Gemini Near-infrared Spectroscopy of Luminous z~6 Quasars: Chemical Abundances, Black Hole Masses, and MgII Absorption

Abstract

We present Gemini near-infrared spectroscopic observations of six luminous quasars at z=5.8$\sim$6.3. Five of them were observed using Gemini-South/GNIRS, which provides a simultaneous wavelength coverage of 0.9--2.5 $\mu$m in cross dispersion mode. The other source was observed in K band with Gemini-North/NIRI. We calculate line strengths for all detected emission lines and use their ratios to estimate gas metallicity in the broad-line regions of the quasars. The metallicity is found to be supersolar with a typical value of $\sim$4 Z_{\sun}, and a comparison with low-redshift observations shows no strong evolution in metallicity up to z$\sim$6. The FeII/MgII ratio of the quasars is 4.9+/-1.4, consistent with low-redshift measurements. We estimate central BH masses of 10^9 to 10^{10} M_{\sun} and Eddington luminosity ratios of order unity. We identify two MgII $\lambda\lambda$2796,2803 absorbers with rest equivalent width W_0^{\lambda2796}>1 \AA at 2.2<z<3 and three MgII absorbers with W_0^{\lambda2796}>1.5 \AA at z>3 in the spectra, with the two most distant absorbers at z=4.8668 and 4.8823, respectively. The redshift number densities (dN/dz) of MgII absorbers with W_0^{\lambda2796}>1.5 \AA are consistent with no cosmic evolution up to z>4. Comment: 33 pages (including 7 figures and 6 tables), AJ in press

Full text

Page 1

arXiv:0707.1663v1 [astro-ph] 11 Jul 2007
GEMINI NEAR-INFRARED SPECTROSCOPY OF LUMINOUS z ∼ 6
QUASARS: CHEMICAL ABUNDANCES, BLACK HOLE MASSES, AND
Mgii ABSORPTION1
Linhua Jiang1, Xiaohui Fan1, Marianne Vestergaard1, Jaron D. Kurk2, Fabian Walter2, Brandon
C. Kelly1, and Michael A. Strauss3
ABSTRACT
We present Gemini near-infrared spectroscopic observations of six luminous quasars
at z = 5.8 ∼ 6.3. Five of them were observed using Gemini-South/GNIRS, which
provides a simultaneous wavelength coverage of 0.9–2.5 µm in cross dispersion mode.
The other source was observed in K band with Gemini-North/NIRI. We calculate line
strengths for all detected emission lines and use their ratios to estimate gas metallicity
in the broad-line regions of the quasars. The metallicity is found to be supersolar with
a typical value of ∼ 4 Z⊙, and a comparison with low-redshift observations shows no
strong evolution in metallicity up to z ∼ 6. The Feii/Mgii ratio of the quasars is
4.9 ± 1.4, consistent with low-redshift measurements. We estimate central BH masses
of 109to 1010M⊙ and Eddington luminosity ratios of order unity. We identify two
Mgii λλ2796,2803 absorbers with rest equivalent width Wλ2796
and three Mgii absorbers with Wλ2796
0
> 1.5˚ A at z > 3 in the spectra, with the two
most distant absorbers at z = 4.8668 and 4.8823, respectively. The redshift number
densities (dN/dz) of Mgii absorbers with Wλ2796
evolution up to z > 4.
0
> 1˚ A at 2.2 < z < 3
0
> 1.5˚ A are consistent with no cosmic
Subject headings: galaxies: high-redshift — galaxies: active — quasars: emission lines
— quasars: absorption lines — quasars: general
1Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721
2Max-Planck-Institut f¨ ur Astronomie, K¨ onigstuhl 17, D-69117 Heidelberg, Germany
3Department of Astrophysical Sciences, Princeton University, Princeton, NJ 08544
1Based on observations obtained at the Gemini Observatory (acquired through the Gemini Science Archive), which
is operated by the Association of Universities for Research in Astronomy, Inc., under a cooperative agreement with
the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), the Particle Physics
and Astronomy Research Council (United Kingdom), the National Research Council (Canada), CONICYT (Chile),
the Australian Research Council (Australia), CNPq (Brazil) and CONICET (Argentina).

