Discovery of Weak O VI Absorption in Underdense Regions of the Low-Redshift Intergalactic Medium

TL;DR

This study tackles the problem of missing baryons by probing metal enrichment in the underdense low-redshift IGM using O VI. By stacking the O VI 1032, 1038 doublet at the rest-frame locations of 396 Lyα absorbers with ${ m log} N_{ m HI} < 14.5$ (and no individual O VI detections) in 82 high-S/N HST/COS spectra, the authors reveal a statistically significant O VI signal with ${ m log}(N_{ m OVI}) oughly 12.14$ and $W_r^{1032} oughly 1.7$–$2.0$ mÅ. The O VI absorption is stronger for higher H I column density and shows a marginal signal for broad Lyα absorbers; Si III is detected while Si II is not. Galaxy cross-correlations indicate the metal-bearing gas largely resides in the diffuse IGM rather than CGM, yielding metallicity estimates around ${ m log}(Z/Z_igodot) oughly -1.9$ (photoionization) or $ oughly -3$ (CIE), thereby constraining the metal-enrichment history of the underdense IGM and suggesting the absence of a universal metallicity floor above $ obreak Z obreak / obreakodot obreaksim 0.01$.

Abstract

We search for weak O VI absorption in the low-redshift intergalactic medium (IGM) using 82 high signal-to-noise quasar spectra obtained with the Cosmic Origins Spectrograph on board the Hubble Space Telescope. From this dataset, we compile a clean sample of 396 intervening Lyman-alpha (Lya) absorption lines with H I column densities log (N_HI) < 14.5, all of which lack individual O VI absorption with log (N_OVI ) > 13. We perform a spectral stacking analysis at the expected location of the O VI doublet, revealing O VI absorption with a statistical significance greater than 5$σ$, and measure an equivalent width of 1.7 $\pm$ 0.3 mA, corresponding to log (N_OVI ) = 12.14 $\pm$ 0.08. The stacked O VI absorption signal associated with strong Lya absorbers (13.5 <= log N_HI < 14.5) is significantly stronger than that associated with weaker Lya absorbers (12.5 <= log N_HI < 13.5). For the subset of 81 broad Lya absorbers (BLAs; b(HI) > 45 km/s), we obtain a marginal $\sim$3 $σ$ O VI detection. Other than Si III, detected at 5$σ$, no associated metal lines are found. Cross-correlation of the Lya absorbers with galaxies indicates that 93% of these absorbers are not associated with bright galaxies within 1 Mpc, implying that the detected O VI originates in the diffuse IGM rather than the circumgalactic medium. The stacked O VI signal suggests characteristic metallicities of $\sim 0.01\,Z_{\odot}$ under photoionisation and $\sim 0.001\,Z_{\odot}$ under collisional ionisation conditions, though these estimates are model-dependent and assume that O VI and H I trace the same phase. This study provides the first observational evidence for metal absorption in low-column-density Lya systems that individually exhibit no detectable metals, placing important constraints on the metal enrichment of the underdense IGM.

Figures (6)

• Figure 1: Histogram showing the column density distribution of our final, clean 396 weak H$\;$ absorbers that defines our stacking sample to detect weak O$\;$ IGM absorption.
• Figure 2: Stacked O$\;$ absorption profiles obtained through three methods: median (top panels), S/N-weighted mean (middle panels), and 5$\sigma$ clipped mean. Vertical dashed lines represent the expected O$\;$ doublet locations, while the shaded regions encompass the $\pm$50 km s$^{-1}$ around absorption features. The left-hand panels depict the stacked profiles with pseudo-continuum (green curve), while the right-hand panels show the pseudo-continuum normalized stacks. All three methods yield O$\;$ absorption with a statistical significance of $>5\sigma$ (legends indicate the combined significance of detection for both lines in each method, see Section \ref{['sec:results']}).
• Figure 3: Similar to Fig. \ref{['fig.sample_stack']}, this figure displays stacked O$\;$ absorption profiles obtained using the S/N-weighted mean method, with results presented separately for two bins of H$\;$ column density: $12.5 < {\rm log}\,N_{\rm H\,I} <13.5$ (top panels, 3.3$\sigma$ detection) and $13.5 < {\rm log}\,N_{\rm H\,I} <13.5$ (bottom panels, 6.6$\sigma$ detection).
• Figure 4: Pseudo-continuum normalized S/N-weighted mean stacks for Si$\;$$\lambda$1193 (top) and Si$\;$$\lambda$1206 (bottom). The number of Ly$\alpha$ absorbers, the upper limit in the Rest-frame Equivalent Width (REW) and column density is given in the bottom right side of each panel.
• Figure 5: Column densities of H$\;$ and O$\;$ for individual absorbers from Danforth16 are shown in cyan circles. The weak O$\;$ absorbers obtained from our stacking analysis for the two bins of log $N_{\rm H\,I}$ are shown as red stars.