The large-scale cross-correlation of Damped Lyman Alpha Systems with the Lyman Alpha Forest: First Measurements from BOSS

TL;DR

This work reports the first measurement of the cross-correlation between the Ly$\\\a$ forest and DLAs using BOSS DR9, detecting the signal on scales up to $40 \, h^{-1} \, { m Mpc}$ and confirming a gravitational origin through redshift-space distortions. By combining the cross-correlation with independently measured Ly$\\\a$ forest bias, the authors infer the DLA bias $b_D=(2.17\pm0.20)\beta_F^{0.22}$, implying DLAs predominantly live in massive halos around $M_h\sim 6\times10^{11} M_\odot$ with cross sections $\Sigma(M)\propto M^{\alpha}$ where $\alpha\approx1.1$ for $\beta_F=1$. The results require DLAs to be extended atomic gas in halos of ~$10^{12} M_\odot$, with typical DLA sizes $>20$ kpc and containing ~10% of halo baryons, posing new constraints on high-redshift galaxy formation and feedback models. The study demonstrates the power of Ly$\\\a$–DLA cross-correlation as a probe of halo mass, gas distribution, and the baryon census in the early universe, and motivates further refinements with future data to map redshift evolution and metallicity dependence.

Abstract

We present the first measurement of the large-scale cross-correlation of Lyman alpha forest absorption and Damped Lyman alpha systems (DLA), using the 9th Data Release of the Baryon Oscillation Spectroscopic Survey (BOSS). The cross-correlation is clearly detected on scales up to 40 Mpc/h and is well fitted by the linear theory prediction of the standard Cold Dark Matter model of structure formation with the expected redshift distortions, confirming its origin in the gravitational evolution of structure. The amplitude of the DLA-Lyman alpha cross-correlation depends on only one free parameter, the bias factor of the DLA systems, once the Lyman alpha forest bias factors are known from independent Lyman alpha forest correlation measurements. We measure the DLA bias factor to be b_D = (2.17 +/- 0.20) beta_F^{0.22}, where the Lyman alpha forest redshift distortion parameter beta_F is expected to be above unity. This bias factor implies a typical host halo mass for DLAs that is much larger than expected in present DLA models, and is reproduced if the DLA cross section scales with halo mass as M_h^alpha, with alpha= 1.1 +/- 0.1 for beta_F=1. Matching the observed DLA bias factor and rate of incidence requires that atomic gas remains extended in massive halos over larger areas than predicted in present simulations of galaxy formation, with typical DLA proper sizes larger than 20 kpc in host halos of masses ~ 10^12 solar masses. We infer that typical galaxies at z ~ 2 to 3 are surrounded by systems of atomic clouds that are much more extended than the luminous parts of galaxies and contain ~ 10% of the baryons in the host halo.

Figures (17)

• Figure 1: Quasar rest-frame wavelength distribution of the DLAs that satisfy the criteria of redshift and column density range, continuum to noise ratio and velocity separation from the background quasar. The red histogram (8189 systems) contains also quasars with BAL systems with a Balnicity index $BI < 1000 \, {\rm km\, s}^{-1}$, while the blue histogram (7458 systems) contains no systems flagged as BAL. Vertical lines show the cut applied to our final sample, $1005 {\, \rm\AA} \le \lambda_r \le 1037 {\, \rm\AA}$, reducing to 6780 the final number of DLAs used in this study.
• Figure 2: Left panel: distribution of the 6780 DLA redshifts. Right panel: neutral hydrogen column density distribution. The vertical line on the right panel corresponds to the standard lower limit for DLAs, $\log N_{HI} > 20.3$.
• Figure 3: Measured mean transmission as a function of redshift, with (solid black line) and without (dashed red line) the Mean Transmission Correction, compared to the measurement of 2008ApJ...681..831F (blue dotted line).
• Figure 4: Measured variance per pixel with MTC (solid black line) and without (long-dashed red line). The inferred intrinsic variance using the MTC assuming the corrected noise variance provided by the pipeline is shown by the short-dashed green line, and the intrinsic variance assumed for the weights [eq. (\ref{['eq:intrinsic']})] is the dotted blue line.
• Figure 5: The 12 sub-samples in which we split the DR9 quasar sample, $9$ in the northern Galactic hemisphere (left) and $3$ in the southern one (right).