To reconstruct the history of galaxy formation, we need to have some way of connecting present-day galaxies with their progenitors to high redshift. This could be done most directly by measuring galaxy masses. Two possible approaches are: (1) the use of galaxy-galaxy lensing, where lensed images of background galaxies are used to probe the masses of foreground galaxies, and (2) spectroscopic measurements of internal galaxy dynamics (see below). We will also need to use moderate resolution spectroscopy (R = 1000-5000) to measure redshifts and the detailed properties of large numbers of galaxies (> 50,000/sq. deg.) to faint magnitudes (R ~ 26.5). Large galaxy samples are needed because the history of galaxy formation involves multiple evolutionary threads (mass assembly and morphological evolution, star formation and chemical enrichment history, and possibly AGN (active galactic nuclei) activity), which depend on multiple parameters (e.g., mass and environment as a minimum) as functions of cosmic time. Thus, for example, the input data required for the reconstruction must sample, at a minimum, a range of star formation rate (5 bins), morphology (4 bins), and chemical abundance (4 bins) as functions of galaxy mass (4 bins), redshift (5 bins), and environment (3 bins). With a minimum of 100 galaxies per bin, a minimum sample of 500,000 objects is required. The faint magnitude limit is needed in order to probe, at high redshift, the equivalent of L* densities in the present-day universe. By probing comparable number densities at each intervening epoch, we will able to make robust evolutionary connections between present-day galaxies and their progenitors.
In this section, we give some benchmarks for our current limited ability to study detailed galaxy properties (star formation rates, metallicities, stellar ages, and internal galaxy dynamics) at z > 1, and describe several ways in which faint spectroscopy will be critical for the further advancement of our understanding of galaxy formation. In particular, we will be able to extend current studies on the galaxy luminosity function to encompass a statistically significant sample of objects. We will also be able to extend current studies to include very faint, new galaxy populations at z > 6. In addition, we will be able to study familiar galaxies at z > 0 at high spatial resolution.
Examples of the detailed galaxy properties that can be studied using seeing-limited, primarily optical spectroscopy include the following:
Star Formation Rates: Recent work has led to a provocative first look at the star formation history of the universe (e.g., Steidel et al.; see Figure 1).1 The results imply a relatively constant star formation rate over the first half of the history of the universe, followed by a gradual tapering off until the present era. There are significant limitations to our current understanding (e.g., the statistics are poor at all redshifts) and many significant issues remain (e.g., when and under what conditions did the first stars form?). Moreover, because we currently use global measures of star formation integrated over all (primarily bright) galaxies at each redshift, our current understanding can say little about the dependence of the star formation history on galaxy properties such as mass and morphological type. As a result, our current understanding says little about the evolution of individual systems such as the Milky Way. Thus, one role of the GSMT is to chart out the detailed star formation history of galaxies as a function of mass, environment, and morphological type in order to place the Milky Way in context.
Metallicity: High signal-to-noise spectra can be used to estimate the metallicity of the gas and stars in galaxies at z > 3. These measurements are typically beyond the capability of existing 8-10m telescopes, because the galaxies at z ~ 3 are typically too faint to allow the measurement of spectra with signal-to-noise ratios exceeding 2-3 (L* corresponds to R ~ 24.5 at z ~ 3). A demonstration of the kind of measurements that will be routinely possible with a GSMT is provided by LRIS/Keck observations of the unique gravitationally lensed galaxy MS 1512-cB58 (z = 2.73; AB6540 = 20.4, magnification factor ~ 30). From their high signal-to-noise spectrum of this system, Pettini et al.2 were able to estimate that both the stars and gas in the galaxy are metal enriched at a level 0.3-0.5 of solar, based on the strength of CIV (three times ionized carbon) P Cygni absorption and the absorption strengths of weak interstellar absorption lines (see Figure 2). With the sensitivity of the GSMT, we will be able to make similar measurements for a larger, statistically significant sample of galaxies. We will also be able to measure metallicities more directly by using weaker transitions that, unlike CIV, are not contaminated by ISM (Interstellar Medium) absorption.
