Copyright © 2005 by Annual Reviews. All rights reserved
|
| Annu. Rev. Astron. Astrophys. 2005. 43:
769-826 Copyright © 2005 by Annual Reviews. All rights reserved |
There is evidence for winds in the spectra of several z > 1
galaxies. Low-ionization interstellar absorption lines that are
blueshifted by hundreds of km s-1 relative to systemic
velocities, and Ly
emission lines similarly shifted redward, have been detected in most
z ~ 3 - 4 Lyman break galaxies (LBGs; e.g.,
Lowenthal et al. 1997;
Pettini et al. 2000,
2001,
2002;
Adelberger et al. 2003;
Shapley et al. 2003),
in several gravitationally lensed
Ly-emitting
galaxies at z ~ 4 - 5 (e.g.,
Franx et al. 1997;
Frye, Broadhurst, &
Benitez 2002),
and in many luminous IR galaxies at z
2 (e.g.,
Smail et al. 2003;
Swinbank et al. 2005).
Ly emission with red
asymmetric or P Cygni-type profiles is also commonly seen in z
5
Ly-emitting galaxies
(e.g.,
Dey et al. 1998;
Ellis et al. 2001;
Dawson et al. 2002;
Ajiki et al. 2002).
A few isolated z > 1
line-emitting "blobs" and extended circumnuclear nebulae with
complex structured velocity fields have been tentatively interpreted
as spatially resolved large-scale winds driven by starbursts or AGNs
(e.g.,
Francis et al. 2001;
Ohyama, Taniguchi,
& Shioya 2004
and references therein).
Prevalent winds in high-z galaxies, particularly in LBGs, are
to be expected given their large surface densities of star formation:
the typical size of the star-forming cores in LBGs is
rhalf
~ 1.6 h70-1 kpc (e.g.,
Giavalisco et al. 1996)
and their SFRs are ~ 1 - 100
M
yr-1. Thus, they have star formation surface densities > 1
M
yr-1
kpc-2. However, one should be cautious when applying local wind
criteria to high-z galaxies because there are significant structural
differences in starbursts between the two epochs. The energy
injection zone in local starbursts is much smaller than the galaxy
half-light radius and generally centered on the nucleus, whereas the
starbursts in LBGs appear to be galaxy wide. The prominent disks of
local starbursts were probably not yet assembled in z ~ 3 LBGs,
perhaps resulting in more spherical winds than in local starbursts.
A well-studied wind in the gravitationally lensed LBG
MS 1512-cB58
(Pettini et al. 1998,
2000,
2002)
has bulk outflow velocity ~ 255
km s-1 on the basis of the positions of the low-ionization
absorption lines relative to the rest-frame optical emission lines.
Pettini et al. (2002)
note that its outflow kinematics are remarkably
symmetric: The profiles of the absorption lines from the different
ionization stages are broadly similar, spanning a range of velocities
of ~ 1000 km s-1, whereas the receding portion of the outflow
as mapped by the redshifted backscattered
Ly emission also has
essentially the same kinematics. The derived mass outflow rate
(~ 70 M
yr-1) exceeds the SFR of this galaxy (SFR
40
M
yr-1), so
this outflow may have had a strong impact on the chemical evolution of
the host galaxy. The entrained wind material appears to have enough
energy to escape the gravitational potential of
MS 1512-cB58
(Mbaryons ~ 1010
M), but the
large / Fe
ratio in the ISM of this object suggests that at least some of the
material made by previous stellar generations is retained.
The properties of the outflow in MS 1512-cB58 seem to be typical of
those in LBGs, although the data on fainter objects are necessarily
more uncertain
(Pettini et al. 2001;
Adelberger et al. 2003;
Shapley et al. 2003).
Large velocity shifts averaging ~ 650 km s-1 are found by
Shapley et al. (2003)
between the low-ionization lines
and the Ly emission
lines of LBGs. A comparison with the study of
Pettini et al. (2001)
suggests that these velocity offsets reduce to
~ 300 km s-1 when measured relative to the more reliable
nebular lines. This value is slightly higher than the outflow
velocities found in low-z galaxies of similar SFR's
(Section 4.4).
