Limits on the Star Formation Rates of z>2 Damped Ly$\alpha$ Systems from H$\alpha$ Spectroscopy

Andrew J. Bunker¹ Department of Astrophysics, University of Oxford, OX1 3RH, UK

Stephen J. Warren Imperial College, Department of Physics, Blackett Laboratory, Prince Consort Road, London SW7 2BZ, UK

D. L. Clements Institut d'Astrophysique Spatiale, Bâtiment 121, Université Paris XI, F-91405 Orsay CEDEX, France

Gerard M. Williger NASA Goddard Space Flight Center, Code 681, Greenbelt, Maryland 20771, USA

Paul C. Hewett Institute of Astronomy, Madingley Road, Cambridge CB3 0HA, UK

(1)Present address: NICMOS Postdoctoral Researcher, Department of Astronomy, University of California at Berkeley, 601 Campbell Hall, Berkeley CA 94720, USA
email: bunker@bigz.Berkeley.EDU

         

 

Abstract:

We present the results of a long-slit K-band spectroscopic search with CGS4 on UKIRT for H$\alpha$ emission from the objects responsible for high-redshift (z>2) damped Ly$\alpha$ absorption systems. The objective was to measure the star-formation rates in these systems. However, no H$\alpha$ emission was detected above our $3\,\sigma$ limits of $f\mathrel{\hbox to 0pt{\lower 3pt\hbox{$\mathchar''218$ }\hss} \raise 2.0pt\hbox{$\mathchar''13C$ }}10^{-19}$Wm^-2, corresponding to star formation rates $\mathrel{\hbox to 0pt{\lower 3pt\hbox{$\mathchar''218$ }\hss} \raise 2.0pt\hbox{$\mathchar''13C$ }}10\,h^{-2}\,{\rm M_{\odot}}\,{\rm yr}^{-1}$ (q₀=0.5). These upper limits are more meaningful than those from searches for Ly$\alpha$ emission because the H$\alpha$ line is unaffected by resonant scattering. For q₀=0.5 our limits are in conflict with the star formation rates predicted under the assumption that the high-z DLAs are the fully-formed galactic-disk counterparts of today's massive spiral galaxies. Deeper spectroscopy is needed to test this picture for q₀=0.0. A programme of NICMOS imaging observations currently underway, combined with VLT spectroscopy, will provide a detailed picture of the link between DLAs and young galaxies.

galaxy formation, quasar absorption lines, damped Ly$\alpha$ systems, star formation rates, infrared spectroscopy

Introduction

The history of star formation in the Universe is a topic of enormous current interest (e.g., Madau et al. 1996). The damped Ly$\alpha$ absorption systems (DLAs, Wolfe et al. 1986) contain most of the neutral gas in the Universe, and from their redshift distribution, and the measured column densities, the evolution in the co-moving density of neutral gas $\Omega_g$ can be measured (e.g., Lanzetta et al. 1991). Then the analysis of the variation of $\Omega_g$ with redshift allows the measurement of the history of star formation in the Universe (Pei & Fall 1995) provided the consequences of dust obscuration are accounted for.

This approach to the history of star formation unfortunately tells us nothing about how galaxies are assembled. One school of thought has DLAs being the (large) progenitors of massive spiral disks (e.g., Lanzetta et al. 1991; Prochaska & Wolfe 1997). However, Møller & Warren (1998) have recently shown that the impact parameters of the few detected galaxy counterparts of high-redshift DLAs are small (in the context of this debate) and that the space density of the DLAs at high-redshift is probably much higher than the space density of spiral galaxies today.

Searches for Star Formation in Damped Systems

Here we present the results of a spectroscopic survey for H$\alpha$$\lambda$656.3nm emission from 8 damped absorption systems at 2.0<z<2.6, along the line-of-sight to 6 high-redshift quasars. At these redshifts H$\alpha$ appears in the near-infrared K-window. The results are relevant to the debate on the nature of the DLAs, for if the DLAs are the counterparts of today's spiral galaxies the associated H$\alpha$ emission should be detectable. The rate of depletion of the cosmic density of neutral gas can be used to compute a universal star formation rate. The average star formation rate in each DLA depends then on their space density, so that high measured rates of star formation would provide support for the view that the DLAs are massive galaxies already in place at high-z. A low measured star formation rate on the other hand would be in agreement with the hierarchical picture.

There have been extensive searches for Ly $\alpha\,\lambda$121.6nm emission from DLAs but with limited success (e.g., Smith et al. 1989; Hunstead, Pettini & Fletcher 1990; Lowenthal et al. 1995). This is generally thought to be due to the fact that resonant scattering greatly extends the path length of Ly$\alpha$ photons escaping through a cloud of neutral hydrogen so that even very small quantities of dust can extinguish the line (Charlot & Fall 1991). Because the effective extinction can be extremely large this has the consequence that non-detections do not provide any useful information on the star formation rates in the DLAs. The H$\alpha$ line, although intrinsically weaker by a factor $\sim 10$, lies at a longer wavelength where the extinction is smaller, and is not resonantly scattered. In consequence a search for H$\alpha$ emission from DLAs may be more efficient than a search for Ly$\alpha$.

