas, the required ionization correction is minor, even under
  strong radiation fields such as the Fermi Bubbles (R. Bordoloi et al. 2017; T. Ashley
  et al. 2022). Under a typical metagalactic UV background, it is generally less than
  10% and does not significantly alter the inferred oxygen abundance. In addition,
  oxygen does not deplete onto dust grains, unlike Si or C. For these reasons, we
  consider the approximation NO,total ≈ NOI to be valid for our analysis.\r\n\r\nWe
  compute masses in two radial bins: 0–100 kpc and 100–300 kpc, respectively. The
  total oxygen mass within radius is obtained by summing the masses of the constituent
  annuli. This yields a strict lower limit of  ≳  2 ×106M⊙ on the total oxygen mass
  within = 300 kpc.\r\n\r\nWe compare the inferred oxygen mass in the extended environment
  of our sample with the total oxygen content in the interstellar medium (ISM) of
  a typical z ∼ 6 galaxy. In computing this quantity, we make the following assumptions.
  First, we assume that all galaxies in this work lie on the typical z ∼ 6 mass–metallicity
  relation, derived by stacking [O iii] emitter galaxies (J. Matthee et al. 2022),
  and we adopt the gas-phase metallicity from this relation for each galaxy.\r\n\r\nBecause
  direct measurements of the total H2 and H i gas content are not available at these
  redshifts, we extrapolate the molecular gas fraction from L. J. Tacconi et al. (2018)
  to z ∼ 6 and adopt a conservative H i gas fraction based on local observations (S.
  Huang et al. 2012). We note that high-z galaxies are likely more gas-rich than their
  local counterparts; therefore, the inferred ISM gas masses should be considered
  lower limits.\r\n\r\nWith these caveats in mind, we estimate the total ISM oxygen
  mass as\r\n\r\nwhere we define . Here, XO is the gas-phase oxygen abundance derived
  from the mass–metallicity relation, and and MHI are the molecular and atomic gas
  masses derived as described above. For our galaxy sample, the mean stellar mass
  is , corresponding to an ISM oxygen mass of MO,ISM  ≳  6.8 × 105 M⊙. Remarkably,
  this indicates that the typical ISM oxygen content is comparable to the oxygen mass
  in the extended environment around these galaxies, implying that the circumgalactic
  or intergalactic medium at z ∼ 6 contains a substantial reservoir of oxygen. While
  these estimates depend on several specific assumptions, the inferred ISM and CGM
  oxygen masses appear to be of similar magnitude.\r\n\r\nAt low redshift, both L*
  and sub-L* galaxies similarly exhibit substantial metal content in their circumgalactic
  medium (CGM), with the CGM containing at least as much metal as the ISM of their
  host galaxies (J. Tumlinson et al. 2011; R. Bordoloi et al. 2014). A key difference
  is that, whereas the low-z CGM is typically gravitationally bound to its host galaxy,
  at high redshift much of the gas exists in the extended environment of galaxy overdensities
  rather than being bound to individual galaxies (see also R. Bordoloi et al. 2024).\r\n\r\n4.2.
  Expected Incidence of O i Absorbers\r\nIn this section, we report the expected redshift-path
  incidence of O i absorbers around [O iii] emitting galaxies detected in the EIGER
  survey. We define the number of O i absorbers associated with each galaxy overdensity
  per unit absorption distance above a detection threshold (WOI ≥ 0.05Å ) as\r\n\r\n\r\nHere,
  Rmax = 300 kpc is the survey radius adopted in this work, and is the mean O i absorber
  detection completeness (see Section 3.5; 91.4% for WOI ≥ 0.05 Å). nOD is defined
  as\r\n\r\n\r\nngal is the number density of [O iii] emitting galaxies per unit comoving
  volume, estimated from the [O iii] luminosity function at z ∼ 6. And 〈overdensity
  size〉 is simply the mean number of galaxies in each galaxy overdensity empirically
  measured to be 〈overdensity size〉 = 2.67. fc is the O i absorption covering fraction
  within 300 kpc for WOI ≥ 0.05 Å (). (The covering fraction within 300 kpc is independent
  of the definition of the primary member.)\r\n\r\nTo estimate the number density
  of galaxies per unit comoving volume (ngal), we integrate the z ∼ 6 [O iii] luminosity
  function derived from the EIGER survey, with a minimum luminosity of Lmin = 1042.2
  ergs−1 (J. Matthee et al. 2022):\r\n\r\n\r\nwhere Φ(L) is the [O iii] luminosity
  function defined as\r\n\r\n\r\nHere, Φ* is the characteristic number density, L*
  is the characteristic luminosity, and α is the faint-end slope. We adopt Φ* = 10−4.02
  Mpc−3, L* = 1043.33 ergs−1, and α = −1.92, values derived from detailed completeness
  calculations of the EIGER [O iii]-emitting galaxies (R. Mackenzie et al. 2026, in