The equivalent widths of blueshifted low-ionization interstellar
absorption lines in LBGs anti-correlate strongly with the strength of
the Ly emission, and
correlate with the color excess
E(B - V). The latter suggests that dust is
entrained with the
outflowing neutral gas and is sufficiently widespread to redden the
host LBG; this effect is also observed in local wind galaxies
(Section 4.10). The former is easy to
understand if we recall that the
equivalent widths of saturated absorption lines (as is the case here)
depend only on the covering fraction of the absorbing material and the
range of velocities over which this material is absorbing. Larger
covering factor or absorbing velocity interval implies that less
Ly emission can escape
from the wind. This effect may also
explain the redder UV continuum and the larger kinematic offsets
between Ly and the
interstellar absorption lines observed in
weak Ly-emitting LBGs
(Shapley et al. 2003).
Quantifying the environmental impact of LBG winds comes by probing the environment with spectra of background QSOs. Around six z ~ 3 LBGs, Adelberger et al. (2003) find hints of an H I deficit within comoving radius 0.5 h70-1 Mpc. They favor a scenario whereby LBG winds influence the nearby IGM directly over a proximity effect caused by LBG-ionizing radiation. The excess of absorption-line systems with large C IV columns that they find near LBGs is interpreted as further evidence for chemical enrichment of the IGM by LBG winds, although it could also be attributed to debris from tidal interactions (e.g., Morris & van den Berg 1994).
In a recent extension to this survey,
Adelberger et al. (2005)
find that the relationship between C IV column density and clustering
around LBGs is still present, although at a lower level than in the
original survey. Most galaxies now appear to have significant H I
within 1 h-1 Mpc of their centers, but some fraction
of them (~ 1/3) still show a significant deficit. The LBGs with strong
H I absorption have roughly as much absorption as expected from
windless, smoothed particle hydrodynamics (SPH) simulations of
CDM universes (e.g.,
Croft et al. 2002;
Kollmeier et al. 2003),
but LBGs without much H I are not present in these simulations.
These results are at odds with the original suggestion of large,
spherical windblown cavities around LBGs
(Adelberger et al. 2003).
They are qualitatively more consistent with the idea that winds emerge
along paths of least resistance, possibly avoiding large-scale
filaments (e.g.,
Theuns et al. 2002).
6.2. QSO Absorption-Line Systems
Large-scale GWs at all redshifts can be detected by their absorption
of background QSO continuum light (e.g.,
Rauch 1998).
Comparisons between the number densities of Mg II absorbers and
star-forming galaxies and the properties of local outflows (e.g.,
Bond et al. 2001)
suggest that there are enough Mg II absorbers to account
for the expected properties of winds at 1 < z < 2. But a more
detailed analysis of absorption-line spectra is needed to answer this
question quantitatively. Possible absorption signatures of winds or
superbubbles include broad (few hundred km s-1) and complex
profiles pointing to strong nonrotational kinematics, a pairwise
absorption pattern that straddles weak absorption near the kinematic
center of the line, -rich
abundances, and large cloud-to-cloud
variations in metallicity and ionization level. Such signals are
evident in several very strong Mg II absorbers
(Bond et al. 2001)
and at lower column densities (e.g.,
Rauch et al. 2002;
Zonak et al. 2004).
Detailed comparison of the kinematics of absorbers with those of the Mg II-absorbing galaxies can further test for winds. In a study of five Mg II-absorbing galaxies at redshifts 0.44 < z < 0.66 and of absorbers with projected impact parameters from those galaxies of 15 - 75 h-1 kpc, Steidel et al. (2002) find that halo rotation sometimes dominates radial infall or outflow even for gas far from the galactic plane. But in the z = 0.7450 Mg II galaxy toward quasar Q1331+17, no absorbing gas is detected at the projected velocity of the disk rotation (Ellison, Mallén-Ornelas, & Sawicki 2003). The motion of the absorbing gas in this galaxy is consistent with a large (~ 30 h70-1 kpc) superbubble expanding at ~ ± 75 km s-1. Clearly, more data of high quality are needed for a final verdict (see, e.g., Côté et al. 2005).