Detecting the IGM metal enrichment with the 2-point correlation function of the flux. Application to the UVES deep spectrum

TL;DR

This study tests whether the two-point correlation function (TPCF) of the transmitted flux in the CIV forest can reveal intergalactic medium (IGM) metal enrichment when dominated by very weak absorbers. Using an ultra-high-S/N UVES spectrum of HE0940-1050 at $z\sim3$, the authors perform Voigt-profile line fitting, compute the flux TPCF, and implement a deabsorption procedure to suppress circum-galactic medium (CGM) contributions, supplemented by 1000 mock realizations to estimate uncertainties. They find a clear CIV peak in the complete spectrum, but the peak vanishes once CGM-linked absorbers are removed; adding weak systems via mocks or synthetic spectra does not recover a significant IGM signal, highlighting sensitivity limits for a single line of sight. The results suggest that detecting IGM CIV enrichment with this statistical approach requires multiple lines of sight and careful treatment of spectral features, with implications for constraining IGM metallicity (e.g., [C/H] around $-3.5$ to $-3.8$ under simple enrichment models) and the inferred volume filling factors of enriched regions.

Abstract

The distribution and the abundance of metals in the intergalactic medium (IGM) have strong implications on galaxy formation and evolution models. The ionic transitions of heavy elements in quasar spectra can probe both the mechanisms and the sources of chemical pollution, but high resolution, high signal-to-noise ratio (S/N) spectra are required, as the IGM absorbers can be too weak for direct detection. In this work, we investigate the IGM metallicity, focusing on the detection of the weak absorption lines. We exploited the cosmological tool of the two-point correlation function (TPCF) and applied it to the transmitted flux in the CIV forest of the ultra-high S/N UVES spectrum of the quasar HE0940-1050 (z~3). We also "deabsorbed" the strongest circum-galactic medium (CGM) systems to reveal the underlying IGM signal. For each test, we generated 1000 mock spectra with shuffled line positions to derive an estimate for the TPCF noise level. The TPCF shows a clear peak at the characteristic velocity separation of the CIV doublet, but when deabsorbing the CGM contribution (i.e. all metal lines and CIV lines associated with logN(HI)>14.0), the peak is no longer significant at 1$σ$, although 7 weak CIV systems remain visible. Even adding up to 135 additional weak mock CIV systems (logN(HI)<14.0) to the spectrum does not produce a significant CIV peak. Eventually, when we create a synthetic spectrum with gaussian distributed noise and same S/N as the complete spectrum, we remove the signal caused by the spectral intrinsic features and thus find a peak compatible with a metallicity of -3.80<[C/H]<-3.50. We conclude that the TPCF method is not sensitive to the presence of the weakest systems in the real spectrum, despite the extremely high S/N and high resolution of the data. However, the results of this statistical technique would possibly change when combining more than one line of sight.

Figures (10)

• Figure 1: CIV column density, $\log N_{\rm{CIV}}$, versus the associated column density of neutral hydrogen, $\log N_{\rm{HI}}$. The vertical dotted lines mark the values of $\log N_{\rm{HI}}= 13.0, 13.5, 14.0$ and $14.8$. Blue data points represent the detected CIV systems identified by D16; white triangles indicate the upper limits on $\log N_{\rm{CIV}}$, which we used to derive the mock measurements (see Sect. \ref{['sect:test_uplim']}). The red triangles represent the upper limits in the range between $\log N_{\rm{HI}}= 14.0$ and $\log N_{\rm{HI}}= 14.8$, not included in our tests. The green dashed line indicates the relation $\log N_{\rm{CIV}} = \log N_{\rm{HI}} - 2.5$ which is used, in the range $13 < \log N_{\rm{HI}} < 14$, to create the mock systems for one of our tests, see Sect. \ref{['sect:test_uplim']}.
• Figure 2: TPCF computed on the complete spectrum, in black. The dashed vertical lines are marking the velocity separation of the most common doublets: CIV ($\sim$ 500 km s$^{-1}$), that is the one we are interested in, but also Mg$\,\rm II$ ($\sim$ 770 km s$^{-1}$) and Si$\,\rm IV$ ($\sim$ 1933 km s$^{-1}$). The dark-gray and light-gray shaded regions indicate the uncertainties of $1\, \sigma$ and $3\, \sigma$ respectively.
• Figure 3: TPCF for the spectrum where all metal lines have been deabsorbed except for CIV systems associated with $\log N_{\rm{HI}} < 14.8$ (left) and $\log N_{\rm{HI}}< 14.0$ (right) Ly$\alpha$ lines, plotted together with the $1\,\sigma$ and $3\,\sigma$ shaded regions obtained from the corresponding set of mock spectra.
• Figure 4: Left: TPCF computed on the spectrum in which we included the mock measurements derived from the upper limits in addition to the CIV systems associated with $\log N_{\rm{HI}}< 14.0$. Right: TPCF for the case in which we set the column density value of the mock CIV measurements such that $\log N_{\rm{CIV}} = \log N_{\rm{HI}} - 2.5$ in the range $13 < \log N_{\rm{HI}} < 14$. The shaded coloured regions indicate the 1 $\sigma$ and 3 $\sigma$ regions obtained from the distributions of the TPCF values computed on the corresponding sets of mock spectra.
• Figure 5: Left: TPCF computed on the synthetic spectrum (with same S/N as the original one) in which we included the CIV systems associated with $\log N_{\rm{HI}}< 14.0$. Right: TPCF for the case in which we also added the 135 mock measurements derived from Eq. \ref{['eq:logNHI_logNCIV_relation']} in addition to the weak systems. The shaded coloured regions indicate the 1 $\sigma$ and 3 $\sigma$ uncertainty regions obtained from the distributions of TPCF values computed on the corresponding sets of mock spectra.