  preparation). This results in ngal = 8.586 × 10−04 Mpc−3. Using these values in
  Equation (6) yields , for O i absorption around galaxies at a mean galaxy redshift
  of 6.001.\r\n\r\nAll calculations in this section are performed with a detection
  threshold of WOI > 0.05Å, since the survey is highly complete down to this equivalent-width
  limit and because most of the literature surveys we compare against adopt the same
  threshold.\r\n\r\nWe emphasize that this approach differs from blind QSO absorption-line
  surveys (e.g., G. D. Becker et al. 2006; L. Christensen et al. 2023; A. M. Sebastian
  et al. 2024; A. Sodini et al. 2024), where all O i absorption-line systems along
  the full quasar redshift path are counted to compute dN/dX directly, under the assumption
  that the absorber cross section is contributed by all galaxies. In contrast, our
  method begins by defining the absorber cross section around galaxy overdensities
  and then estimating the corresponding contribution to dN/dX from the galaxy overdensities
  under study. Comparing these two approaches constrains the fraction of O i (or other
  metal) absorption systems that arise in the extended halos of these galaxies. Figure
  6 shows these literature measurements as colored circles, while the contribution
  to dN/dX around z ∼ 6 galaxy overdensities is indicated by an orange square. To
  quantify the relative contribution of galaxy overdensities, we fit a single power
  law to all literature values at z  >  4 (blue dashed line with shaded region). Comparing
  the dN/dX inferred from quasar absorption-line surveys at z ∼ 6 to our measured
  value, we find that O i dN/dX around z ∼ 6 EIGER galaxy environments account for
  35% ± 16% of all O i absorption-line systems detected in blind quasar surveys. This
  implies that [O iii] emitting galaxy environments at z ∼ 6 contribute roughly one-third
  of the O i absorption systems at that epoch. Although the EIGER QSOs were selected
  to have known absorption systems along their sight lines, the EIGER galaxy survey
  around these sight lines is highly complete and independent of this selection. Therefore,
  we can create an unbiased measurement of the absorption cross section (dN/dX) around
  these galaxy overdensities. We justify this comparison to blind absorption surveys
  in Section 2.\r\n\r\nZoom InZoom OutReset image size\r\nFigure 6. Incidence of O
  i absorbers per comoving path length, dN/dX. Values are derived from all O i absorption
  lines along high-z quasar sight lines. Literature measurements are shown from G.
  D. Becker et al. (2019; blue circles), L. Christensen et al. (2023; purple circles),
  and A. M. Sebastian et al. (2024; red circles). The blue dashed line shows a power-law
  fit to the z  > 4 literature data, with the shaded region marking the 1σ uncertainty.
  The orange square gives the inferred dN/dX when including only O i absorption associated
  with host galaxies (see Section 4.2), indicating that only ∼35% of O i absorbers
  at z ∼ 6 arise in galaxy environments.\r\n\r\nDownload figure:\r\n\r\nStandard image\r\nHigh-resolution
  image\r\nThis in turn suggests that ∼65% of the O i absorption systems detected
  along quasar sight lines at the epoch of reionization are not associated with [O
  iii] emitter galaxy overdensities. Either the bulk of O i absorption resides in
  isolated pockets of neutral IGM gas outside extended galaxy halos, or it is contributed
  by even lower-luminosity, less massive galaxies () that fall below the EIGER detection
  limits. Such systems may also be detected in galaxy halos that are oxygen-poor or
  lack strong [O iii] emission features, since our galaxy identifications primarily
  rely on [O iii] and Hβ emission lines.\r\n\r\nSimulations suggest that O i absorption
  should primarily arise in galactic halos rather than in the diffuse IGM, since metals
  are not widely distributed and ionization fronts propagate more quickly than metals
  (B. D. Oppenheimer et al. 2009; K. Finlator et al. 2013). Very low-mass protogalaxies
  forming their first stars may blur the boundary between IGM and CGM, enriching their
  surroundings with metals, especially if their star formation is bursty and prolific.
  If sufficiently numerous, such systems could contribute a nonnegligible fraction
  of oxygen in the early Universe.\r\n\r\nRecent observations support this possibility.
  K. Nakajima et al. (2025) reported a strongly lensed (μ = 98) galaxy at z ≈ 6.6
  with a stellar mass upper limit of only ∼2700 M⊙ and a dynamical mass ∼100 times
  larger. The source shows emission and may be dominated by Population III stars.