Age Estimates: Stellar population studies can also be carried out from spectroscopy with signal-to-noise ratios adequate to measure spectral breaks and stellar absorption lines with rest-frame equivalent widths of a few Angstroms or less. Reasonably high signal-to-noise spectra can be used to age date the stellar populations of galaxies. For example, spectral breaks in the optical (rest-frame UV) spectrum of the weak radio source LBDS 53W091 imply an age 3.5 Gyr. The detection of a 3.5 Gyr old system at z = 1.5 implies a high formation redshift z > 5 for the system and places useful constraints on cosmological parameters3 (see Figure 3). To date, stellar ages have been measured only for a handful of galaxies, in part because such measurements test the limits of 8-10-m telescopes. Thus, one role of the GSMT is to enable similar measurements for a larger number of galaxies. If we can use the GSMT to measure stellar population ages for a sample of galaxies, e.g., evolving ellipticals, we will have an independent constraint on the evolutionary history of the galaxy population, (i.e., when the bulk of the stars formed and the chemical enrichment history of the population). This is similar to the situation for metallicity measurements.
Galaxian winds: Outflows from galaxies are also expected to have significant implications for the star formation and chemical enrichment history of galaxies and the chemical enrichment of the IGM (Intergalactic Medium). Recent developments suggest direct evidence for this process at work: rest-frame UV lines in z ~ 3 star forming galaxies and radio galaxies are observed to have broad blue-shifted absorption components (e.g., Q000-263 D6-; see Figure 4)4. One interpretation is that these features are produced by galaxian winds, possibly driven by supernovae and stellar winds, and/or an origin in an AGN. Are such powerful winds a product of the evolution of galaxies of all masses? Can metals ejected by winds significantly enrich the IGM? If powerful winds are also observed in galaxies at higher redshift, they may be able to explain the observed metal enrichment of the IGM at z ~ 3. If winds also punch holes in galactic disks, allowing the escape of ionizing photons, that may provide an explanation for the reionization of the universe at high redshift. With the GSMT, we will able to address these questions by measuring mass loss and momentum input rates of winds from galaxies over a range of galaxy mass and redshift.
Sensitivity Estimates: The same observational program for large scale structure described in Section 2.1.2 would provide much of the data needed to address the issues discussed in this section. For example, the resolution (R = 5000) and signal-to-noise (s/n = 20) that are needed to probe the structure in the IGM are comparable to those needed to study the metallicity of the gas and stars in individual galaxies (as in Figure 2). Such studies could be carried out easily for all Lyman break galaxies (~ 25 AB mag) using the same data set described in Section 2.2.2. Thus, detailed studies of galaxy metallicities, winds, and other galaxy properties could be obtained without additional observational effort for the 250,000 objects that will be studied at R = 5000 in the program described in Section 2.1.2. As another example, the resolution (R = 500-1000) and signal-to-noise (s/n = 5) needed to probe the stellar ages of galaxies (as in Figure 3) are comparable to those that will be obtained in the faint galaxy portion (27.5 AB mag) of the survey described in Section 2.1.2. Thus, stellar ages and other galaxy properties could be obtained without additional observational effort for the 2.5 million objects that will be studied at R = 1000 in the program described in Section 2.1.2.
A high throughput GSMT, with diffraction-limited capability and instruments that potentially employ OH suppression, has the potential to study very faint, new galaxy populations such as the first objects to illuminate the universe.
