cold gas at high redshift: proceedings of a workshop celebrating the 25th anniversary of the...

COLD GAS AT HIGH REDSHIFf

ASTROPHYSICS AND SPACE SCIENCE LIBRARY

VOLUME 206

Executive Committee

W. B. BURTON, Ste"ewacht, Leiden, The Netherlands J. M. E. KUUPERS, Faculty of Science, Nijmegen, The Netherlands

E. P. J. V AN DEN HEUVEL, Astronomical Institute, University of Amsterdam, The Netherlands

H. VAN DER LAAN, Astronomical Institute, University of Utrecht, The Netherlands

Editorial Board I. APPENZELLER, Landessternwarte Heidelberg-Konigstuhl, Germany

J. N. BAHCALL, The Institute for Advanced Study, Princeton, U.S.A. F. BERTOLA, Universittl di Padova, Italy

W. B. BURTON, Sterrewacht, Leiden, The Netherlands J. P. CASSINELLI, University of Wisconsin, Madison, U.S.A.

C. J. CESARSKY, Centre d' Etudes de Saclay, Gif-sur-Yvette Cedex, France J. M. E. KUIJPERS, Faculty of Science, Nijmegen, The Netherlands

R. McCRAY, University of Colorado, JILA, Boulder, U.sA. P. O. MURDIN, Royal Greenwich Observatory, Cambridge, U.K.

F. PACINI,Istituto Astronomia Arcetri, Firenze, Italy V. RADHAKRISHNAN, Raman Research Institute, Bangalore, India

F. H. SHU, University of California, Berkeley, U.SA. B. V. SOMOV, Astronomical Institute, Moscow State University, Russia

R. A. SUNYAEV, Space Research Institute, Moscow, Russia S. TREMAINE, CITA, University of Toronto, Canada

Y. TANAKA, Institute of Space & Astronautical Science, Kanagawa, Japan E. P. J. VAN DEN HEUVEL, Astronomical Institute, University of Amsterdam,

The Netherlands H. VAN DER LAAN, Astronomical Institute, University of Utrecht,

The Netherlands N. O. WEISS, University of Cambridge, U.K.

COLD GAS AT HIGH REDSHIFT

Proceedings of a Workshop Celebrating the 25th Anniversary of the Westerbork Synthesis Radio Telescope,

held in Hoogeveen, The Netherlands, August 28-30, 1995

Edited by

M. N. BREMER P.P. VANDER WERF H. J. A. ROTIGERING

Leiden Observatory, The Netherlands

and

C. L. CARILLI

National Radio Astronomy Observatory, Socorro, New Mexico, U.S.A.

KLUWER ACADEMIC PUBLISHERS DORDRECHT I BOSTON I LONDON

A C.I.P. Catalogue record for this book is available from the Library of Congress.

ISBN-13: 978-94-010-7273-1 DOl: 10.1007/978-94-009-1726-2

e-ISBN-13: 978-94-009-1726-2

Published by Kluwer Academic Publishers, P.O. Box 17,3300 AA Dordrecht, The Netherlands.

Kluwer Academic Publishers incorporates the publishing programmes of

D. Reidel, Martinus Nijhoff, Dr W. Junk and MTP Press.

Sold and distributed in the U.S.A. and Canada by Kluwer Academic Publishers,

101 Philip Drive, Norwell, MA 02061, U.S.A.

In all other countries, sold and distributed by Kluwer Academic Publishers Group,

P.O. Box 322, 3300 AH Dordrecht, The Netherlands.

Printed on acid-free paper

All Rights Reserved © 1996 Kluwer Academic Publishers

Softcover reprint of the hardcover 1st edition 1996 No part of the material protected by this copyright notice may be reproduced or

utilized in any form or by any means, electronic or mechanical, including photocopying, recording or by any information storage and

retrieval system, without written permission from the copyright owner.

CONTENTS

Preface ................................ xi

Introduction: Cold Gas at High Redshift

Cold Gas at High Redshift. . . . . . Colin A. Norman and Robert Braun

Cold Gas and Evolution at Low to Moderate Redshift

CO in Ultraluminous and High z Galaxies . . . . . . . . . . . . 25 N.Z. Scoville, M.S. Yun, and P.M. Bryant

Ultraluminous Infrared Galaxies: Dissipation in Forming Spheroidal Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . .. ;n

Paul P. van del' Werf

The Neutral Hydrogen Distribution in Luminous Infrared Galaxies 47 J.E. Hibbard and M.S. Yun

Molecular Gas and Dust in Infrared Luminous Galaxies . . 55 U.Lisenfeld, R.E. Hills, S.J.E. Radford, and P.M. Solomon

The Evolution of the Far-infrared Galaxy Population. 61 Michael Rowan-Robinson

The European Large Area ISO Survey: ELAIS S.J. Oliver

77

Keck Observations of J-lJy Radio Sources: Hints to Galaxy Evolution 85 James D. Lowenthal and David C. Koo

VI

Theoretical Aspects

Small Scale Structure and High Redshift HI. . . . . . . . . . .. 93 D.H. Weinberg, L. Hernquist, N.S. Katz, and J. Miralda-Escude

Are the Lyman Alpha Forest "Clouds" Expanding Pancakes? Some Theoretical Implications of the Recent Size Determinations of Lyo: Absorbers ....................... , 109

M.G. Haehnelt

On the Distribution of Intergalactic Clouds Stanislaw Bajtlik

Disk Galaxies at z = 0 and at High Redshift . G. Kauffmann

Warm Gas at High Redshift. Clues to Gravitational Structure Formation from Optical Spectroscopy of Lyman Alpha Ab-

115

121

sorption Systems . . . . . . . . . . . . . . . . . . . . . 137 Michael Rauch

Gas in Clusters

H I Imaging of Clusters. Jacqueline van Gorkom

An H I Survey of the Bootes Void A. Szomoru

An H I Study of Ursa Major Spirals. Dark Matter in Spirals and

145

159

the TF -relations ........................ 165 M.A. W. Verheijen

H I at High Redshift A. G. de Bruyn

Butcher-Oemler Effect and Radio Continuum K.S. Dwarakanath and F.N. Owen

Warm Molecular Gas in AGNs and Cooling Flows Walter Jaffe, Malcolm Bremer, and Roderick Johnstone

The Search for Cold Gas in the Intracluster Medium Christopher P. O'Dea and Stefl A. Baum

171

195

199

X-ray Observations of Cold Gas in Clusters . . . . . . . . . . . . 205 R.M. Johnstone

Absorption Measurements

Absorption Measurements of Molecular Gas F. Combes and T. Wiklind

ANew Molecular Absorption Line System. The Gravitational

Vll

215

Lens PKS 1830-211 at z = 0.88582 . . . . . . . . . . . . .. 227 T. Wiklind and F. Combes

Deep HST Imaging of a Damped Lyman a Absorbing Galaxy at z = 2.81 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2~{3

P. M ¢ZZer and S.J. Warren

Associated X-ray Absorption in High Redshift Quasars. Martin Elvis and Fabrizio Fiore

Opacity of Singly Ionized Helium from Very Tenuous Intergalactic

239

Absorbing Gas ........................ , :l45 W. Zheng

Heavy Elements in the Lyman-a Forest: Abundances and Clus-tering at z = 3 ......................... 249

D.S. Womble, W.L. W. Sargent, and R.S. Lyons

Absorption Lines from Cold Gas in Extragalactic Superbu bbles. Ti II and Ca II Absorption Towards the Superbubhle LMC2 in the Large Magellanic Cloud ................ 255

Adeline Caulet

Gravitational Lenses and Damped Lya Systems. A. Smette

261

HI 21 em Line Observations of Damped Lya Systems. . . . . .. 267 C. Carilli, W. Lane, A.G. de Bruyn, R. Bral1,n, and G./(. Miley

A Radio Search for High Redshift H I Absorption . . . . . . 279 J.N. Chengalur, A.G. de Bruyn, R. Braun, and C. Carilli

Testing z ~ 0 Analogs for the Damped Lyman a Absorbers 285 F.H. Briggs and E. Sorar

Vlll

Interstellar Medium in Distant Galaxies

Molecular Gas in High Redshift Galaxies .... Simon J. E. Radford

CO, C I and (Possibly) HCN in the Cloverleaf Quasar Richard Ba'rvainis

301

Searching for Molecular Gas in a Radio Galaxy at Redshift ;{.8 :W5 R.J. Ivison, P. Papadopoulos, E.R. Seaquist, and S.A. Eales

Thermal Emission from Dust in High-z Galaxies .. . . 311 David H. Hughes

Submillimetre Observations of QSOs at Redshifts z > 4 :{25 K. G. Isaak, R. E. Hills, S. Wlzitlzington, and R. G. McMahon