  This provides direct evidence of oxygen production in extremely low-mass systems.
  A population of such ultralow-mass dwarf galaxies could plausibly account for the
  ∼65% of O i absorbers detected along quasar sight lines that are not associated
  with galaxies in our survey. Our sensitivity limit of (J. Matthee et al. 2022) means
  such dwarfs remain undetected. However, these galaxies may lack a dense enough CGM
  to shield neutral O i from the UV background; thus, even if they produce oxygen,
  neutral O i may not survive (K. Finlator et al. 2013). Interestingly, simulations
  by S. Kušmić et al. (2024) showed that the contribution of metal ions, including
  O i, is only weakly correlated with the stellar mass of the galaxy, and the contribution
  is indistinguishable when comparing across galaxy luminosities. Additionally, they
  argue that due to the steep slope of the stellar mass function, lower-mass, lower-luminosity
  galaxies contribute the majority of absorbing metal ions in the CGM and IGM.\r\n\r\nLater
  simulation works by C. Doughty et al. (2018); C. Doughty & K. Finlator (2019) suggest
  that as the Epoch of Reionization continues, O i will reduce overall but be preferentially
  found in higher column density regions close to halos, as they have a better ability
  to self-shield from ionizing photons. In fact, C. Doughty & K. Finlator (2019) found
  that at z = 6 the majority of O i mass was located in the ISM of galaxies and only
  ∼30% of O i mass was found in the CGM and IGM. In contrast, in this work, we find
  that a substantial (comparable to ISM mass) oxygen mass exists outside galaxies;
  however, we stress that our mass measurements are order-of-magnitude estimates.
  A more direct comparison between these simulations and this work is the covering
  fraction. They report ∼5%–25% covering fraction at z = 6 for O i with out to 102
  pkpc. In the EIGER survey, we report a covering fraction using a minimum threshold
  of WOI ≥ 0.05 Å (or approximately logNOI > 13.7 cm−2) out to 300 kpc. Given the
  large measurement uncertainties and different definitions of associated galaxies,
  the covering fractions are qualitatively consistent between the two works.\r\n\r\nThe
  simulations of K. Finlator et al. (2013) suggest that O i absorbers with are most
  likely associated with halos of mass ∼109 M⊙, where self-shielding is efficient.
  They further predict that only ∼13% of all O i absorption systems should reside
  in halos of 1010–1011.5 M⊙, the mass range of galaxies probed in this work. By contrast,
  we measure a higher dN/dX of ∼35%, but qualitatively our results remain consistent
  with the idea that roughly two-thirds of O i absorbers are associated with undetected
  ∼109 M⊙ halos. Future observations targeting lower-luminosity galaxies and higher
  redshifts will be essential to test these predictions.\r\n\r\nAfter the initial
  posting of this manuscript, G. Pruto et al. (2026) presented a simulation-based
  study that adopts the observational compilation shown in Figure 6 as a reference
  and explores the corresponding behavior in numerical models. Our minimum Wr threshold
  corresponds to ; therefore, the most appropriate comparison is with their sample.
  Qualitatively, the two sets of results are in agreement. A key distinction lies
  in the definition of galactic environment. We characterize environments in terms
  of galaxy overdensities, which we argue is the most appropriate description of where
  O i absorbers reside at z ≈ 6, whereas G. Pruto et al. (2026) define the CGM based
  on the virial radii of halos with . As a result, a direct quantitative comparison
  is not straightforward. Nevertheless, their inferred dN/dX corresponds to an incidence
  rate of O i absorbers of ∼58% ± 11%, which is consistent with our measurements within
  the uncertainties. This agreement may depend on their tighter constraints on absorber
  location and on the adopted minimum Wr or threshold. G. Pruto et al. (2026) further
  report that ∼80% of O i absorbers with are predicted to lie outside the CGM, also
  in qualitative agreement with our findings.\r\n\r\n4.3. Metal Abundances in O i
  absorber population\r\nIn this section, we examine the relative abundances and metal
  column density ratios of O i absorbers associated with galaxy overdensities, and
  compare them to O i absorbers without any associated galaxies. To enable a fair
  comparison, we restrict our analysis to the six quasar sight lines from the EIGER
  survey and identify all O i absorption systems in the redshift range where an [O
  iii]-emitting galaxy could be detected (5.9 ≤ zOI ≤ zQSO). We detect 10 O i absorption
  systems in total, 5 of which have associated galaxy overdensities (Table 1) and
  5 are without any identified galaxy association.\r\n\r\nWe first compare the ionic
  column density ratios in these two populations. Figure 7 (top panels) shows the
  ratios of NSiII/NOI and NCII/NOI as functions of NOI. Systems located near galaxy
  overdensities (blue squares) systematically show higher Si ii/O i and C ii/O i ratios
  compared to systems without associated overdensities (green crosses). This offset
  could reflect differences in metallicity, or alternatively, the effects of ionization.