Evidence for High-z Populations: A variety of observational evidence suggests the existence of a significant population of luminous objects at very high redshift (z » 3). One line of evidence comes from the metallicity and ionization properties of the IGM. For example, the Ly forest at z » 3 is found to be metal enriched at a level ~ 0.01 of solar over a wide range of column densities, from the high column densities NHI = 1021 cm-2 of damped Ly systems, which are believed to be the (post-collapse) progenitors of present-day disk galaxies, to the much lower column densities NHI = 1015 - 1016 cm-2 that characterize the IGM overdensities from which galaxies eventually form.5 The similarity in the level of enrichment for such a wide range in column density is interpreted as evidence for an early phase of metal enrichment at high redshift. In addition, the absence of Gunn-Peterson troughs in high redshift quasar spectra indicates that the IGM was reionized at z 5. The increasing rarity of QSOs (quasi-stellar objects) at z > 3 may indicate the need for a distinct high redshift population of objects capable of producing the requisite ionization.
Additional evidence comes from the properties of high-z galaxies. The z ~ 3 star forming (Lyman break) galaxies are known to be enriched in metals (e.g., Steidel et al.)6, implying significant metal enrichment at z > 3. Moreover, the tight photometric sequences of galaxies in low- to intermediate-z clusters suggest that the elliptical galaxy population forms at high redshift. Finally, some z ~1.5 ellipticals are found to have old stellar populations (= 3.5 Gyr)3, which also suggests a formation redshift z > 5.
To this, we can also add the empirical evidence of the known existence of z > 5-6 QSOs (e.g., z = 6.28)7 and galaxies (e.g., z = 5.74)8. In particular, the curious fact that 2 of the 6 galaxies known at z > 5 were discovered serendipitously is highly suggestive of a significant population of galaxies at z > 5.
Expected Number Densities: Estimates by Haiman & Loeb9 indicate that it may be possible to directly observe the high-z population that is responsible for the reionization and metal enrichment of the IGM if an early generation of star clusters is the source of reionization (see Figure 5). Haiman & Loeb9 have also explored the possibility that high redshift QSOs contribute to the reionization of the IGM (Figure 6).
What Will They Look Like? Popular techniques for the detection of high-z objects attempt to detect the light from a young stellar population, either via the bright UV continuum, which bears the characteristic imprint of HI absorption by the IGM, or via strong emission lines such as Lya and Ha that are produced in HII regions. These techniques have led to the identification a small sample of z > 5 galaxies and QSOs. Observations of known z ~ 5 objects provide a useful guide to how we might expect to detect reionization sources and also to what we might be able to deduce about them. For example, in the case of the serendipitous R-band "dropout" 0140+326 RD1, the first spectroscopically confirmed galaxy at z > 5, Dey et al.11 were able to deduce from the rest-frame UV spectral energy distribution the redshift of the system (z = 5.34), and that the object is a star forming galaxy rather than an AGN and could provide constraints on the star formation rate and galaxy mass. As intriguing as these observations are, they also indicate the limitations of current 8-10-m telescopes for the study of the high-z universe. For example, the spectrum of 0140+326 RD1 shown in Figure 7 required an integration time of ~10 hrs with the Keck telescope. Observations of significantly higher signal-to-noise will be difficult to obtain with the current generation of 8-10-m telescopes.
Sensitivity Estimates: For the estimates given below, the total efficiency of the spectrograph and telescope system is assumed to be 0.5, with diffraction-limited performance and 50% Strehl. For the H band, the sky background is assumed to be either H = 14.8, the mean ground-based background including the OH lines, or H = 18.5, approximating the background level between the OH lines, as would be appropriate for an OH suppression spectrograph. The target is assumed to be a point source.
Objects at the 10 nJy flux level (29 AB mag) can be observed spectroscopically at R = 500 with s/n = 3 per resolution element at H in 8 hrs of integration time, assuming a sky background of H = 14.8. If OH suppression techniques are used so that the background is H = 18.5, s/n 10 in the same integration time. At this flux level, the number density of z > 10 star clusters and quasars is expected to be >100 and ~1 per sq. arcmin per logarithmic flux interval, respectively (cf. Figures 5 and 6). Thus, multiple targets would be accessible within a < 1' field.