1.25 mm Detection of 7 Radioquiet QSOs with Very High z .. :{31 A. Omont, R.G. McMahon., P. Cox, E. Kreysa, and J. Berge-ron

Radiative Transfer Models for IRAS F10214+4724 and other Hy-perluminous Galaxies. . . . . . . . . . . . . . . . . . :{:n

S.M. Green and M. Rowan-Robiniwn

IR and X-rays from IRAS F10214+4724: a Hidden AGN? 34;{ G.L. Granato, A. Franceschini, and L. Danese

Gas and Dust in high z radio galaxies . . . . . . . . . . . . 349 P.J. McCarthy

Kinematics and H I Absorption in Lya Halos around z > 2 Radio Galaxies . . . . . . . . . . . . . . . . . . . . . . . . . . ;{6;{

Huub Rottgering

The Lya Velocity Field of the Radio Galaxy 4C41.17 (z = :{.8) with TIGERjCFHT . . . . . . . . . . . . . . . . . . ;{67

B. Rocca- Volmerange

The Effect of a Surrounding Cooling Flow on a Powerful Radio Source . . . . . . . . . . . . . . . . . . . . , . . . . . :H3

M.N. Bremer, A.C. Fabian, and C.S. Crawford

An HST Look at Dust in 3CR Galaxies .... S.A. Baurn, S. de Kojj, W. Sparks, J, BiTetta, D. Golornbek, D. Macchetto, G. Miley, and P. McCarthy

Detailed Studies of the Lyman Alpha Kinematics in 2104-242 . ;{85 A.M. Koekernoer, W.J.M. van Breugel, P.J. McCarthy, and J. Bland-Hawthorn

Orientation Effects in Quasar Spectra: Dust and Obscuration . . ;{91 Joanne C. Baker and Richard W. Hunstead

IX

Effects of Dust and Resonance Scattering on the UV Spectrum of Radio Galaxies . . . . . . . . . . . . . . . . . . . . . . . .. :{97

M. Villar-Martin, L. Binette, and R.A.E. FosbuTY

HST Observations of Radio Galaxies at z"" 1 P.N. Best

Instrumental Developments

........ 403

Studies of Cold Gas in the Early Universe with Large Millimeter Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411

Robed L. Bmwn

Strategies for Galaxy Surveys in the Submillimetre Waveballd . . 42;{ A. W. Blain

Considerations for Detecting CO in High Redshift galaxies. . . . 129 Frank P. ismel and Paul P. van deT Wer!

Future Possibilities for Detecting H I at High Redshift .. . . . . 4:H Robed Bmun

SPH Simulations ofthe Early Universe. Performance of the Dwinge-100 Square Kilometer Array . . . . . . . . . . . . . . . . . . 451

D.R. ingmm, N. Katz, D.H. WeinbeTg, and L. Hernq'llist

Searches for H I Emission from Protoclusters using the Giant Metrewave Radio Telescope - Observational Strategies . . . . . 457

C. SwaTup

Index ................................. 46:1

PREFACE

Recent years have seen increasing evidence that the main epoch of galaxy formation in the universe may be directly accessible to observation. Angular fluctuations in the background relict radiation have been detected by various ground-based instruments as well as by the COBE satellite, and suggest that the epoch of galaxy formation was not so very early. Combined optical and radio studies have found galaxies at redshifts above 2.0, systems that at least superficially show the characteristics expected of large galaxies seen only shortly after their formation. And absorption lines in the spectra of quasars seem to be telling us that most cold gas at early to intermediate cosmological epochs was in clouds having roughly galaxy sized masses. What kinds of new observations will best help us study this high redshift universe in future? What new instruments will be needed? These are questions that loom large in the minds of the Dutch astronomical community as we celebrate 25 years of operation of the Westerbork Synthesis Radio Telescope. Celebration of this Silver Jubilee has included a birthday party (on 23 June, 1995), a commemorative volume looking at both the history and the future of the facility ("The Westerbork Observatory, Continuing Adventure in Radio Astronomy," Kluwer 1996), and an international workshop, held in the village of Hoogeveen on 28-30 August, 1995. That workshop, the proceedings of which are presented in this volume, focussed on one of the main scientific activities of the instrument, indeed of the Dutch community in general, exploration of the extragalactic universe via the observation of cosmic gas in its various manifestations. Of course, astronomy these days seldom confines itself to a single portion of the electromagnetic spectrum, so the scope of the workshop covered not only studies at radio frequencies but also in millimeter, infrared, optical and even higher frequency bands. In the event, 79 participants from 9 countries came to Hoogeveen to consider what can be learned about the early universe from the study of cold gas. As readers of the proceedings will agree, the state of knowledge in the field was well summarized at the workshop, even though the definition of "cold" used by participants clearly covered quite a broad range of temperatures! The interest of the Dutch community in the field was if anything strengthened and broadened at the workshop and our thinking about the future stimulated. On the short term that future will in any case include an upgraded and

xii

modernized Westerbork Synthesis Radio Telescope. Among other new features our telescope will soon be equipped with tunable receiver systems capable of observing the 21 cm line of neutral hydrogen at most redshifts out to beyond about 4.5. Which observations are most urgent with this new capability'! Which will complement most effectively IIew capabilities at other wavelengths? The workshop gave us a lot to think about. On the intermediate term we are looking forward first to the ESO Very Large Telescope, and then to the coming generation of large aperture synthesis arrays at millimeter wavelengths. How should we plan the evolution of our community to make optimum use of these powerful new capabilities? Again, the workshop gave us much to mull over. And in the long term, a very large new decimeter-wave instrument seems unavoidable if we are to probe the epoch of galaxy formation in detail. The astronomical community in the Netherlands plans to playa major role in bringing that telescope into being. The workshop helped us focus our thoughts on the power of the instrument for cosmology and galaxy evolution studies, and in the process yielded several new collaborations that will help define the detailed specifications of the instrument. All in all, a very successful workshop. On behalf of all those present let me formally record here our appreciation to those who did most of the organizational work - Malcolm Bremer, Huub Rottgering, Paul van der Werf, Chris Carilli, Bjorn Heijligers, Ronnie Hoogerwerf, Hedy Versteege-Hensel, Alain Smette, Rene Genee and Nico de Vries. The workshop was organized under the auspices of the European Association for Research in Astronomy (EARA), which consists of the Institute of Astronomy (Cambridge), the Institut d' Astrophysique (Paris) and the Leiden Observatory. The Scientific Organizing Committee consisted of Frank Briggs (chair), .Jacqueline van Gorkom, Chris Carilli, Richard Hills, Dick Hunstead, Malcolm Longair, George Miley, Alain Omont, Martin Rees, Huub Rottgering, Michael Rowan-Robinson, Nick Scoville, and Paul van der Werf.

Harvey Butcher Director, NFRA

Acknowledgement. The organizers would like to thank the Netherlands Foundation for Research in Astronomy, the Leids Universiteits Fonds and the Leids Kerkhoven-Bosscha Fonds for providing the financial support that made this workshop possible.

INTRODUCTION: COLD GAS AT HIGH REDSHIFT

COLD GAS AT HIGH RED SHIFT

COLIN A. NORMAN

Dept. of Physics and Astronomy Johns Hopkins University and Space Telescope Science Institute

AND

ROBERT BRAUN Netherlands Foundation for Research in Astronomy

Abstract. We discuss the current observational and theoretical issues concerning cold gas at high redshift and present simulations showing how a number of observational issues can be resolved with planned future instrumentation.

1. Introduction

The observable history of the universe is dominated by a long phase from the epoch ofrecombination (at redshift 1500) to the reheating and reionization phase (perhaps near redshift 7) when the entire intergalactic medium is cold neutral gas. Current limits from QSO absorption line studies place this epoch above red shift 5. The fluctuations in this gas are so small that it is difficult to see either in emission or absorption (Scott and Rees 1990). However, it is an interesting scientific goal to try to observe this cool component of the intergalactic medium at high redshift. The only objects we know something about at the highest redshifts are the quasars. The space density of high redshift quasars clearly exhibit a steep rise and fall about a redshift of 2-3 (Shaver 1995) and the rise may be associated with the onset of galaxy formation.