  However, since the majority of these O i systems lack detectable high-ionization
  transitions, it is reasonable to assume that neutral oxygen and singly ionized silicon
  and carbon dominate the elemental reservoirs (see also L. Christensen et al. 2023).
  In those cases, the observed ionic ratios provide direct tracers of relative abundances.
  Two exceptions are found along the J0100 and J1148 sight lines, which do exhibit
  Si iv and C iv absorption; for these systems, the degeneracy between chemical composition
  and ionization corrections cannot be resolved. It is plausible that, on average,
  O i absorption found near galaxy overdensities has higher elemental abundance ratios
  as the non-galaxy-associated absorbers either have significantly lower-luminosity
  and less massive galaxies, which would be expected to produce less metal, or there
  are truly no galaxies associated with this population.\r\n\r\nZoom InZoom OutReset
  image size\r\nFigure 7. Top panels: ion column density ratios of Si ii to O i (left)
  and C ii to O i (right) as a function of the detected O i column densities, shown
  in logarithmic units. Absorbers associated with galaxy overdensities (blue squares)
  and those not associated with galaxy overdensities (green crosses) exhibit clear
  variations in their column density ratios. One sight line with no detected Si ii
  absorption is indicated with a downward arrow, where a 2σ upper limit on the column
  density is assumed. Bottom panel: relative abundance ratios [Si/O] vs. [C/O] with
  respect to solar abundance. The background contours show the IMF-weighted abundance
  distributions from Population II and Population III SN models. Orange and pink circles
  indicate abundance ratios predicted by low-metallicity Population II SN models.
  The abundance ratios of several O i absorbers associated with galaxy overdensities
  are consistent with a significant contribution from Population III SNe.\r\n\r\nDownload
  figure:\r\n\r\nStandard image\r\nHigh-resolution image\r\nWe next compute the relative
  abundances [Si/O] and [C/O], presented in the bottom panel of Figure 7. These abundances
  are defined as\r\n\r\n\r\nwhere NSi,tot and NC,tot represent the summed ionic column
  densities. For systems with detected high-ionization absorption (C iv or Si iv),
  we include both low- and high-ionization states (NSiII + NSiIV and NCII + NCIV),
  marking these measurements as lower limits (blue upward arrows in the figure) since
  additional undetected ionization states may be present. We note that in the two
  systems (a and b) exhibiting high-ionization absorption (see Table 1), the low-
  and high-ionization absorption components are offset in velocity by Δv = –66 km
  s−1 and –69 km s−1, respectively. These offsets are within ≲1.5 resolution elements
  of the instrument, and therefore we cannot statistically distinguish whether the
  low- and high-ionization components are kinematically distinct. This is suggestive
  of a complex, potentially multiphase gaseous environment in the outskirts of these
  galaxies. For systems without detected high-ionization transitions, we adopt NSiII
  and NCII as the dominant contributions. We assume NO,tot ≈ NOI, as ionization corrections
  for O i are typically very small at high O i column densities (see Section 4.1 for
  discussion). We adopt solar reference abundances of = −1.18 ± 0.05 and (M. Asplund
  et al. 2021).\r\n\r\nFigure 7 (bottom panel) reveals a striking separation between
  O i absorbers associated with galaxy overdensities (blue squares) and those without
  galaxy associations (green crosses). Systems near galaxies are found in the upper
  left of the [Si/O]–[C/O] diagram, with two cases exhibiting high-ionization absorption
  (lower abundance limits, blue arrows), which will offset them even farther from
  the locus of nongalaxy absorbers.\r\n\r\nWe place these abundance patterns in the
  context of chemical enrichment models from different stellar populations (see also
  L. Christensen et al. 2023). The contours and colored circles in Figure 7 highlight
  predicted [C/O] and [Si/O] yields from representative Population II and Population
  III supernova (SN) models, weighted by a Salpeter IMF. The weighting by number of
  stars predicted by an IMF is done to represent the typical observational relative
  abundance distribution in these models. These are not exhaustive but span the dominant
  theoretical expectations.\r\n\r\nPopulation III pair-instability SNe (PISNe; 140–260M⊙;
  light green contours) predict narrow [C/O] ≃ –0.7 and high [Si/O] values, inconsistent
  with any of our observed systems. Such PISNe are also expected to produce substantial
  amounts of iron-peak elements, leading to elevated Fe/O ratios (J. Tumlinson et
  al. 2004). We will investigate Mg ii and Fe ii absorption associated with z ∼ 6
  galaxies in an upcoming work (J. Higginson et al. 2026, in preparation). Models
  of massive Population III stars (10–100 M⊙) produce both hypernovae (gray contours;
  E > 1052 erg) and standard core-collapse SNe (light blue contours; A. Heger & S.