Page 2

– 2 –
1.INTRODUCTION
In recent years we have witnessed the discoveries of quasars at z ∼ 6; at this epoch the universe
was less than one Gyr old (e.g. Fan et al. 2003, 2006; Cool et al. 2006; McGreer et al. 2006). These
high-redshift quasars are among the most luminous objects known and provide direct probes of
the early universe when the first generations of galaxies and quasars formed. Quasar activity
and the formation processes of galaxies and supermassive black holes (BHs) are closely correlated
(e.g. Kauffmann & Haehnelt 2000; Wyithe & Loeb 2003; Hopkins et al. 2005; Croton et al. 2006),
so these z ∼ 6 objects are essential for studying the accretion history of BHs, galaxy formation,
and chemical evolution at very early epochs. High-redshift luminous quasars harbor BHs with
masses of 109–1010M⊙and emit near the Eddington limit (e.g. Barth et al. 2003; Vestergaard 2004;
Jiang et al. 2006; Kurk et al. 2007), revealing the rapid growth of central BHs. Their emission line
ratios show solar or supersolar metallicity, as is found in low-redshift quasars (e.g. Barth et al.
2003; Dietrich et al. 2003a; Freudling et al. 2003; Maiolino et al. 2003), indicating that there was
vigorous star formation and element enrichment in the first Gigayear.
Understanding quasars requires observations from X-ray to radio wavelengths; each spectral
region is due to different physical mechanisms and probes different regions of the active nucleus. In
particular, the rest-frame UV spectrum, from the Lyα emission line to the Feii bump at 2000–3000
˚ A, contains strong diagnostic emission lines and provides key information on the physical conditions
and emission mechanisms of the broad-line region (BLR). The bulk motions of the BLR can be used
to determine the mass of the central BH, while chemical abundances in the BLR are important in
understanding the history of star formation in the host galaxy. Photoionization models show that
various emission-line ratios provide reliable estimates of metallicity in the BLRs of quasars (e.g.
Hamann et al. 2002; Nagao et al. 2006). Strong permitted line ratios, such as Nv λ1240/Civ λ1549
(hereafter Nv/Civ), Nv λ1240/Heii λ1640 (hereafter Nv/Heii), and Feii/Mgii λ2800 (hereafter
Feii/Mgii), are relatively easy to measure even in z ∼ 6 quasars (e.g. Hamann & Ferland 1993,
1999). Fe abundance is of particular interest (e.g. Hamann & Ferland 1993, 1999): most of the
Fe in the solar neighborhood is generated by Type Ia supernovae (SNe Ia), whose precursors are
believed to be long-lived intermediate-mass stars in close binaries, so appreciable Fe enrichment
can only happen on a timescale of one Gyr after the initial starburst (e.g. Greggio & Renzini 1983).
On the contrary, the production of α elements such as Mg and O is dominated by SNe of types II,
Ib and Ic, which explode very soon after the initial burst. Therefore, the Fe/α ratio is expected to
be a strong function of age in young systems, and thus it puts a useful constraint on the age and
chemical enrichment of the gas in the quasar environment.
At low redshift, elemental abundances measured from strong emission and intrinsic absorption
lines have revealed that quasar environments have solar or supersolar metallicity.
measurement of these spectral lines is difficult, as the rest-frame UV waveband is redshifted to the
NIR range. NIR spectroscopy has been obtained for a few z ∼ 6 quasars (e.g. Goodrich et al. 2001;
Pentericci et al. 2002; Barth et al. 2003; Freudling et al. 2003; Maiolino et al. 2003; Stern et al.
2003; Iwamuro et al. 2004). These observations only provide NIR spectra with low signal-to-noise
At z ∼ 6,