In adiabatic models, where massive pancake structures formed and subsequently lumps of order the size of galaxies fragmented out of their collapse, the atomic masses of the cool gaseous pancake structures were es-

3

M. N. Bremer et al. (eds.), Cold Gas at High Redshijt, 3-21. © 1996 Kluwer Academic Publishers.

4 COLIN A. NORMAN AND ROBERT BRAUN

timated to be up to '" 1014 - 1015 M (0). If such masses of diffuse atomic gas existed at z '" 3.5, they would already have been detected by current searches (Wieringa, De Bruyn and Katgert 1992 and references therein). Their non-detection can now be understood in the light of the constraints set by microwave background studies and related research on the fluctuation spectrum (c.f. Scott, Silk and White 1995). The relative smoothness of the density fluctuations and the essentially mandated bottom-up nature of the galaxy formation process greatly limits the possibilities for directly observing proto-cluster size fluctuations in the cool gas phase. More ingenious methods, which probe both smaller and larger angular scales and in particular smaller masses, are likely to be required as discussed later.

Great hopes for this meeting lay in a number of reported observations of molecules observed at high redshift. However, while there are still very interesting as yet unconfirmed claims oflarge molecular masses of CO at high red shift the only well confirmed CO observations seen in emission are due to the two well known gravitational lensed objects the Cloverleaf and FSC 10214+4724 (Barvainis 1995, Scoville et at. 1995, and Frayer 1995). There have also been several detections of CO in absorption against background radio sources in the mm band but also associated with lensing (Combes and Wiklind, these proceedings).

With combined Keck and HST data, remarkable progress has been made in the study of the absorption lines of QSOs and the objects that are associated with the absorbing material. The population of Lyman Alpha clouds can have a number of progenitors as we shall discuss. Both the Damped Lyman Alpha (DLA) systems and the Lyman Alpha forest lines may account for a significant fraction of the currently observed baryonic content of galaxies (c.f. Storrie-Lombardi et at. 1995).

In this review of a very large subject we focus on a brief observational and theoretical overview of the subject of cool gas in the universe. In particular we present 9 figures that show how current and planned future instrumentation can detect and image cool gas at high red shift and indicate how such observations may help resolve some of the key issues.

2. Neutral Gas

There are now several new aspects to the study of Lyman Alpha absorption systems (c.f. Meylan 1995). From the point of view of this workshop it seems most interesting to emphasize that recently there has been a significant change in ideas about the origin of the Lyman Alpha systems. In particular they seem to have correlation scales of order'" 1 Mpc and cannot be associated directly with individual galaxies. Structure formation can produce sheetlike debris of low column density that can account for

COLD GAS AT HIGH REDSHIFT 5

many of the properties of the absorption lines. More generally, remarkable simulations presented at this meeting indicate that the distribution of the absorbers can be obtained in N-body /SPH simulations.

..........

10-12

10- 13

10- 14

10- 15

10- 18

N 10-17 e CJ ........ 10- 18

.......... 3, 10- 19 ....

10-20

10-21

10-22

10-23

10-24

.. , ............................... , ............................... 1 ............... \, .............. I .............................. I ................. .. · . . . . .. . . . ..

l "-i- l l HI ~bs. lin 2~ H<>:urs 1 l ............................................................................................

l l "'+ l s .. ~oo m~y. TB~103 ~. I1V~10 k~/s l .! ........ : ..... ~.! ........ : ........ ! ........ : ........ ! ........ : ........ ! ........ : ..

: : ~ : : : : : : : .j ....... + ...... ·~····f····· .. j ..... ···f·· .~. i······· -r .... ~i······· 'f'

· . • , , ,t7l ' • ~. •

t:::::::t:::::::l:::.:t~:::::L::::::t:::r:l:::::::t::::rt::::::t : : : : -f.- : : : : : .. . .. .. .. .. .. .. .. .. .. . .. , .. . . ..

·, .. ·······:········;········:········;·+····~···· .. ··i· ....... : ......... , .......... :.. ~ : ~ ~ ~ +~ ~ ~ ~ ~ · ........... . ...................................................................................................................................................... .. . .. . .. . .. .. .. .. • • .. • • I • .. • I .. .. .. . .. . . .. .. .. .. . .. . .. .. .. .. .. .. · .. . . .. . . . . . .. .. .. .. .. .. .. . .. . .. , .............................. ,. ....................... ,. ........................ , ..................... , ............ . · . .. . . . .. . . .. · . . .. . .. .. .. .. .. .. . .. . .. .. .. . .. .. .. . . . .. . . .. .. .

.. ~ ........ ~ .......... ~ .......... ~ .......... ~ ......... ~ ....... ~ ......... ~ ............ ~ ............ ~. · . . . .. . . . . . · . . .. . . .. . .. .. .. • • • .. • • I • • .. . . . . . . .. .. . · .. . .. . .. .. . . .. . ~ .. ...... -r ....... ~ ..... ···f ...... ··~······"·f·· ...... ~ ... ·····f··· +·r ·······f· , , , , , , , . .J:.. • .

. ~ ........ : ........ ~ ......... ; ........ ~ .......... :- ........ ~ ......... :- ....... -·t t-t=' ...... :-. · .. .. .. .. .. . .. .. .. .. .. .. . .. .. . . . · .. , , , ... t· · .. . . . .. . .. .. ..

.~ ......... :. ....... ~ ......... :. ........ ~ ........ :. ....... ~ ......... :. ....... ~ ..... .... :..

Figure 1. We show the limiting column densities that we can expect to detect with current and planned instrumentation in the red-shifted 21 cm line, A background flux of 100 mJy is assumed, as well as a hydrogen spin temperature of 1000 K and a linewidth of 10 km 13- 1 , We compare these limits with the known distribution (from Petitjean et al. 1993) of the column density of absorbers derived from QSO absorption line studies.

At low red-shifts, it is clear that HI emission maps going to ever fainter column densities such as the map of M81 by Yun et ai. (1994) are a most interesting compliment to the rapidly advancing knowledge we are obtaining from Keck and HST on the low column density environments of galaxies. There is no substitute for an unbiased spatial tracer of column density like that of an optically thin emission line. Unfortunately, the column density sensitivity in HI emission at a fixed physical resolution diminishes at least as rapidly as Ding, so that only the highest column density disks will remain


accessible out to large distances, and then only with the largest possible collecting areas, as we will see below.

However, lower column densities of HI can still be probed out to large distances using the HI 21 em line in absorption. In Figure 1 we show the 50' limiting column densities in the 21 cm line that we can expect to detect with the up-graded, frequency-agile WSRT (Westerbork Synthesis Radio Telescope) and the proposed SKAI (Square Kilometer Array Interferometer, described by Braun in these proceedings) in a 24 hour integration. These limits were calculated with the conservative assumptions that only a relatively faint background source of 100 mJy flux be available and that the mean spin temperature of the gas be 1000 K. A brighter background source or cooler spin temperature result in a linear improvement of the column density limit. These detection limits are superposed on the observed number distribution of Lyman Alpha absorbers as function of column density (from Petitjean et al. 1993).

It is dear from the figure that the WSRT will allow access to the entire distribution of Damped Lyman Alpha systems (NH I > 1020 .2 cm- 2 ), while the SKAI will also permit study of much of the column density range of the Lyman limit systems (1020 .2 > NHI > 1017 cm-2 ). The 21 cm data provide additional insight into the physical properties of the absorbing gas via an estimate of the effective spin temperature, as well as providing an opportunity to image absorber structure at milli-arcsec resolution utilizing VLBI (Very Long Baseline Interferometry). The equation relating the HI column density and the 21 cm line opacity is:

327rkvi1 Ts J rdV 3hc3 A21

1.83 X 1018 Ts J rdV cm- 2

independent of red shift , for V in units of km s-1.

(1)

(2)

Direct imaging in HI 21 cm emission is the only reliable method for determining atomic gas masses. Current efforts have been limited both by instrumental sensitivity and by accessible frequency coverage to redshifts less than about 0.1. The situation is summarized in Figure 2, where "Detection" and "Imaging" atomic masses are shown as a function of redshift for the WSRT and the SKAI for an integration time of 100 hours. "Detection" has been defined as requiring a 50' signal in a single 50 km S-1 velocity channel, while "Imaging" has been defined as requiring a 50' signal in each of 6 independent 50 km s-1 velocity channels. The dotted line between red-shifts of 0.2 and 2.5 for the WSRT indicates the frequency range where receiver systems, while available, are not yet optimized and

........ ~ 1013

II (;)

::r: ...; 1012

II (;)

C

010 11

~ --rn rn 1010 ttl ~ -::r: tID 109 s:: .... ...., .-S 108 .-.....J


fiI in 100 Hout;s .................................................. ,................................................................................... .......... .. .... . . . .

1070.01 0.1 1 10 Red-shift

7

Figure 2. We show the detection (5lT in 50 km s-l) and imaging (5lT in six channels of 50 km S-I) limits of atomic gas mass as a function of redshift with current and planned instrumentation.

are about a factor of 2-3 worse than shown. The atomic gas masses of the well-known nearby systems M33 and MIOI have been included for reference in the figure, as well as the atomic gas mass of the ultra-luminous FIR galaxy, III Zw 35.