  E. Woosley 2010), spanning a broader region overlapping some of our measurements.
  We also consider Population II models: low-metallicity core-collapse SNe from M.
  Limongi & A. Chieffi (2018; orange contours, Population II SNe (L18)) occupy −0.25
  ≲ [C/O] ≲ 0.5 and −0.4 ≲ [Si/O] ≲ 0.4, overlapping parts of the Population III range,
  while the C. Kobayashi et al. (2006) yields (pink and orange circles for hypernovae
  and SNe, respectively) tend toward lower [Si/O] and [C/O].\r\n\r\nAgainst these
  predictions, the contrast between the two absorber populations is clear. All nongalaxy
  O i systems (green crosses) fall at the lower left of the diagram, consistent with
  enrichment from both Population II and/or Population III SNe models. This regime
  also overlaps with ultra-metal-poor damped Lyman-α absorbers (DLAs) at z ∼ 3 (−0.5
  ≲ [Si/O] ≲ 0, −0.5 ≲ [C/O] ≲ 0; R. Cooke et al. 2011; R. J. Cooke et al. 2017).
  By contrast, three of the five galaxy-associated absorbers lie well outside both
  the locus of Population II predictions and the locus of ultra-metal-poor DLA relative
  abundances. These systems are best explained if Population III enrichment played
  a major role, suggesting that they may represent a distinct population from ultra-metal-poor
  DLA host galaxies (see also A. Sodini et al. 2024).\r\n\r\nSimilarly, A. Sodini
  et al. (2024) found that the [C/O] and [Si/O] abundance ratios for high-redshift
  O i absorber systems are consistent with significant enrichment from Population
  III, with the majority explained by ≥ 30% contributions. They also find a large
  spread in the implied Population III contribution; several systems require ≥ 60%
  contributions, while others fall significantly below the average. This is a strong
  indication of heterogeneous chemical enrichment by Population III stars during the
  epoch of reionization.\r\n\r\nThe absorber toward J159 at z = 5.91 shows [C/O] =
  0.20, [Si/O] = 0.55 (labeled c), and no evidence of high-ionization absorption.
  These abundance ratios are best explained by contributions from Population III SNe
  or hypernovae. The absorber toward J1030 at z = 6.07 ([C/O] = −0.38, [Si/O] = 0.65,
  labeled e) also shows no high-ionization absorption. The relative abundances in
  these two predominantly neutral systems are therefore quite robust. Interestingly,
  the J1030 system is not reproduced by any single Population II or Population III
  model, suggesting that it may have experienced multiple enrichment episodes, producing
  unusually high [Si/O] coupled with low [C/O].\r\n\r\nTwo cases require special mention.
  The absorber toward J0100 at z = 5.94 shows [C/O] ≥ 0.42 and [Si/O] ≥ 0.67 (labeled
  a), with strong C iv and Si iv absorption. Since these are lower limits, ionization
  corrections (for carbon and silicon) would push this system further upward, strengthening
  the case for Population III contributions. This observation is best explained by
  contributions from Population III hypernovae models. The J1148 absorber at z = 6.01
  shows [C/O] ≥ 0.02 and [Si/O] ≥ 0.20 (labeled b), again with high-ionization lines;
  correcting for ionization would shift it diagonally upward in the diagram. Taken
  together, four of the five galaxy-associated absorbers show relative abundances
  that cannot be explained by Population II models alone, pointing instead to significant
  enrichment from Population III SNe or hypernovae.\r\n\r\nThus, most galaxy-associated
  O i systems carry signatures of Population III enrichment. These results suggest
  that O i systems tracing galaxy overdensities at z ∼ 6 carry chemical signatures
  of Population III enrichment, likely from multiple SNe events rather than a single
  explosion. The elevated [Si/O] and [C/O] ratios in these environments imply that
  [O iii]—emitting galaxies at this epoch may retain significant Population III contributions,
  and that metal-line absorbers around such overdensities provide a powerful probe
  of the first generations of stars.\r\n\r\n4.4. Comparison to the z ≈ 0 Universe\r\nAt
  z ∼ 6, we cannot directly detect the local environments where O i absorption systems
  arise. However, at z ∼ 0, there are instances where the environments giving rise
  to O i absorption are well understood. Here, we discuss several such instances and
  compare them with z ∼ 6 observations. Such comparisons can provide useful insights
  regarding the nature of local environments within which O i systems exist at high
  z. We focus on O i absorption associated with local blue compact dwarf (BCD) galaxies
