4.6. Details on the Ly Emission Line in Very Distant Galaxies
The classic proposal by Partridge & Peebles (1967) that the Ly emission line might carry a fair fraction of the escaping bolometric luminosity of a young-star-rich galaxy is now testable. The review by Pritchet (1994) is also strongly recommended. Of course these early predictions did not reflect the possible presence of dust. Since the 1990s various searches have been initiated for Ly-emitting galaxies at large redshifts. Initially all of these searches led to negative results (eg, Thompson & Djorgovski 1995).
However, deeper photometric and spectroscopic searches of the last 6-7 years have yielded a modest number of "safe" Ly emitters - often (at the largest z's) the line being the only measurable spectral feature. The peak flux from a distant Ly emission line galaxy can often exceed the (redward) continuum level by a factor greater than 10! Of course the line from a faint system still has to compete with the strong telluric sky emission bands of OH and O2. Space-spectra won't deal with such a bright near-IR sky, and that will be advantageous.
Successful Ly searches include Cowie et al. (1998); Hu et al. (1998), Pascarelle et al. (1998), Hu et al. (1999), Steidel et al. (2000), Kudritski et al. (2000), Fynbo, Möller, and Thomsen, (2001).
There are three modes of Ly detection used with success in the past few years. They are narrow-band photometric excesses at fixed wavelengths (redshifts), a Ly forest (Lyman breaks in the continua) plus emission at the line, and serendipitous or fortuitous detections on multi-slit spectrograms. The issues we may face for each/all of the sub-types include the emission line strength and shape, the luminosity function of Ly emitters (and their surface densities), the effect of widespread neutral gas and dust, and the termination of the "dark ages" before or during the re-ionization epoch. Many of these topics have been addressed recently by Stern & Spinrad (1999); Rhoads et al. (2003); Ellis et al. (2001); Hu et al. (1999); Hu et al. (2002a), and in a predictive manner by Stiavelli (2002).
I suggest a few specific points where new observations and interpretations may be of substantial interest. For example, we'd like to confirm or deny that strong emission line Ly galaxies (z 4) obey the same luminosity function distribution as do photometrically selected Lyman break systems at z = 3 and z = 4 (cf. Steidel et al. 1999; Giavalisco 2002).
The difficulty in a present-sample comparison between Lyman break galaxies and Ly emitters is that (at high luminosities, at least) only a modest fraction of Lyman break (continuum selected) galaxies have strong Ly emission lines (W0 > 20 Å, say). Among the Ly-emitting systems (narrow-band or serendipitous detections) many candidates have very faint continua and would be missed in normal broad-band photometry. This latter bias is stressed by Fynbo et al. (2001). Indeed, Rhoads et al. (2003) found that if they summarized the line/continuum ratio in Ly galaxies, the equivalent widths occasionally "rose" to W0 1000 Å but more frequently to 190 Å. 60% of the Ly emitters studied by Malhotra & Rhoads (2002) had observed equivalent widths (hereafter EW) > 240 Å. For Ly-break systems, Shapley et al. (2001) find their 60th-percentile line to be a marginally-detectable 20 Å EW. The Shapley galaxies are at a slightly lower redshift; that difference is not critical.
If the above trend of lower-continuum-luminosity galaxies (z > 4) having stronger Ly-emission were to continue, we might diagnose this systematic as a trend toward lower metallicities for lower masses. But there are other possibilities; the Ly-emission line may as easily depend upon physical outflows (galactic winds), which in turn could have some total mass-dependence (or merger timing).
To get some idea as to the evolution of the luminosity function of young galaxies, we can compare the surface densities of distant galaxies. Pritchet (1994) made a first approximation to this. We utilize the Steidel et al. (1999) luminosity function zero point, and the "predictions" by Lanzetta et al. (1999) and Stern & Spinrad (1999) for a constant (with z) luminosity function. The cumulative surface density of identified z 4.5 galaxies in the HDF(N) is about 1.5 / '. These galaxies constitute a sample of continuum galaxies (photo-zs) and emission line galaxies with I814 26.5. This is very close to the "prediction" of the Lanzetta (unevolved) surface density (also see Ouchi et al. 2002).
The Lanzetta (1999) surface density curves do suggest a drop in the faint galaxy surface densities for the extreme case, z 6; that is not surprising at about I814 = 26. Still at slightly fainter magnitude levels a measure of the z 6.0 density by broad-band/narrow-band photometry may be a viable check on the luminosity function zero point and its shape (Lehnert & Bremer 2003).
What is the best physical interpretation of the very large EWs of Ly often measured for galaxies at z > 3?
The Ly-emitting galaxies with line EW in excess of 200 Å (rest-frame) (Malhotra & Rhoads 2002) are difficult to explain with a conventional O-B star mass function and ionizing spectra that are similar to those anticipated in extant solar-abundance models. The models rarely (and temporally) exhibit W0 150 Å (e.g., Charlot & Fall 1993). To decrease the observed Ly EW would be easy; as the dominant resonance line it is scattered frequently, and the resulting "random spatial walk" at the center of this line, coupled by small amounts of dust, can easily and drastically reduce the emission measure. It would, of course, also depend on the geometry.
To obtain a higher EW and/or higher flux in Ly, one can call upon three scenarios:
(a) A "tilted" mass function, with more O stars than found in local HII regions, as an ad hoc premise.
(b) We can also reduce the heavy element abundances in our models, and this allows an increase in the number of ionizing photons per O star. A recent paper by Schaerer (2002) considers the temporal evolution of the Ly line from model stellar populations ranging down from solar metal-abundances to very low metallicities (below the abundance level of the most metal-poor stars and gas in relatively nearby star-forming systems). We amplify this discussion below.