As can be seen in the figure, gas-rich systems will soon be detectable out to red-shifts of a few tenths with the WSRT. The SKAI, on the other hand, will allow detection of even low mass spirals like M33 to z > I and gas-rich systems to z = 3 or more.

Since every narrow velocity interval is so sparsely populated with condensed atomic gas (at least since the epoch of re-ionization) observations of this type will not be source confusion limited, even with only a modest angular resolution of several arcmin. (This same comment applies to all emission line tracers of high redshift gas, except perhaps where the red-


shift has placed the line frequency near another emission line of Galactic or terrestrial origin.)

An important point to note is that the explicit redshift dependence of the equation for the atomic gas mass in terms of the observed 21 cm line integral is not often stated. For clarity, we give the equation below for an optically thin distribution of neutral hydrogen:

167rmHD'i J S dV 3hcA21 {1 + z) II

(3)

5 D'i J 2.35 X 10 (1 + z) SlIdV (4)

for the luminosity distance DL in Mpc, and the line integral in Jy km s-l.

3. Molecular Gas

Molecular hydrogen gas is seen directly in the optical band in only one high redshift QSO absorption-line system 0528-25 at redshift 2.8. In the millimeter band, four objects have now been observed which show high redshift absorption in various molecules (CO, HCO+, HCN, O2) generally associated with absorbers in gravitationally lensed systems (c.f. Combes and Wiklind, these proceedings). Actual conditions in proto-galaxies, etc. are not yet clear enough to make solid predictions, but it is obvious that molecular studies at high redshift have much to tell us in the near future. With conditions similar to, say, our Galaxy, gas phase and surface reactions produce molecular species readily on short time scales rv 106 - 107 yr. Molecular hydrogen will form rapidly once the density and column density are high enough. A thorough discussion of the physical conditions and the constraints imposed on the H2 species is given in Black et al. (1987). Shielding by dust may be a crucial ingredient but probably the most important parameter is the strongly evolving background radiation field.

The beautiful data on the two lensed objects that show CO emission at high redshift are well described at this meeting by Barvainis and Scoville for the Cloverleaf and F10214+4724 respectively.

Frayer (1995) reviews the current evidence for detection of CO emission at high redshift. Only very tentative detections have yet been made in nonlensed systems. When detected at modest redshift (z rv 0.1), the empirical Galactic conversion factor suggests molecular gas masses of a few times 1010 M0 concentrated within regions of a few kpc in diameter (Scoville et al. 1991). However, it has not yet been demonstrated that a similar conversion factor of CO luminosity to molecular hydrogen mass need apply under the extreme physical circumstances of circumnuclear starbursts. Even when

-~ 1013

II o

::z:: ~ 1012

II o

C

01011

::s ......., rn : 1010

::s N

::z:: b.O s:: .-

0001 .-S .-~


~o in 8 Hour~ ..................................................................... · . · . · .

..... 33 :

1 Red-shift

Figure 3. We show the detection (50" in 50 km S-1) and imaging (50" in six channels of 50 km S-1) limits of molecular gas mass utilizing the CO 2-->1 transition as a function of redshift with current and planned instrumentation.

multiple CO line transitions are observed, it is worrisome that they need not necessarily arise from regions sharing the same physical conditions, so that the line ratios may remain difficult to interpret.

An impression of the current and future capabilities for imaging molecular mass at high redshift via associated CO emission is given for an 8 hour integration in Figure 3. The empirical Galactic conversion factor (eg. Scoville et al. 1991) gives:

4 Dl J MH2 = 1.2 X 10 (1 + z) SC01-+0dV (5)

for the luminosity distance DL in Mpc, and the CO 1-+0 line integral in Jy km s-l. If another CO line transition is used then the constant in eqn. 5 should be scaled in accordance with the ratio of line luminosities


(in erg s-1). We have defined "detection" and "imaging" as before for the atomic gas mass above and assumed that the CO 2--+ 1 transition would be observed with a 3 times higher line luminosity than that of CO 1--+0. Nearby normal spirals and two ultra-luminous FIR systems are included for reference in the figure. The question marks are used to indicate the uncertainty in assigning a molecular mass to the observed emission line luminosity in the case of the extreme starburst systems.

From Figure 3 it is clear that the MMA (the proposed NRAO Millimeter Array) should allow study of unlensed ultra-luminous systems out to redshifts greater than 1, although normal spirals will only be accessible out to about a tenth.

105

qH in 24 Hour~ -If.) 10" l'-II

0 = .....; 1000 II

0 C

j 100 ........

S 10 =' ...J

= 0 bD 1 r:: .....

-0-) ..... 13 0.1 .....

...J ........................ ~ ......... -NGC253 .. : . .

1 Red-shift

Figure 4. We show the detection (5IT in 50 km s-l) and imaging (5IT in six channels of 50 km S-l) limits for OH mega-maser emission as a function of redshift with current and planned instrumentation.

Megamasers are frequently seen in association with relatively edge-on starburst galaxies and it may well be worth searching systematically for


megamasers at high redshift since their luminosities are so large. Megamaser emission in the OH, H2 CO and H20 lines is a valuable probe of circumnuclear kinematics with ultra-high angular resolution and once the theory adequately catches up with the observations it may also be of help in understanding the physical conditions in extreme star-bursting systems.

In Figure 4 we show the detection and imaging limits of OH megamaser emission in a 24 hour integration. Reference luminosities of the sources in III Zw 35 and IR20100-4156 are indicated as well as the kilomaser emission seen in NGC 253. Comparable luminosities to those of OH in III Zw 35 are also seen in the H2 CO and H2 0 megamaser sources (Henkel and Wilson 1990, Baan et al. 1993, Henkel et al. 1984). With the added frequency coverage of the upgraded WSRT, megamaser emission should already be detectable to red-shifts greater than about 1, while the added sensitivity of the SKAI should allow such sources to be studied in detail at any redshift.

There have been reports of ultra cold gas that could constitute a significant fraction of the dark matter in the Universe ( Lequeux, Allen and Guilloteau 1993, Pfenniger and Combes 1994, Gerhard and Silk 1995). The absence of such gas in the absorption line spectra of QSOs indicates that the covering factor of this gas in a sight-line to a distant QSO is less than :s; 1 %. This limit may be a severe constraint on the proposal that such cold gas is a major constituent of the Universe.

4. Dust

Some time ago, Ostriker and Heisler (1984) proposed that the observed fall off in QSO number density might be due to obscuration by dust. The excellent study of Shaver (1995) shows that this is not the case. More moderate obscuration is probably present giving variations in the inferred number counts as a function of redshift for QSOs of less than order unity. This is consistent with calculations done by Fall and Pei (1994).

The importance of radio surveys cannot be underestimated here since a complete radio survey can be used independent of the dust obscuration and as noted by Shaver (1995) quasars at red shift z = 6 can be easily seen once the target radio source is known.

Submillimetre observations at high redshift (Isaak et al. 1994, McMahon et al. 1994) show that dust masses at redshifts z = 4 - 5 of order 108 Me and temperatures of say 60 J( can already be detected. Protogalaxies may have such dust masses after an initial burst of star formation and a more or less immediate ( 106 - 107 yr) giant and supergiant dust producing phase. Conversion of observed continuum luminosities to actual dust masses remains very tricky while the emission spectrum is only poorly sampled and there may well be multiple temperature components present.


-LO 1034 l.'-\I

0 ::I: Dus~ Cont. in 8 H~urs

~ 1033 ................................................................................................ ~ .............................................. .. · . · . \I · .

0 c: N 1032 ::I: .........

rI.I

~ 1031 s.. CU -8 1030 ::s

....:I N ::I: 1029 0 0 t"J t\2

btl 1028

s:: .--' .-S

... M33 :

.-

....:I 0.1 1 Red-shifl

Figure 5. We show the limiting luminosity for a continuum observation at 230 GHz for detection (5cr) and imaging (30cr) of the dust continuum as a function of redshift for current and planned instrumentation. The solid curves are for a modified black body (v1. 5 B(T, v)) using a dust temperature of 60 K, while the dashed curves are for a dust temperature of 30 K.