Objects at the characteristic JWST imaging flux limit ~1 nJy (31.4 AB mag) can be observed spectroscopically at R = 500, with s/n = 3 per resolution element at H in ~8 nights of integration time, using OH suppression so that the background is H = 18.5. At this flux level, the number density of z > 10 star clusters and quasars is expected to be >1000 and >10 per sq. arcmin per logarithmic flux interval, respectively (cf. Figures 5 and 6). Again, multiple targets would be accessible within a < 1' field. Although JWST would be able to detect these objects, obtaining spectra for them would be prohibitively time-consuming, requiring ~100 times longer integration times than with the GSMT.
With its high sensitivity and angular resolution, the GSMT will open another new window on galaxy evolution: the study of the spatially resolved kinematic and metallicity structure, and the stellar content of distant galaxies (angular scales 0.1-0.01"). With spatially resolved, moderate resolution (R = 2000) spectroscopy, we would be able to address issues such as the evolution of galaxy metallicity and mass-to-light ratio in the context of star formation and morphological evolution. In addition, galaxy masses measured using galaxy kinematics would complement masses measured through lensing. The high angular resolution capability of a GSMT in the near infrared (1--2.5µm) would enable the following kinds of measurements.
Star Formation Rates: In the redshift range z 1-3, traditional optical emission line diagnostics such as H and H will fall in the 1-2.5µm region, in which high angular resolution capability will be possible with a GSMT. The use of these traditional diagnostics to measure star formation rates offers several advantages. Because studies of star formation rates in galaxies at low redshift are often based on H, using the same diagnostics at higher redshift does not introduce an additional technique-dependent systematic bias. In addition, extinction-corrected star formation rates can be obtained from the measured Balmer decrement.
Dynamical Masses: In the same redshift range, strong emission lines such as H and [OIII]5007 can also be used to measure dynamical masses for star forming galaxies. As discussed below, the measurement of galaxy masses over a range of redshifts is one of the most critical (and least well-known) measurements that is needed to unravel the history of galaxy formation. Masses measured with these HII region diagnostics should be more reliable than those measured with optical (rest-frame UV) interstellar absorption lines, because the dynamics reflected in the rest-frame UV lines may be dominated by non-gravitational motions, e.g., due to supernovae and galaxian winds.
Current attempts to measure dynamical masses for z ~ 3 Lyman break galaxies using Keck and the VLT (0.6" slit)13 find a relatively narrow range of [OIII] line widths, 50-100 km/s. Because the typical half-light radii of Lyman break galaxies is 0.2-0.4" in both the rest-frame UV and the optical (based on WFPC2 and NICMOS (near-infrared camera and multi-object spectrometer) imaging with HST)14,15 the measured line widths correspond to dynamical masses of ~1010M if the line widths are entirely gravitational. The relatively small masses that are obtained are perhaps surprising given the popular interpretation that Lyman break galaxies are the progenitors of the bulk of bulge-dominated galaxies. More likely, the measured widths of the emission lines reflect only the mass contained within the central high surface brightness regions of the galaxies (see Figure 8).
Thus, a potential role of the GSMT is to provide high angular resolution spectroscopic capability that can measure the velocity field as a function of position in the galaxy in order to confirm the gravitational nature of the velocity field. In addition, with the higher sensitivity of the GSMT, it may be possible to extend the study of velocity fields to lower surface brightness regions in order to better sample the depth of the galaxian potential.