  and the Magellanic Stream (MS), owing to their low metallicities and well-characterized
  local environments.\r\n\r\nThere is evidence that O i is found preferentially in
  the H ii regions of local BCDs, with their H i regions being pristine or showing
  very low levels of enrichment (D. Kunth et al. 1994; M. Pettini & K. Lipman 1995;
  T. X. Thuan & Y. I. Izotov 1997). In particular, T. X. Thuan & Y. I. Izotov (1997)
  report that, to varying degrees, low-ionization metal lines (e.g., O i and Si ii)
  reside in the ionized hydrogen regions of BCDs rather than the neutral hydrogen
  regions. They report . Compared to the z ∼ 6 measurements in panel (b) of Figure
  7, the O i column density of the BCD is comparable, but the Si ii-to-O i ratio is
  inconsistent with the z ∼ 6 values. On the other hand, V. Lebouteiller et al. (2004)
  find that the H ii region of the BCD I Zw 36 has significantly higher metallicity
  than its H i region. However, inferred from velocity offsets, O i in that system
  is associated with the H i region. The reported O i column density of that BCD ()
  is approximately 2 orders of magnitude higher than the z ∼ 6 sample. Given the available
  evidence, we conclude that the observed z ∼ 6 O i absorption systems are not similar
  to local BCD systems and are unlikely to originate in H ii regions.\r\n\r\nThe MS
  is a filamentary trail of gas in the Local Group. Multiphase absorption-line observations
  from A. J. Fox et al. (2013) and P. Richter et al. (2013) support the argument that
  the MS was tidally stripped from the Magellanic Clouds approximately 2 Gyr ago.
  Inferences from the abundance ratios of alpha elements in the MS are best explained
  by contributions from massive stars via Type II SNe when the MS was still part of
  the Magellanic Clouds, although the presence of significant dust depletion makes
  further inferences into enrichment history difficult. An additional study of the
  MS (N. Kumari et al. 2015) reports abundance ratios in a compact high-velocity cloud
  (CHVC) that is believed to be associated with the MS. They find [Si/O] = −0.04 ±
  0.25 and [C/O] = −0.09 ± 0.23. When comparing these with the z ∼ 6 observations
  presented in panel (c) of Figure 7, we see that the CHVC abundance ratios occupy
  a different region of parameter space than the z ≈ 6 O i absorbers within identified
  galactic overdensities showing Population III enrichment signatures (points a, c,
  e). This discrepancy provides evidence that these z ∼ 6 O i absorbers with Population
  III enrichment signatures are unlikely to arise in environments similar to the MS.
  However, the ionic ratios of O i absorption systems not associated with galaxy overdensities
  (green crosses) are consistent with this measurement. With the caveat that MS measurements
  do contain dust depletion effects, it is plausible that such O i systems may trace
  IGM structures with densities comparable to those of the MS at high z.\r\n\r\n5.
  Conclusion\r\nWe have conducted a survey of neutral O i absorption systems associated
  with galaxies during the Epoch of Reionization in the EIGER survey, using six high-redshift
  quasars as background sight lines. Our study focuses on the distribution of neutral
  oxygen around -emitting galaxies within 300 kpc and their connection to galaxy overdensities.
  The main results are as follows:\r\n\r\n1.  \r\nFive O i absorbers are detected
  among 15 galaxy overdensities, spanning z = 5.91–6.16 with a mean redshift 〈z〉 =
  6.001.\r\n2.  \r\nThe O i covering fraction within 300 kpc of the closest galaxy
  is , with for R ≤ 50 kpc and for 50 ≤ R ≤ 300 kpc, where the covering fraction is
  calculated using the galaxy in an overdensity with the lowest impact parameter.\r\n3.
  \ \r\nWe find  ≳ 2 × 106 M⊙ of oxygen within 300 kpc, a mass comparable to the ISM
  oxygen content of z ∼ 6 galaxies—assuming minimal ionization correction (see Section
  4.1 for justification)—suggesting that the circumgalactic environment already holds
  metal reservoirs rivaling those inside galaxies.\r\n4.  \r\nO i absorbers are detected
  as far as ∼8 R/Rvir from galaxies, in sharp contrast to low-ionization tracers such
  as Mg ii, which are typically confined within ≲R/Rvir.\r\n5.  \r\nThe incidence
  of galaxy-associated O i absorbers is \r\n \r\n. Comparison with blind quasar absorption-line
  surveys indicates that these systems represent only ∼35% of all O i absorbers at
  z ∼ 6, implying that the majority likely arise in lower-mass galaxies below current
  detection limits or in dense neutral IGM pockets that survived reionization.\r\n6.