(c) Finally, sometimes a strong Ly emission line is the signature of an AGN. However, "real" AGN spectra, from QSOs down to modest-luminosity accretions, usually produce a broader Ly emission line (v 1000 km s-1) than seen in normal galaxies (v ~ 500 km s-1). They usually, but not always, also show C IV (moderately broad) 1549 Å. So most of the narrow-line Ly galaxies must have a line powered by the UV flux from OB stars. This is confirmed by the lack of hard X-ray flux in LALA galaxies at z 4.5 (Malhotra et al. 2003), indicating they are not obscured AGN.
The previously-mentioned Schaerer paper (Schaerer 2003) predicts EW of ~ 240-350 Å for metallicities down to Z = 4 × 10-4 (down from solar by a factor of ~ 50 times). Stiavelli (2002) shows even larger EW for Ly in metal-poor OB stars. Conceivably the initial stellar mass function (IMF) could also vary and be itself slanted toward higher masses because of the lower abundances. So the pairing of low abundance and a structure favoring massive O stars might allow EW to match most of the Ly galaxies selected by Malhotra & Rhoads (2002) and by Rhoads et al. (2003). An almost-practical spectroscopic test of this idea can be made by examining the UV HeII transition at 01640 Å. This line is much weaker than Ly in star-forming populations - with EW ~ 5 Å anticipated at low abundances of the metals. At higher abundances (near solar) it will be even weaker. Thus higher S/N spectrograms will be required in practice to use this He II feature in Ly "test galaxies".
The shape of the Ly emission line in distant star-forming galaxies is peculiar and may turn out to be an interesting guide to the circumgalactic medium as well as to galaxian winds or sporadic outflows.
The asymmetry of the Ly line has been noted by Kunth et al. (1998) and Pettini et al. (2001); it is also mentioned by Stern & Spinrad (1999). We have utilized the broad red wing of the Ly line and its sharp ISM/IGM cutoff on the blue side as a secondary criterion for assuming a single strong emission line is to be identified as Ly. This is opposed to the profile of the [O II] ]3727 doublet - unresolved in most lower-spectral-purity observations of faint objects. Recent work by E. Landes, S. Dawson, and the author has compared a spectal asymmetry index (a lambda-space ratio) for ten strong Ly emission lines; this particular index is small for a symmetric line and large for a red winged emission. Out of a sample of seven medium-resolution spectra of galaxies with a "solid" [O II] identification (z = 1.0) the asymmetry index averages 0.9 ± 0.1, while the 10 bonafide Ly galaxies, with <z> 4 display a larger range of index, from 1.0 to 2.3, with none less than unity. Seven of the Ly systems are clearly asymmetric with a noticeable red wing (see Fig. 6).
The Ly line is usually steeply declining on its blue side; we'll soon come back to this observation. So deciding whether an emission line is [O II] at a modest z or Ly at a large z, can often be helped by measuring the asymmetry. Of course a Ly (bigger redshift) decision based upon a large line asymmetry index becomes a sufficient, but not necessary condition for claiming the Ly identification.
The astrophysics behind the red wing of Ly has been well expounded by Tenorio-Tagle et al. (1999), Ahn, Lee & Lee (2002), and Dawson et al. (2002). The scenario here is a mini-galaxy scale outflow of neutral and partly ionized matter; the blueward velocity component being absorbed by external and expanding neutral H gas between us and the outflow. The backscattered component can be sufficiently redshifted off of the receeding wind, and hence avoid immediate absorption. This will impose a broadened red wing to the Ly line.
On the blue side of Ly we have a rapid decrease in intensity, a very sharp cutoff to the galaxy emission line at a slightly smaller redshift. The actual galaxy systemic velocity is likely to be near but blueward of the line peak, rather than its bisector at about half of maximum intensity.
In any case the Ly H absorption can take place in neutral circumgalactic gas, and in putative cluster gas, and also, at slightly lower redshift, neutral H clouds in the IGM - the well-studied Ly forest.
One interesting semi-quantitative aspect of the blue side cutoff is the difference we have noticed between the blue edge of Ly in QSO spectra and that of the normal distant galaxies, highlighted in this review (see Fig. 7). A new type of "proximity effect" seems in place, in the sense that the galaxy Ly profile on the short wavelength side is extremely steep, going from the line peak to near zero intensity in v1 = 100 km s-1, on our few echelle (higher spectral resolution) observations of the brightest distant systems (in their Ly line). The profile on the blue side of the strong emission line in QSO spectra (also z > 4) is moderately steep, but has a typical v2 800 km s-1, but often > 1000 km s-1.
Figure 7. The steepness of the ultraviolet side of the Ly emission line in a QSO (z = 5.09) and a faint galaxy, RD1 (z = 5.34, Dey et al. 1998). The very sharp and rapid decline of the blue side in the distant galaxy may be indicative of nearby (surrounding?) neutral gas. On the other hand, the QSO presumably ionizes much of any circumgalactic H originally present (with a small v) Thus the QSO line and continua are detectable to v2 2500 km s-1. Reductions and Figure by S. Dawson. |
Our interpretation of this systematic difference between UV-luminous QSOs and UV-fainter galaxies is straightforward. In proximity to the luminous ultraviolet radiation field of the QSOs H is very thoroughly ionized and thus doesn't absorb Ly photons at small v. On the other hand, a galaxy's UV ionizing radiation may not escape (or fully escape - see Dawson et al. 2002). Thus the rapid decline on the blue side of Ly may simply augur the existence of neutral gas in the circumgalactic environment near the galaxy. The effect may increase with redshift, but this is not yet well documented. This trend is potentially of interest in our present and future attempts to document the degree of IGM ionization near active objects and also on a diffuse, larger scale. Our coverage in redshift implies that we are looking back close to the re-ionization redshift, between z = 6 and z = 20, apparently.