In Figure 5 we illustrate the possibility for detecting dust continuum emission via an 8 hour observation at 230 GHz with the heterodyne receiver system of the JCMT and the proposed MMA. Dust emission spectra were calculated with a modified black body (v1.5 B(T, v)) using dust temperatures of 60 K (solid curves) and 30 K (dashed curves). The ratio of rest-frame 100 flm to 230 GHz flux density in these cases is 1800 and 360 respectively. "Detection" is defined as requiring a 50' signal and "Imaging" a 300' signal. Reference luminosities of normal spirals and the ultra-luminous systems III Zw 35 and B1202-0725 are indicated. Single (sub- ) millimeter dishes will be able to do better than the indicated heterodyne .JCMT performance through the use of high bandwidth bolometric detectors (such as SCUBA). This detection method is not yet applicable to coherent, high


resolution imaging with an interferometric array, although the development of "hot electron bolometers" may make this a possibility in the future.

From the figure it is clear that the dust continuum becomes more accessible to a 230 GHz observation beyond z = 1 for dust temperatures greater than about 30 K. However, current sensitivities will limit detection to extreme systems, like B1202-0725, and even then these will be preferentially found at z < 0.3 and perhaps at z > 3. The MMA will allow comparable detections on less extreme systems (with LFiR "only" f'V 1012 L0)' Dust continuum from normal spirals like M33 and M101 will still only be accessible in the local universe.

-~ 1 033 o:::rr----r---r--r-T-rT"T'Tr---r--T""'"'I-r-rT'TTT""--r--r-""T"""Ir-r"I"TTT-::::O

II o

::z:: ....; 1032

II o

C

til 1031

::z:: ""-rn ';;D 1030

s.. Q) -E 1029

.3 til

::z:: 1028 t!J

~ -tID 1027 C ....

"""" .... E .... ~

NT ~ont. in 24 Hqurs 81202 ...

1 Red-shift

Figure 6. We show the limiting luminosity for a continuum observation at 1.4 GHz for detection (50") and imaging (300") of the non-thermal continuum associated with massive star formation as a function of redshift for current and planned instrumentation. The curves assume a power law spectrum of the form S oc v-O 7.

Many authors have pointed out the excellent correlation of dust continuum emission and non-thermal radio continuum emission based on the


many thousands of nearby galaxies detected with IRAS. Although still not understood in detail, there appears to be a strong coupling of both emission tracers to the massive star formation rate. With this in mind, we have illustrated in Figure 6 the limiting 1.4 GHz luminosities of current and planned instrumentation as a function of redshift. A power law emission spectrum of the form S ex: v-O.7 has been assumed. Detection and imaging of luminous star-bursting systems should already be possible with the VLA out to z > 1. The greater sensitivity of the SKAI will allow even normal spiral galaxies to be visible out to cosmological distances.

In this case, of observing the faint continuum emission from distant sources, it is critical that enough angular resolution be employed so that source confusion does not limit the sensitivity of an observation. The deepest existing radio continuum observations, as well as experience with the HST, suggest that angular resolutions of 0.1-1 arcsec are sufficient to completely circumvent the problem of source confusion. It is for this reason that the curves in Figure 6 have been drawn for the VLA (the NRAO Very Large Array) and SKAI, where such angular resolutions will be achievable, rather than for the WSRT, for which continuum source confusion at faint flux levels will be a limitation.

5. Cosmology: The Cool Gas History of the Universe

Observational tests for the detection of the cool pre-ionization (z 2: 5) IGM have been considered by Scott and Rees (1990, also see Kumar et at. 1995). If the hydrogen spin temperature is greater than the CMB temperature (TR) at these epochs an emission signature from neutral hydrogen would be expected. Proto-cluster mass enhancements are likely to have total masses of 1015 M8 on proper scales of less than about 3 Mpc, corresponding to less than about 15 arcmin at z = 6. The instrument best-matched to this problem would have a comparable beam size of some 15 arcmin at an observing frequency of 200 MHz. The necessary telescope diameter of some 350 m corresponds roughly to that of the individual elements of the SKAI.

An observing mode that is being envisioned for SKAI is one whereby the auto-correlations of the individual elements are incoherently summed to give a VN increased sensitivity over an individual element, which still falls short by VN from the sensitivity of the coherently combined data, but has a factor of about 104 greater brightness sensitivity. In this mode atomic gas masses of 6 X 1011 M8 could still be detected at z = 6. As long as the mass fraction of neutral atomic hydrogen is greater than about 6 X 10-4 then proto-cluster enhancements should be seen. In the case of the GMRT (the Giant Meterwave Radio Telescope), the limiting neutral atomic fraction for proto-cluster detection is about 0.025 (Kumar et at. 1995).


An interesting alternate possibility is that the HI may be observable in absorption, particularly for high baryon density Universes where the effect of collisions can drive down the spin temperature, Ts, below the cosmic background radiation temperature, TR. For a clumpy gas distribution at high redshift the resulting spin temperature and column density variation can produce a patchy structure across the sky. Since the HI brightness temperature is given by:

(6)

the absorption detection signature will be a factor of TR/Ts stronger than that in emission. In the most extreme scenarios, Scott and Rees predict TR/Ts = 10, resulting in an easily detectable signal for instruments like SKAI and possibly the GMRT operating near 200 MHz.

Generally, massive structures are needed to produce a currently observable effect. However from the work of Steinmetz (1995) and Kauffman (these proceedings) it is clear that in the standard bottom up scenarios there are not many really big lumps of neutral gas at high redshift but there are many small lumps clustering up to large scales but only at the present epoch. A particularly interesting way to view this is with the tree diagrams in Lacey and Cole (1994) that indicate how the dark matter halos put themselves together hierarchically to form larger galaxies.

6. Active Galaxies, Radio Galaxies and Quasars

Although it has proved exceedingly difficult to detect the extended gaseous halo structure around protogalaxies (Djorgovski et al. 1995) it has been far more productive to look around active galaxies. Large masses of ionized and cool gas have been found around high redshift radio galaxies (c.f. McCarthy 1989). Similarly interesting results can be found in Rottgering et at. (1995) and Van Ojik (1995).

Very recently, however, detections of luminous Lyman Alpha emission from protogalaxies at high red shift are emerging from detailed studies using HST and Keck (Giavalisco et at. 1995, Moller et ai. 1995). Typically, we might expect the masses of HI associated with the Lyman Alpha emitter to be of order 1010 M8 although this depends on a number of uncertain parameters such as the ionization balance, etc. Figure 2 suggests that such atomic gas masses should be detectable with SKAI out to z '" ;i.

7. Galaxy Formation: Can it be Observed as Cool Gas?

Interestingly, there now seems to be a consensus building about the pattern of galaxy evolution from combined HST and ground based (CFHT, Keck)


-.... 0 - : : : Messier . 3 Spectr;a : : • • .; ................ ~ ••••••••••••• -t ...... ••• .. ·····t .. ··· .......... t .. • ........... -t .••••• - •••• ··too· .. _·

: 15 : Iiili ~: ~ :R 8 81il8 II: !ilii!ill all. III Ii ~:m:ll • ,. I U "... ,1 .. ....... .. ". " ... 1: .... '" .... ,u. .u

II .... .: ..................... .: ...................... -~ .. _ ........... -- ...... : -........ _ ............ ~ ............ -... -...... ;. .............. -_ ...... -~ ........ .. • • , .. , I •

Il:: : : : : :..,: .. ~ ............. ~ ............. ~ ............. : .......... ~ ...... ~ .... :'IIL' .......... : .... . .; 'gj I:: III ... ....

: : : : ...: : : · . . ....... _ ..... .. : .. ···· ...... ·:·· .. ··· ...... :· .. ··· ...... ·: .. ·~~t···· .... ··? .. ···~ .. ·· .. : ..... ..-·~--····-·- .. ···1"·--···------1·--·-·-------1-··;;""nr----too ... -I:: .... ..1.GIIIII.~ .... J.--.......... L ............ i ..... -........ i.

: ........... -~ ...i. ...... ...... .. .: ...... ~!"~e7. .. = .. 1 .. ~~.~ ••. ! ........... .

>. '1 i i "" i -rt.i 1 r::: cu

I"Il 0.1 cu c 0.01 ::3

0.001

:Messier .3 Spectr;a : :

-.... 'gj I:: III ... ....

:F::E:I:::J:::':Jr : ~. 1 : ~ 11UI: ; ; .; ............. .

..... I:: .... >. '1

..T .. ··· ...... T .. ··· ...... r .. ··· ...... ·: .. ··· .. ;~4·· .. ·· ...... \~· ........ T··· . _. r--"""·- _. _. 'Or" --_ ..... --- --1-·-" -- ---_.- "1---;;"miIii -. roo-~ _. -- .. -: ~t········· ·r···· ··~······-······1----····---·-~·-····-······!-··;;",.DA-'.' - .... ·f .. ·· __ ····-r····· -rt.i

I:: cu

I"Il

....; 1 I:: 0 0.1 tJ

0.01

0.001

Figure 7. Simulated spectra of the low mass spiral galaxy M33 are shown for frequencies between about 108 and 1014 Hz after being red-shifted to z = 0.25, 1 and 4, under the assumption of no spectral evolution. Instrumental sensitivities (10-) of existing and planned instruments are overlaid for both spectral line observations (top panel) and continuum observations (bottom panel). Spectral line IDs for some of the major emission lines are indicated at z = o.