  \ \r\nRelative abundance and ionization ratios ([Si/O], [C/O]) show that galaxy-associated
  O i systems are chemically enriched, with signatures in several systems consistent
  with contributions from Population III stars. In particular, we identify four galaxy
  overdensities whose outskirts exhibit absorption-line abundance ratios consistent
  with contributions from Population III nucleosynthesis.\r\n\r\nOverall, these results
  indicate that neutral O i at z ∼ 6 primarily traces extended overdensity environments
  rather than the CGM of individual halos. Moreover, the chemical abundance patterns
  of galaxy-associated O i systems suggest contributions from Population III stars,
  likely arising from multiple SNe events rather than a single explosion. The elevated
  [Si/O] and [C/O] ratios imply that [O iii]-emitting galaxies at this epoch may retain
  significant Population III enrichment. Consequently, metal-line absorbers around
  such overdensities offer a powerful probe of the first generations of stars, providing
  insight into early chemical enrichment and the role of primordial stellar populations
  in shaping galaxy evolution during the Epoch of Reionization.\r\n\r\nAcknowledgments\r\nThis
  work is based on observations made with the NASA/ESA/CSA James Webb Space Telescope.
  The data were obtained from the Mikulski Archive for Space Telescopes at the Space
  Telescope Science Institute, which is operated by the Association of Universities
  for Research in Astronomy, Inc., under NASA contract NAS 5-03127 for JWST. These
  observations are associated with program #1243. This paper includes data gathered
  with the 6.5 m Magellan Telescopes located at Las Campanas Observatory, Chile. This
  work was supported by a NASA Keck PI Data Award, administered by the NASA Exoplanet
  Science Institute. Data presented herein were obtained at the W. M. Keck Observatory
  from telescope time allocated to the National Aeronautics and Space Administration
  (NASA) through the agency’s scientific partnership with the California Institute
  of Technology and the University of California. The Observatory was made possible
  by the generous financial support of the W. M. Keck Foundation. The JWST data presented
  in this article were obtained from the Mikulski Archive for Space Telescopes (MAST)
  at the Space Telescope Science Institute. The specific observations analyzed can
  be accessed via DOI: 10.17909/y94q-h212. This work was partially supported by the
  National Science Foundation under grant No. AST-2108931. Jorryt Matthee and Ivan
  Kramarenko acknowledge funding from the European Union (ERC, AGENTS, 101076224).\r\n\r\nFacilities:
  JWST - James Webb Space Telescope (NIRCam), Magellan: Baade - (FIRE), VLT: Kueyen
  - (X-shooter), and Keck: I - (HIRES, MOSFIRE).\r\n\r\nSoftware: astropy (Astropy
  Collaboration et al. 2013, 2018, 2022), rbcodes (R. Bordoloi et al. 2025), rbvfit
  (R. Bordoloi & J. Higginson 2025)."
article_number: '49'
article_processing_charge: Yes
article_type: original
arxiv: 1
author:
- first_name: Jack
  full_name: Higginson, Jack
  last_name: Higginson
- first_name: Rongmon
  full_name: Bordoloi, Rongmon
  last_name: Bordoloi
- first_name: Robert A.
  full_name: Simcoe, Robert A.
  last_name: Simcoe
- first_name: Jorryt J
  full_name: Matthee, Jorryt J
  id: 7439a258-f3c0-11ec-9501-9df22fe06720
  last_name: Matthee
  orcid: 0000-0003-2871-127X
- first_name: Daichi
  full_name: Kashino, Daichi
  last_name: Kashino
- first_name: Ruari
  full_name: Mackenzie, Ruari
  last_name: Mackenzie
- first_name: Ivan
  full_name: Kramarenko, Ivan
  id: 9a9394cb-3200-11ee-973b-f5ba2a8b16e4
  last_name: Kramarenko
  orcid: 0000-0001-5346-6048
- first_name: Simon J.
  full_name: Lilly, Simon J.
  last_name: Lilly
- first_name: Anna-Christina
  full_name: Eilers, Anna-Christina
  last_name: Eilers
- first_name: Rohan P.
  full_name: Naidu, Rohan P.
  last_name: Naidu
- first_name: Minghao
  full_name: Yue, Minghao
  last_name: Yue
citation:
  ama: Higginson J, Bordoloi R, Simcoe RA, et al. EIGER. VIII. First stars signatures
    in the connection between O I absorption and galaxies in the epoch of reionization.