Our Near-IR Spectroscopic Survey

  

Figure: The measured $3\,\sigma$ upper limits to the SFRs in ${\rm M_{\odot }\,yr}^{-1}$ for the sample of 8 DLAs for $q_{\circ }=0.5$. The curves are the predicted survey-averaged SFRs from the closed-box models of Pei & Fall (1995), under the assumption that the DLAs are the progenitors of present-day spiral galaxy disks. The predictions of the hierarchical picture will lie below these curves in proportion to the ratio of the space density of galaxy sub-units at any redshift to the space density of spiral galaxies today.

With the CGS4 spectrograph on the 3.8-m UK Infrared Telescope (UKIRT) we have undertaken a search for line emission from 8 high-redshift DLAs (2.0<z<2.6) near the expected wavelength of H$\alpha$. The observations and data reduction are detailed in Bunker (1996). No lines were detected at $>3\,\sigma$ significance above the quasar continuum. Our long-slit K-band spectra were typically 1-hour each and used the 2.5-arcsec wide slit ( $10\,h^{-1}$kpc at $z\approx 2.3$, q₀=0.5 & $h=H_{0}\,/\,100{\rm km\,s}^{-1}\,{\rm Mpc}^{-1}$ throughout). With a 3-arcsec extraction width, the $3\,\sigma$ upper-limits lie in the range $(1.0 - 1.6)\times 10^{-19}$Wm^-2 for spectrally-unresolved line emission. The resolution of CGS4 in this configuration is $650\,{\rm km\,s}^{-1}$ FWHM.

We use the upper limits on H$\alpha$ line luminosities to constrain the star formation rates in these systems, based on the prescription of Kennicutt (1983), where a star formation rate (SFR) of $1\,{\rm M_{\odot}\,yr}^{-1}$ generates a line luminosity in H$\alpha$ of $11.2\times 10^{33}$W. The limits to the SFRs lie in the range $(8.4 - 18)\,h^{-2}\,{\rm M_{\odot}\,yr}^{-1}$, although the conversion between H$\alpha$ line luminosity and SFR is somewhat uncertain and depends on the assumed IMF.

Testing the Large Disk Hypothesis

We compare the measured upper limits to the star formation rates in our sample against predictions based on the assumption that DLAs are the progenitors of today's spiral galaxies. We begin with the analysis by Pei & Fall (1995) of the observed rate of decline of the cosmic density of neutral gas $\Omega_g({\rm obs})$ measured from surveys for DLAs. At any redshift $\Omega_g({\rm obs})$ will be an underestimate of the true cosmic density $\Omega_g({\rm true})$ because quasars lying behind DLAs will suffer extinction, and may therefore drop out of the samples of bright quasars used to find the DLAs. Pei & Fall correct for this bias, accounting in a self-consistent manner for the increasing obscuration as the gas is consumed and polluted as star formation progresses. In this way they determine the evolution of $\Omega_g({\rm true})$, and so the SFR per unit volume.

Although the analysis of Pei & Fall provides the SFR per unit volume at any redshift, it gives no information on the SFR in individual galaxies. For the hypothesis of large disks of constant co-moving space density, assuming that the SFR in a DLA is proportional to the present-day galaxy luminosity L(0), we can predict the SFRs in the population of DLAs at any redshift. Figure 1 plots the results of this calculation, showing the predicted survey-averaged star formation rate for a sample of DLAs for $q_{\circ }=0.5$. Seven of the eight DLAs lie below the curve in this plot. The significance of this result is reduced by the fact that the solid angle over which we have searched for H$\alpha$ emission is smaller than the expected solid angle of the large disks. Despite this we would still have expected to detect 2 or 3 systems with average SFRs twice as large as our upper limits. These results then provide support for the hierarchical picture. For $q_{\circ}=0.0$ deeper limits are required to distinguish between the two pictures. A more detailed treatment of this survey and its implications is given in Bunker et al. (1998).

Future Work

A decisive test can be made with the latest generation of near-IR instrumentation. Deep H-band imaging with HST/NICMOS (Warren et al., GO-7824) will reveal whether the galaxies responsible for damped absorption are indeed in sub-L^* pieces, and the opportunities afforded by the largest ground-based telescopes such as the VLT should enable the accurate measurement of star formation rates in these systems. Combined with spectroscopy of metal lines, in this way we will build up a picture of the history of assembly, gas depletion, and chemical evolution in the population of damped absorbers.

We would like to thank the Organizing Committee of the ``NICMOS & the VLT'' Sardinia meeting. We are grateful for the excellent support we received while observing at UKIRT. We thank Mike Fall, Palle Møller & Hy Spinrad for useful discussions, and Max Pettini & Richard McMahon for details of some of the damped systems included in our survey. AJB was supported by a PPARC studentship while at Oxford, and a NICMOS-IDT postdoctoral position at U.C. Berkeley.