-0 .... II

~

..; 'm s:: ca J.. .... .... s:: .... >. l. -rn s:: cu

fIJ cu s:: ~

....

. 5


1

0.1

0.01

0.001

109mTIm-,,~~-rTnmm~rnmmrr-~Trrmr-rrrr~~~rrmr-~

108

107

. essier ~O 1 Spectra : : .. "; ............. : ........... _-:-_ ...... -----!, .... __ ._----: ........ _----:- .... _-------:- .. _.

o , , , I • •

• , I I • • • · . . .. . · , . . . . . ................... ---_ ... _-----, ...... _---- ... -- ....... -_ ......... _--.- .. ,. ............. ,. .... . • • , I • • •

• • I I • • • .. .... • I • • • • • , . . . . .

• - ... - ••• _ •••• - ..... _ - - _ ••••• - - - _._._. _. __ - _. - -0 __ ••• _. _ ••• __ • _. _ ••••• _____ ~ ••• _ •• _____ ........ .

: • : : ~ I : · .. .

>. ~ 1000

...; s:: o

t..)

100

10

1

0.1

0.01

O.001Uil~~~~~~wlwO~1~OLU~1~O~1l~~WL-LLllum~~llW~LU

Frequency

17

Figure 8. Simulated spectra of the massive spiral galaxy MIDI are shown for frequencies between about 108 and 1014 Hz after being red-shifted to z = 0.25, 1 and 4, under the assumption of no spectral evolution. Instrumental sensitivities (10') of existing and planned instruments are overlaid for both spectral line observations (top panel) and continuum observations (bottom panel). Spectral line IDs for some of the major emission lines are indicated at z = o.


-"" 0 .... II ~

...; .; I:: CIS J.o ..... .... I:: .... >. 1--rn 1 c III fIl 0.1 III I:: 0.01 ~

0.001

-.... . ; I:: CIS J.o ....

.... I:: .... >. 1--rn c

cu fIl

...; 1 c 0 0.1 u

0.01

0.001

Figure 9. Simulated spectra of the luminous starburst galaxy III Zw 35 are shown for frequencies between about 108 and 1014 Hz after being red-shifted to z = 0.25, 1 and 4, under the assumption of no spectral evolution. Instrumental sensitivities (10) of existing and planned instruments are overlaid for both spectrallille observations (top panel) and continuum observations (bottom panel). Spectral line IDs for some of the major emission lines are indicated at z = o.


data (Lilly et al. 1995, Driver et al. 1995, Griffiths et al. 1995). Massive galaxies do not seem to be evolving whereas smaller dwarf irregulars seem to burst into life between a redshift of z = 0.5 - 1 and then fade away by the current epoch. This pushes back the epoch of massive galaxy formation to redshifts of order z 2: 3.

We note in passing that, apart from the standard collapse and infall of HI, there are large masses of hot gas such as those associated with cooling flows in clusters that have cooler denser material at their centers. In fact large HI masses are inferred from shadowing effects (Allen and Fabian 1994). What this might be like at higher redshift has been discussed recently by Nulsen and Fabian (1995). Interesting limits on the cold gas content of the intracluster medium for nearby clusters of galaxies indicate that the total cold neutral gas content in the central regions of such clusters is S; 109 M0 (O'Dea et al. 1995).

Current theories of galaxy formation (c.f. Navarro, Frenk and White 1995) indicate that typical galaxy masses increase as a function of redshift from dwarf galaxy sized objects at redshifts of order a few to more massive galaxies at redshift of order unity to cluster sized objects at the current epoch. We next illustrate how our observational capabilities overlay the red shifted spectral energy distributions of several galaxy types and masses.

In Figure 7 we show a simulated spectral energy distribution of the low mass spiral galaxy M33 redshifted to z = 0.25,1 and 4 assuming no spectral evolution. In the top panel we have overlaid the 10- sensitivity at a spectral resolution of 104 of a variety of existing and planned instruments on these spectra. Comparison of the instrument sensitivities with emission line intensities in the spectra illustrates out to what redshift such an object might be studied. In the lower panel the same spectra are overlaid with 10- continuum sensitivities of the same instruments. In this case the sensitivities should be compared with the flux densities of the continua to assess out to what redshift the object might be studied. Integration times of "one transit" were assumed which were typically 8 hours for ground-based telescopes and 104 seconds for satellite observatories. The various line and continuum emission components in the model spectra are described in detail in Braun (1992). A similar set of redshifted spectra and overlaid instrumental capabilities are shown for the luminous spiral galaxy M101 in Figure 8. In Figure 9 we show the same plots for the ultraluminous FIR starburst galaxy III Zw 35 including its observed megamaser emission in OH and H2 CO.

Comparison of the redshifted model spectra with our current and projected observational capabilities (in Figs. 7-9) gives us grounds for guarded optimism about our prospects for studying the galaxy formation process. Near z = 1 we should be able to give a very good characterization of the types of objects which have formed via their atomic masses and the lumi-


nosities of the molecular, dust and stellar components. The more massive and luminous- end of the distribution can be tracked all the way out to z > 4, while even the low mass end of the distribution should yield its secrets out to z rv 0.5. New instrumentation will be critical to realizing this goal. The unprecedented sensitivity of the VLT and Keck will be necessary to permit optical and near-IR spectroscopy to identify these distant systems. Similarly, the next generation of cm/dm and mm/sub-mm arrays (SKAI and the MMA) will be needed to ascertain the associated cool gaseous masses and its kinematics. And although ISO makes an important contribution to the intervening frequency interval, it is clear that a new mission with SIRTF (or better) sensitivity will be needed to effectively fill in the mid-IR to FIR gap.

Great progress is being made in studying the Universe at high redshift at present by work done with Keck and HST. After completing this paper and contemplating the results of the simulations it is clear that extraordinary progress can be made with the planned instrumentation. It is obvious how the proposed studies at longer wavelengths from low frequency radio to sub-millimeter can give vital information in our quest to understand the physics of the Universe at high redshift when it was a fraction of its current age.

We thank many of our colleagues for stimulating discussions of this interesting subject and, in particular, F. Briggs, G. de Bruyn, R. Ellis, A. Fabian, and M. Rees.

References

Allen, S. W. and Fabian, A. C. 1994, MNRAS 269, 409. Baan, W.A., Raschick, A.D. and Uglesich, R. 1993, ApJ 415,140. Barvainis, R. in IAU 170, CO: Twenty Five years of Millimeter-wave Spectroscopy, in

press. Black, J.R., Chafee, F. H. and Foltz, C.B. 1987, ApJ 317, 442. Braun, R. 1982, NFRA Note: Planning for the next decade and beyond. Cen, R., Miralda-Escude, J. Ostriker, J.P. and Rauch, M. 1994, ApJ 437, L9. Driver, S. et al. 1995, ApJ, in press. Fall, S.M. and Pei, Y.C. 1993, Ap J 402, 479. Frayer, D. T. PhD Thesis 1995, University of Virginia Gerhard, O. and Silk, J. 1995, ApJ submitted. Giavalisco, M. et al. 1995, ApJ, in preparation. Griffiths, R. et al. 1995, ApJ, in press. Henkel, C. and Wilson, T.L. 1990, A&A 229, 43l. Henkel, C., Giisten, R., Downes, D., Thurn, C, Wilson, T.L. and Biermann, P. 1984,

A&A 141, Ll. Isaak, K.G., McMahon, R.G., Hills, R.E. and Withington, S. 1994, MNRAS 269, L28. Katz, N. et al. 1995 ApJ, in press. Kumar, A., Padmanabhan, T. and Subramanian, K. 1995, MNRAS 272, 54. Lequeux, J., Allen, R.J. and Guilloteau, S. 1993, A&A 280, L3. Lilly, S. et al. 1995, ApJ, in press.