    <i>The Astrophysical Journal</i>. 2026;999(1). doi:<a href="https://doi.org/10.3847/1538-4357/ae3e81">10.3847/1538-4357/ae3e81</a>
  apa: Higginson, J., Bordoloi, R., Simcoe, R. A., Matthee, J. J., Kashino, D., Mackenzie,
    R., … Yue, M. (2026). EIGER. VIII. First stars signatures in the connection between
    O I absorption and galaxies in the epoch of reionization. <i>The Astrophysical
    Journal</i>. IOP Publishing. <a href="https://doi.org/10.3847/1538-4357/ae3e81">https://doi.org/10.3847/1538-4357/ae3e81</a>
  chicago: Higginson, Jack, Rongmon Bordoloi, Robert A. Simcoe, Jorryt J Matthee,
    Daichi Kashino, Ruari Mackenzie, Ivan Kramarenko, et al. “EIGER. VIII. First Stars
    Signatures in the Connection between O I Absorption and Galaxies in the Epoch
    of Reionization.” <i>The Astrophysical Journal</i>. IOP Publishing, 2026. <a href="https://doi.org/10.3847/1538-4357/ae3e81">https://doi.org/10.3847/1538-4357/ae3e81</a>.
  ieee: J. Higginson <i>et al.</i>, “EIGER. VIII. First stars signatures in the connection
    between O I absorption and galaxies in the epoch of reionization,” <i>The Astrophysical
    Journal</i>, vol. 999, no. 1. IOP Publishing, 2026.
  ista: Higginson J, Bordoloi R, Simcoe RA, Matthee JJ, Kashino D, Mackenzie R, Kramarenko
    I, Lilly SJ, Eilers A-C, Naidu RP, Yue M. 2026. EIGER. VIII. First stars signatures
    in the connection between O I absorption and galaxies in the epoch of reionization.
    The Astrophysical Journal. 999(1), 49.
  mla: Higginson, Jack, et al. “EIGER. VIII. First Stars Signatures in the Connection
    between O I Absorption and Galaxies in the Epoch of Reionization.” <i>The Astrophysical
    Journal</i>, vol. 999, no. 1, 49, IOP Publishing, 2026, doi:<a href="https://doi.org/10.3847/1538-4357/ae3e81">10.3847/1538-4357/ae3e81</a>.
  short: J. Higginson, R. Bordoloi, R.A. Simcoe, J.J. Matthee, D. Kashino, R. Mackenzie,
    I. Kramarenko, S.J. Lilly, A.-C. Eilers, R.P. Naidu, M. Yue, The Astrophysical
    Journal 999 (2026).
date_created: 2026-10-12T12:15:27Z
date_published: 2026-02-24T00:00:00Z
date_updated: 2026-10-14T07:58:57Z
day: '24'
ddc:
- '520'
department:
- _id: JoMa
- _id: GradSch
doi: 10.3847/1538-4357/ae3e81
external_id:
  arxiv:
  - '2510.05220'
file:
- access_level: open_access
  checksum: 1b8caf640c32d0d27a078bfb27ffcf8a
  content_type: application/pdf
  creator: dernst
  date_created: 2026-10-14T07:50:51Z
  date_updated: 2026-10-14T07:50:51Z
  file_id: '23136'
  file_name: 2026_AstrophysicalJour_Higginson.pdf
  file_size: 2808503
  relation: main_file
  success: 1
file_date_updated: 2026-10-14T07:50:51Z
fulldoi: https://doi.org/10.3847/1538-4357/ae3e81
has_accepted_license: '1'
intvolume: '       999'
issue: '1'
language:
- iso: eng
month: '02'
oa: 1
oa_version: Published Version
project:
- _id: bd9b2118-d553-11ed-ba76-db24564edfea
  grant_number: '101076224'
  name: Young galaxies as tracers and agents of cosmic reionization
publication: The Astrophysical Journal
publication_identifier:
  eissn:
  - 1538-4357
  issn:
  - 0004-637X
publication_status: published
publisher: IOP Publishing
quality_controlled: '1'
scopus_import: '1'
status: public
title: EIGER. VIII. First stars signatures in the connection between O I absorption
  and galaxies in the epoch of reionization
tmp:
  image: /images/cc_by.png
  legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode
  name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)
  short: CC BY (4.0)
type: journal_article
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 999
year: '2026'
...