Chemical Abundances: Metallicity measurements could potentially be used to help place galaxy populations in an evolutionary context. For example, Carollo & Lilly16 have used the technique described below to study the metallicities of star forming galaxies in the redshift range 0.5 < z < 1.0. From their preliminary results, they argue that this galaxy population appears to have a fairly high metallicity, i.e., ~0.4 solar. Therefore, this population is less likely to be comprised of dwarf galaxies brightened by a burst of star formation, and more likely to be the progenitors of present-day, massive, metal-rich galaxies. The gas phase metallicities of distant galaxies in the redshift range z 2-4 can be probed using nebular analysis techniques originally developed for use in local HII regions. One such popular technique measures the O/H abundance ratio using the familiar suite of rest-frame optical diagnostics [OII]3272, H, [OIII]4363, [OIII]4959,5007, [NII]6584 which will appear in the near-infrared at z = 1-4. A recent analysis by Kobulnicky et al.17 has shown that even if they are spatially unresolved, emission line spectra can provide a reasonably reliable estimate of nebular abundances. They estimate that the strong line index R23 = (I3727 + I4959 + I5007)/H can produce an oxygen abundance accurate to ±0.2 dex if the [OII]3272, H, and [OIII]4959,5007 lines are measured with s/n 8. Using this technique, Pettini et al.11 were able to conclude that Lyman break galaxies do not have super-solar abundances, but are significantly more metal-rich than damped Ly systems at the same epoch.
The only apparent difficulty with this approach is that the R23 index is double-valued, and either the measurement of [OIII]4363 (to measure temperature) or the [NII]H (metallicity sensitive) line ratio is needed to break the degeneracy. Resolving this degeneracy is one of the factors that limits current metallicity measurements. As is well-known, [OIII]4363 is typically very weak, ~1% the strength of H,17 and often [NII]6584 is not in a favorable part of the atmosphere (either due to OH emission or telluric absorption) at redshifts where the other diagnostics are accessible. More generally, these observations challenge the sensitivity of current 8-10-m telescopes even when measuring the stronger line diagnostics and when observing the integrated light of the galaxy (slit size 0.6'). At the required s/n, observations of the [OII], H, [OIII], H, and [NII] lines typically require one night per galaxy with existing 8-10m telescopes.
Thus, the potential roles of the GSMT are: (1) to make higher spatial resolution measurements of gas phase chemical abundances; (2) to use its increased sensitivity to measure the weak [OIII]4363 line; (3) to make these observations at higher resolution where more of the OH emission is resolved so that more diagnostics are simultaneously accessible; and (4) to employ multiplexing spectroscopic capability (e.g., multislits or multiple deployable integral field units (IFUs)).
Absorption Line Diagnostics: Current IFU observations of galaxies at z = 0 demonstrate the power of these systems to reveal the detailed kinematics and stellar populations of galaxies (see Fig.9)18 from studies of stellar absorption lines.
In order to carry out similar observations at higher redshifts, sensitivity is the primary challenge. These observations are beyond the reach of current 10-m class telescopes, but they may be possible to a limited extent with a 30-m telescope.
Due to the low surface brightness of the outer regions of galaxies, even a 30-m telescope will be able to probe only the high surface brightness regions of galaxies at the full diffraction limit of the telescope (µH = 22 mag/sq.arcsec), e.g., the inner regions of "giant" galaxies at z 1. Such high surface brightness regions appear to be relatively rare. Figure 10 illustrates the fraction of the sky that is accessible to long IFU observations with a 30-m GSMT. In this illustration, a region of the HDF (Hubble Deep Field) is assumed to be observed with an IFU with an aperture size of 0.1" x 0.1" at R = 2000 centered at 1.6µm. The calculated signal-to-noise assumes photon noise from the source and sky (µsky = 18.5 AB magnitudes per square arcsec, appropriate to the H-band sky continuum between the OH lines) is the dominant source of noise. The top panel shows a 75' x 30' subsection of the NICMOS HDF. In the lower panel, pixels are highlighted where a signal-to-noise 10 is obtained on the continuum in a 6 hr observation.
This simulation shows that there are ~10 galaxies per square arcminute whose central few arcseconds are typically above the limiting surface brightness threshold accessible to a 30-m telescope. Because these regions cover approximately 0.7% of the sky, the focal plane need not be fully sampled, and a system of deployable IFUs may be an appropriate instrumentation solution.
Required Instrumentation: Carrying out these programs will require design of deployable integral field units feeding multiple spectrographs. The design of the IFUs will need to enable optimization to the targets-thus requiring pixel sizes from 0.1" x 0.1" for cases now thought typical, to the full diffraction limit (0.02 x 0.02). In Section 4.7 we provide an overview of an instrument concept matched to these requirements.