McCarthy, P. 1989 PhD Thesis, UC Berkeley. McMahon, R.G., Omont, A., Bergeron, J., Kreysa, E. and Haslam, C.G.T. 1994, MNRAS

267, L9. Meylan, G. 1995, QSO Absorption Lines (Springer). M!Illler, P. et al. 1995, ApJ, in preparation. Moore, B., Katz, N. and Lake, G. 1995, ApJ, in press. Navarro, J.F., Frenk, C.S. and White, S.D.M. 1995, MNRAS 275, 56. Nulsen, P.E.J. and Fabian, A.C. 1995, MNRAS, in press. O'Dea, C., Gallimore, J.F. and Baum, S. A. 1995, AJ, 109, 26. Ostriker, J.P. and Heisler, J. 1984, ApJ 278,1. Petitjean, P., Webb, J.K., Rauch, M., Carswell, R.F. and Lanzetta, K. 1993, MNRAS

262499. Pfenniger, D. and Combes, F. 1994, A&A 285, 94. R6ttgering, H., Hunstead, R.W., Miley, G.K. van Oijk, R. and Wieringa, M.H. 1995,

MNRAS, in press. Scott, D. and Rees, M.J. 1990, MNRAS 247, 510. Scott, D., Silk, J. and White, M. 1995, Science 268, 829. Scoville, N.Z., Sargent, A.I., Sanders, D.B. and Soifer, B.T. 1995, ApJ 366, L5 Scoville, N.Z., Yun, M.S., Brown, R.L. and Vanden Bout, P. A. 1995, ApJ Letters, in

press. Shaver, P. A. 1995 in Proceedings of the Texas Meeting, Munich, December 1994, in press Steinmetz, M. 1995, A&A, in press. Storrie-Lombardi, L.J., McMahon, R. G., Irwin, M.J. and Hazard, C. 1995, ESO Work-

shop on QSO Absorption Lines, ed G. Meylan, p. 47. Storrie-Lombardi, L.J., McMahon, R.G. and Irwin, M.G. 1995, MNRAS, in press. Thompson, D.J., Djorgovski, S. and Trauger, J.T. 1995, A&A, in press. van Oijk, R. 1995, PhD Thesis, University of Leiden. Wieringa, M.H., de Bruyn, A.G. and Katgert, P. 1992, A&A 256, 331. Yun, M.S., Ho, P.T.P. and Lo, K.Y. 1994, Nature, 372, 530.

COLD GAS AND EVOLUTION AT LOW TO MODERATE REDSHIFT

CO IN ULTRALUMINOUS AND HIGH Z GALAXIES

N.Z. SCOVILLE, M.S. YUN, AND P.M. BRYANT Owens Valley Radio Observatory Caltech 105-24, Pasadena, CA 91125

Abstract. Dense molecular clouds are critical to the activity occurring in galactic nuclei. They are the active component from which starbursts arise, and this dense interstellar gas may be the fuel for AGNs. In the ultraluminous IRAS galaxies, high resolution millimeter line mapping has shown extremely high gas surface densities in the central kpc, and often a significant fraction (> 25%) of the total molecular line emission from the galaxy arises from these central regions. New maps at resolutions down to 0.8" in the ultraluminous IRAS galaxies reveal kinematic gradients parallel to the major axis of the CO intensity distribution, suggesting that the gas is situated in a central rotating disk. The most extreme central concentrations are seen in Arp 220 and Mrk 231 (Sey 1) which have now been mapped in both the 2.6 and 1.3 mm CO transitions. In both galaxies, the high observed CO brightness temperatures indicate large area filling factors with mean H2 densities exceeding 104cm-3 • To produce the observed luminosities, the star formation rates must be ",100 M0 yr- 1 within the central 500 pc radius. Estimated time scales for both the dynamical evolution and the exhaustion of the observed central ISM are typically 2x108

years. At higher redshift, CO emission has been unambiguously detected in two objects, FSC 10214+4724 (z = 2.3) and H1413+117 (z = 2.5), both of which are probably gravitationally lensed. High resolution mapping of FSC 10214+4724 reveals two components: an unresolved core with 2/3 of the emission and an extended "disk" (9x24 kpc). In H1413+117 our CO(7-6) maps at 0.8" show a morphology similar to the cloverleaf pattern seen in the optical and the relative fluxes of the four components varies with velocity. Models for the CO lensing are consistent with the molecular emission arising from a disk within about 1 kpc in radius of the quasar. These high red shift systems may therefore be early universe counterparts of the ultraluminous IRAS galaxies.

25


26 N .Z. SCOVILLE ET AL.

1. INTRODUCTION

The IRAS survey provided a complete sample ofluminous starburst galaxies out to moderate cosmological distances (z < 0.1); however, high resolution observations of that gas are vital to understand the differences in starburst activity from one galaxy to another. Without such molecular line data, it is not known if the high luminosities are simply due to a large quantity of gas, a more concentrated distribution of molecular gas, or a more efficient trigger for stimulating gas to form stars. It now appears as though all three factors may be at work to varying degrees. Single dish molecular line surveys of samples of IRAS bright galaxies (eg, Sanders, Scoville & Soifer 1991, Tinney et al. 1988, Sage & Solomon 1988) clearly indicated that the more luminous infrared bright galaxies have greater CO luminosities and thus greater overall masses of dense molecular gas. On the other hand, at the highest luminosities (> 1011 L0 ), the ratio LIR/MH2 is enhanced by a factor of 5-50 over that measured in normal spiral galaxies like the Milky Way or M51. At high LIR there is also a higher frequency of interacting and merger systems and an increasing fraction of galaxies with the optical emission line ratios indicative of a relatively hard ionizing spectrum, possibly from an active galactic nucleus (see Sanders et al. 1990). The increasing luminosityto-mass ratio can be interpreted as a greater efficiency for forming stars in those galaxies or an additional non-stellar source of luminosity.

The peaking of quasar density at z ""' 2 as well as the stellar population analysis of galaxies in the local universe suggest that most galaxies must have formed near or prior to the redshift of 2. The identification of the IRAS FSC 10214+4724 with a z=2.3 galaxy (L > 3 X 1014L 0 , Rowan-Robinson et al. 1991) and the subsequent detection of CO emission (MH2 > 1011 M0 - Brown & Vanden Bout 1991, Solomon et al. 1992) have substantiated the hypothesized existence of luminous gas-rich galaxies in the early epochs forming stars at prodigious rates. Given that there has been reasonable heavy element production, CO rotational transitions are excellent probes of cold interstellar medium (ISM) in very distant galaxies. For objects of constant CO brightness temperature in the different rotational transitions, the observed line flux is almost independent of redshift at z = 1-6 due to the factor (1 + z)2 in angular size distance. These molecular transitions are also unambiguous tracers of cold, dense molecular gas where stars are forming.

Many of the most active starburst galaxies are subject to strong dynamical perturbations: they are either interacting galaxy systems or show evidence of a central bar-like mass distribution. In both instances there will be strong non-axisymmetric gravitational forces which disturb the circular rotation. The interstellar gas is fundamentally different from the stars - the

ULTRALUMINOUS AND HIGH Z GALAXIES 27

ISM is extremely dissipative (due to the fact that the gas clouds have much larger cross-section relative to their mean separation and in cloud collisions the gas radiates strongly in shock fronts). The dissipation of orbital energy (and transfer of angular momentum) leads to transfer of gas towards the galactic center, resulting in high central concentrations of active starforming molecular clouds. The studies of ultraluminous IRAS galaxies in the local universe are highly relevant to the search for and our understanding of primeval galaxies since both starbursts and galactic merging must have been prevalent in the early universe.

2. LUMINOUS IRAS GALAXIES

A major direction of extra galactic molecular line research has been measurement of the molecular gas properties and distributions in luminous IRAS galaxies. The IRAS bright galaxy survey (Soifer et al. 1987) included 238 objects in the northern hemisphere, and approximately 150 of these have had single dish CO measurements. The resulting H2 masses, assuming the same CO emissivity per unit mass for the Milky Way GMCs, show a clear trend for increasing masses of molecular gas with increasing far infrared luminosity. For the most luminous objects, the total H2 masses are in the range 2-50 X 109 Me;), that is 1-20 times the H2 content of our Galaxy. CO aperture synthesis mapping has been done on approximately 30 of these galaxies. Early aperture synthesis maps at 4-6" resolution showed that this gas is also much more concentrated in the galactic nuclei than in normal galaxies (see Scoville et al. 1991, Okumura et al. 1991). Typically 50% of the molecular gas was found within radii less than or equal to 1 kpc, and the gas mass fractions in the central regions (MH2! Mdyn) often exceed 50% (as compared with typical values of 5% near the centers of normal galaxies). Recently, several of these galaxies have been observed at 2" resolution in CO (1-0) and at I" resolution in the CO (2-1) transition. We describe three of these objects (VV 114, Mrk 231, and Arp 220) in more detail below.