Page 3

– 3 –
ratios (SNRs) or partial wavelength coverage. Even basic properties such as the average continuum
slope are not well measured. Recently, Kurk et al. (2007) obtained NIR spectra with high SNRs
for a sample of five z ∼ 6 quasars with ISAAC on VLT. They used the ISAAC data to study the
Feii/Mgii ratios, BH masses, and the Str¨ omgren spheres around quasars. In this paper we present
Gemini NIR spectroscopy of six z ∼ 6 quasars. All these quasars were discovered by the Sloan
Digital Sky Survey (SDSS; York et al. 2000) and were selected from Fan et al. 2000, 2001, and
2004 (hereafter quasar discovery papers). They have extensive follow-up observations in X-ray (e.g.
Shemmer et al. 2006), Spitzer bands (e.g. Jiang et al. 2006), millimeter/submillimeter and radio
wavelengths (e.g. Wang et al. 2007), providing a fundamental dataset to study luminous quasars at
z ∼ 6 in the context of the growth of early BHs and their relation to galaxy formation. The Gemini
data have high SNRs and excellent NIR wavelength coverage. They are used to measure the metal
abundances in high-redshift environments and estimate central BH masses using empirical mass
scaling relations (e.g. McLure & Dunlop 2004; Vestergaard & Peterson 2006).
We also use these high-redshift spectra to study the evolution of strong intergalactic Mgii ab-
sorption at z < 6. Quasar absorption lines are an unbiased tracer of intervening gaseous material
along the lines of sight to quasars. Strong Mgii absorption systems are usually directly associ-
ated with galaxy halos or disks (e.g. Charlton & Churchill 1998; Churchill et al. 2000; Steidel et al.
2002). The statistical properties of the Mgii λλ2796,2803 doublet absorbers at low redshift
(z ≤ 2.2) have been well studied (e.g. Steidel & Sargent 1992; Nestor et al. 2005; Prochter et
al. 2006; see a summary of Mgii absorption surveys by Churchill et al. 2005). At higher redshift,
the Mgii doublet is shifted to the NIR wavelength range, where detectors are much less efficient
than are detectors at the optical region. The study of high-redshift Mgii absorbers is therefore
limited by the small number of quasars with good NIR spectra available. Our Gemini/GNIRS
spectra with high SNRs have excellent NIR wavelength coverage, so they are efficient to probe the
Mgii absorption systems at z < 6.
The structure of the paper is as follows. In § 2 we describe our Gemini observations and
data reduction of the six quasars. In § 3 we conduct a detailed spectral analysis, including the
measurement of emission-line strengths. Chemical abundances are calculated and central BH masses
are estimated in § 4. Mgii absorption systems in the quasar spectra are identified and analyzed
in § 5. We give a brief summary in § 6. In this paper we use λ (λ0) to denote observed-frame
(rest-frame) wavelength. We use a Λ-dominated flat cosmology with H0= 70 km s−1Mpc−1, Ωm
= 0.3 and ΩΛ= 0.7 (Spergel et al. 2007).
2. OBSERVATIONS AND DATA REDUCTION
We obtained NIR spectra for the six luminous z ∼ 6 quasars using Gemini-South/GNIRS and
Gemini-North/NIRI. Five of them were chosen to be easily reached by Gemini-South, and were
observed using GNIRS in February and March 2006. The observations on GNIRS were carried out
in cross-dispersion mode, giving a simultaneous wavelength coverage of 0.9–2.5 µm with excellent

Page 4

– 4 –
transmission. At z ∼ 6, strong diagnostic emission lines such as Civ and Mgii are redshifted
to this wavelength range, so GNIRS provides us an efficient way to collect NIR information for
high-redshift quasars. We used the short camera on GNIRS with a pixel scale of 0.15′′/pixel. The
slit length in cross-dispersion mode is 6.1′′and we used a slit width of 0.675′′to optimize the
combination of resolution and light gain. The resolving power is about 800 and the dispersion
varies from ∼3˚ A per pixel at 0.9 µm to ∼6˚ A per pixel at 2.5 µm. This is sufficient to separate
the Mgii λλ2796,2803 doublet absorption lines. We used the standard ABBA nodding sequence
between exposures. The exposure time at each nod position was five minutes and the distance
between the two positions was 3′′. Before or after the exposure of each quasar, a nearby A or F
spectroscopic standard star was observed for flux calibration and to remove telluric atmosphere
absorption. The log of observations is given in Table 1, where redshifts, zAB(AB magnitudes) and
J (Vega-based magnitudes) are taken from the quasar discovery papers.
The GNIRS data were reduced using the Gemini package within IRAF2. Briefly, for each
quasar the NIR spectroscopic data were sky-subtracted for each pair of images taken at the two
nod positions. Then the images were flat-fielded. In cross-dispersion mode, GNIRS uses orders 3–8
to cover 0.9–2.5 µm. A single flat has non-uniform illumination between orders, so three sets of
flats with different exposure times were taken each night. The final flat is extracted from the three
flats so that each order has good illumination but without saturation. After flat-fielding, distortion
correction and wavelength calibration were applied, the frames were combined, and one-dimensional
spectra were extracted. Then the spectra were flux calibrated and corrected for telluric atmosphere
absorption using the spectra of the spectroscopic standard stars. Finally the spectra in different
orders were scaled and connected to a single spectrum. Adjacent orders cover a short common
wavelength range, which enables the individual orders to be scaled to form a continuous, smoothly
connected spectrum. This arbitrarily scaled spectrum was then calibrated to the J broad-band
magnitude listed in Table 1 and thereby placed on an absolute flux scale with an uncertainty of
5∼10%. The typical SNR at 1.2 µm reaches ∼ 8 per pixel.
Figure 1 shows the spectra of the five quasars in the observed frame. The spectra at λ ≤ 1.0
µm are taken from the quasar discovery papers. The spectra have been smoothed by three pixels.
The bottom panel shows the sky transparency at an airmass of 1.0 and water vapor of 1.6 mm.
There are strong telluric absorption bands around 1.4 and 1.9 µm, so we do not show the spectra
in these ranges. In Figure 1 we also show a spectrum of a standard star HIP 54027 calibrated using
the same procedure as we do for our science objects. For comparison, a spectrum (in red) of a black
body with the effective temperature of HIP 54027 is given on top of the spectrum of HIP 54027.
The good agreement between the two spectra indicate that spectra in different dispersion orders
are properly scaled and connected.
2IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the Association of
Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation.