- Steidel, C.C., Adelberger, K.L., Giavalisco, M., Dickinson, M, Pettini, M. "Lyman-Break Galaxies atz>~4 and the Evolution of the Ultraviolet Luminosity Density at High Redshift". ApJ 519, 1 (1999)
- Pettini, M., Steidel, C.C., Adelberger, K.L., Dickinson, M., & Giavalisco, M. "The Ultraviolet Spectrum of MS 1512-CB58: An Insight into Lyman-Break Galaxies". ApJ 528, 96 (2000)
- Spinrad, H., Dey, A., Stern, D., Dunlop, J., Peacock, J., Jimenez, R., & Windhorst, R. "LBDS 53W091: an Old, Red Galaxy at z=1.552". ApJ 484, 581 (1997)
- Pettini, M., Kellogg, M., Steidel, C.C., Dickinson, M., Adelberger, K.L., & Giavalisco, M. "Infrared Observations of Nebular Emission Lines from Galaxies at Z ~= 3". ApJ 508, 539 (1998)
- Songaila, A., & Cowie, L.L. "Metal enrichment and Ionization Balance in the Lyman Alpha Forest at Z = 3". AJ 112, 335 (1996)
- Steidel, CC, Giavalisco, M, Pettini, M., Dickinson, M., & Adelberger, K.L. "Spectroscopic Confirmation of a Population of Normal Star-forming Galaxies at Redshifts Z > 3". ApJ 462, L17 (1996)
- Fan, X. et al. "A Survey ofz>5.8 Quasars in the Sloan Digital Sky Survey I: Discovery of Three New Quasars and the Spatial Density of Luminous Quasars at z~6". AJ, 122 (6), 2833 (2001)
- Hu, E.M., McMahon, R.G., & Cowie, L.L. "An Extremely Luminous Galaxy at Z = 5.74". ApJ 522, L9 (1999)
- Haiman, Z., & Loeb, A. "Signatures of Stellar Reionization of the universe". ApJ 483, 21 (1997)
- Haiman, Z., & Loeb, A. "Observational Signatures of the First Quasars". ApJ 503, 505 (1998)
- Dey, A., Spinrad, H., Stern, D., Graham, J.R., & Chaffee, F.H. "A Galaxy at z=5.34". ApJ 498, L93 (1998)
- Barton, E.J., Geller, M.J., Bromley, B.C., van Zee, L., & Kenyon, S.J. "The Tully-Fisher Relation as a Measure of Luminosity Evolution: A Low-Redshift Baseline for Evolving Galaxies". AJ 121 (2), 625 (2001)
- Pettini, M., Shapley, A.E., Steidel, C.C., Cuby, J.-G., Dickinson, M., Moorwood, A.F.M., Adelberger, K.L., & Giavalisco, M. "The Rest-Frame Optical Spectra of Lyman Break Galaxies: Star Formation, Extinction, Abundances, and Kinematics". ApJ 554, 981 (2001)
- Giavalisco, M., Steidel, C.C., & Macchetto, D. "Hubble Space Telescope Imaging of Star-forming Galaxies at Redshifts Z > 3". ApJ 470, 189 (1996)
- Dickinson, M. "The first galaxies: structure and stellar populations". Philos. Trans. R. Soc. London A, 358 (1772), 2001 (2000)
- Carollo, C.M., & Lilly, S.J. "The Metallicity of 0.5 < Z < 1 Field Galaxies". ApJ 548, L153 (2001)
- Kobulnicky, H.A., Kennicutt, R.C., & Pizagno, J.L. "On Measuring Nebular Chemical Abundances in Distant Galaxies Using Global Emission-Line Spectra". ApJ 514, 544 (1999)
- Davies, R.L. et al. "Galaxy Mapping with the SAURON Integral-Field Spectrograph: The Star Formation History of NGC 4365". ApJ 548, L33 (2001)