Page 5

– 5 –
We obtained a K-band spectrum for SDSS J1623+31123using Gemini-North/NIRI in August
and September 2004 and August 2005. The observations were made with a f/6 camera in spectro-
scopic mode. The long-slit has a length of 110′′and we chose to use a slit width of 0.75′′, which
provides a resolving power of ∼ 500. The observing strategy and data reduction are similar to those
used for the GNIRS data. The standard ABBA mode was used with a nodding distance of 20′′and
an exposure time of five minutes at each nod position. A nearby A or F spectroscopic standard star
was observed before or after the target observation. The NIRI data were also reduced using the
Gemini package within IRAF. After sky-subtraction, flat-fielding and wavelength calibration, the
frames were combined. Then one-dimensional spectra were extracted from the combined images,
and were flux calibrated using the spectra of standard stars. The observing log is given in Table 1
and the spectrum is shown in Figure 2. Since we do not have K-band photometry for this object,
we cannot apply an absolute flux calibration. In Figure 2, we have scaled the spectrum to its J
magnitude assuming a power-law continuum slope of αν= −0.5 (fν∼ ναν).
3. SPECTRAL FIT AND EMISSION-LINE MEASUREMENTS
To analyze broad emission lines such as Civ and Mgii we fit and subtract the UV Feii emission
that contaminates most of the UV-optical spectral region; thousands of Feii lines are emitted at
UV-optical energies that blend with other lines in the spectrum. We use the UV Feii template
presented by Vestergaard & Wilkes (2001) and adopt the fitting procedure described in detail in
§ 4.2 by these authors. Briefly, the fitting is done as an iteration over a power-law fit to the
continuum emission and a model fit to the Feii emission using a scaled and broadened version of
the Feii template. The Feii template is broadened by convolving the template with Gaussians
with a range of sigmas. In the first iteration a power-law is fitted to regions with very little or
no contribution from line emission (e.g. 1275–1295˚ A; 1325–1335˚ A; 1425–1480˚ A; 2180–2220˚ A;
3000–3040˚ A). Upon subtraction of this primary continuum fit multiple broadened copies of the
Feii template are scaled to regions which predominantly contain Fe emission. The broadened,
scaled Feii template that provides the best overall match to the observed Feii emission in the
observed spectrum is then selected and subtracted. Some adjustments of the Feii fitting regions
are sometimes necessary to optimize the fit, as judged by a minimization of the residual flux upon
subtraction of the Feii model. After subtraction of the best matching Feii model from the original
data (i.e., with the continuum not subtracted) another power-law fit to the continuum emission is
performed; the lack of (most of the) contaminating Feii enables the fitting to be done over a broader
wavelength range, often resulting in a better and more realistic continuum fit. After subtraction
of this continuum model, the Feii fitting is repeated. The iteration of the continuum and Feii
fitting is repeated until these individual fits do not change and the subtraction of the Feii model
3The naming convention for SDSS sources is SDSS JHHMMSS.SS±DDMMSS.S, and the positions are expressed
in J2000.0 coordinates as given in Table 1. We use SDSS JHHMM±DDMM for brevity.