2.1. VV 114

VV 114 is an early merger system with LJR = 6 X 1011 Le;) and a very large gas content, MH2 = 5 X 1010 Me;). The two galaxy nuclei are seperated by 6 kpc, and CO (1-0) images show the molecular peak between the optical and infrared nuclei (Yun, Scoville, & Knop 1994). The CO is shown superposed on an r-band image in Figure 1. The molecular gas is extended over"" 5 kpc between the two galaxy nuclei. In fact, the CO distribution is quite similar to that of the non-thermal radio continuum. In this system, the gas is clearly more concentrated toward a central peak than are the two stellar systems. This is interesting because more rapid merging is expected for the

28

[rl <n

~ «

-5

-10

-15

-20 20 10

N.Z. SCOVILLE ET AL.

o ARC SEC

-10 ·20

Figure 1. The integrated CO (1-0) emission (3" resolution) is shown superposed on an r-band image of the interacting system VV 114 (Yun, Scoville, & Knop 1994). The position of the eastern nucleus which is brightest at 2/Lm is indicated by the cross. The molecular gas is distributed in a bar-like configuration between the two galaxy nuclei.

gas than for the stars in interacting systems due to the greater dissipation in the gas.

2.2. MRK 231

Mrk 231 (at 174 Mpc) is one of the few ultraluminous IRAS galaxies in which there is general consensus on the existence of a dust-embedded high


luminosity Sey 1 AGN (Boksenberg et al. 1977). New CO (1-0) and (2-1) images at 0.8-2" resolution resolve the CO source and reveal a velocity gradient parallel to the major axis of the CO distribution (Bryant & Scoville 1995). Virtually all of the single dish CO emission is contained within a radius of 420 pc, and the CO brightness temperatures exceed 30K, indicating nearly complete area filling factors for the molecular gas within this radius. The CO conversion factor in such galactic nuclei was extensively analyzed (Bryant & Scoville 1995) and was found to be within a factor of a few of the Galactic value for Mrk 231. Assuming the Galactic value, the nuclear disk of Mrk 231 must be within 22° of face-on, in order to reconcile the dynamical mass with the mass from the CO line emissivity. This is consistent with the modest extinction measured for the Seyfert/quasar nucleus.

2.3. ARP 220

Arp 220 at 77 Mpc has been imaged in CO (1-0) (Scoville et al. 1991), HeN (Radford et al. 1991) and most recently in CO (2-1) (Scoville, Yun & Bryant 1995). The CO (2-1) maps at I" clearly reveal three components in the dense molecular gas: peaks corresponding to each of the double nuclei (separated by 0.9" at pa = 101°) seen in the near infrared (Graham et al. 1990) and radio continuum and a more extended disk like structure elongated southwest-northeast, similar to the dust lane seen in optical images. In Figure 2, the CO (2-1) line emission integrated over all velocities is shown superimposed on a Ha image of Arp 220. The elongated disk feature exhibits a monotonic velocity gradient parallel to the major axis of the CO intensity distribution, and the dynamical mass determined from the CO is approximately 3x 1010 M0 within 0.9" (360 pc) radius. Bright peaks are also seen in the channel maps coifitident with the locations of the two infrared nuclei. The observed kinematics of the molecular gas are consistent with the radial distribution of the CO emissivity in the sense that a rotation curve derived, assuming that the total mass is dominated by the molecular gas (with little contribution from the stars) is consistent with the observed CO line profiles (Scoville, Yun, & Bryant 1995). The CO to H2 conversion ratio which gives best correspondence between the CO kinematics and the dynamical mass distribution is 40% of the standard Galactic value (Scoville, Yun & Bryant 1995).

In Arp 220 the measured mm-wavelength continuum provides a strong constraint on starburst models since any free-free contribution from the HII associated with a starburst must be less than the measured continuum minus the expected dust contribution. The limit on the free-free continuum implies Q::; 7.5 X 10548-1 . For a standard Miller-Scalo initial mass function


with lower and upper stellar mass cutoffs (ml and m u ), a constant rate of star formation (M) and a burst timescale (tB), the ratio of luminosity to Lyman continuum photon production rate (Q) is

( m )-0.37 (m )-0.96 ( t )0.37 L/Q = 7.2 X 10-44 -Ml __ u_ B 7 L8 sec. 5 8 45M8 5 X 10 yr

Combining this constraint with the dynamical mass limit derived from the CO kinematics, Scoville, Yun & Bryant (1995) find that ml = 5 M 8 ,

mu = 23 M 8 , and M = 90M8 yr- 1 (with a burst timescale of 5 X 107 yr). The narrow range of permissable stellar masses suggests that a starburst origin for a major fraction of the luminosity in Arp 220 is unlikely, although not ruled out. Starbursts with low mass stars (ml = 0.lM8 ) are categorically ruled out since they violate the L / M* constraint by more than an order of magnitude; similarly starbursts with only high mass OB stars are easily ruled out since they violate the L/Q constraint.

3. HIGH RED SHIFT GALAXIES

The Owens Valley Millimeter Array is very well-suited for detecting gasrich high redshift systems. The six 10.4-m diameter telescopes in the array provide an effective collecting area of 510 m 2 - comparable to a 25.5-m diameter telescope. Cross-correlation of the signals as is done in interferometry also significantly reduces the instrumental effects that often plague the single dish observations. The OVRO observations also yield spectra.! map cubes with typical spatial resolution of 2-5", providing spatial information and, when resolved, kinematic information.

:U. FSC 10214+4724

The CO (3-2) emission region the in z = 2.3 IRAS galaxy FSC 10214+4724 is resolved by our 2" resolution observations (see Figure ~{). About 1/3 of the total CO flux originates from an extended structure with a narrow line width while the majority of molecular gas belongs to an unresolved structure (Scoville et al. 1995). The small line width for the extended component may be explained if a kinematically distinct section of the galaxy is strongly amplified and the angular extent is increased by gravitational lensing. Since the molecular and far infrared emission probably arise from the same extended gas and dust distribution in the host galaxy, it is likely both emissions have undergone similar magnifications. If this is the case, then the true luminosity to molecular mass ratio is similar to the apparent ratio (~ 103 L8 MG'l)- this is approximately 200 times greater than that in the Galaxy! At such high luminosity-to-mass ratios, radiation pressure


A f1C SOC

Figure 2. CO (2-1) line emission (I" resolution) integrated over all velocities in Arp 220 is shown superimposed on an HO' image (Scoville, Yun, &. Bryant 1995). The CO emission in the central 500 pc of the galaxy peaks between the two infrared nuclei (spacing 0.95") and is elongated in the outer contours parallel to the dust lane crossing the HO' emission distribution.

acting on the dust may provide significant support against gravity, thus alleviating for the problem that the dynamical mass is apparently less than the molecular gas mass (without resorting to a nearly face-on inclination).

3.2. H1413+117 (THE CLOVERLEAF)

The CO (7-6) emission from HI413+117 (the Cloverleaf quasar) is also resolved by our 0.8" beam (Yun et al. 1995). The morphology of the integrated CO emission (shown in Figure 4) is remarkably similar to that of


·136 kmh -4S kmh

·2

~

·0 + D

Ol 0

+4S kmh +136 kmI.

g 0

~ 0

. ~ 2

U 0

~ -2

.. (0"'), tl

2 0 -2 -4 -(; ARC SEC

Figure 3. New 2" resolution CO (3-2) maps of FSC 10214+4724 with 90 km S-l

velocity channels (Scoville et al. 1995). The cross marks the VLA radio source position. An extended component with a narrow line width is found in V=+45 km/s channel.

optical light, suggesting the CO emitting region is very compact and arises close to the quasar. In CO, the relative brightnesses of the four images seen in the optical varies as a function of velocity and thus the CO enables one to "map" the molecules within the host galaxy, given a suitable model for the lensing. The model presented by Yun et al. (1995) places the molecular gas in a disk within 1 kpc of the quasar nucleus.

The inferred total H2 mass, 2.3 X 1011 Ag -1 h- 2 Me (Ag is amplification factor), is nearly identical to the CO (3-2) measurement by Barvainis et af. (1994), and the CO (3-2) emission must arise from high excitation gas (n> 105 cm-3 , T > 50 K).

The dust masses estimated from the continuum emission are about 2 X

108 and 3 X 108Ag -1h- 2 M e for FSC 10214+4724 and H1413+117 (see Downes et al. 1992, Rowan-Robinson et al. 1993, Barvainis et al. 1992). The inferred gas-to-dust ratios are therefore 500 and 800, respectively - similar to the IR luminous systems in the local universe (e.g. Sanders, Scoville, & Soifer 1991). The dust continuum and CO emission are likely co-spatial in distribution and thus similarly amplified by lensing.


Figure 4. CO (7-6) map of H1413+117 at 0.8" resolution in contours overlaid on the grey scale HST F555W image (Falco 1993).

3.3. SUMMARY OF OBSERVATIONS

The results of the OVRO CO observations on other high redshift systems are summarized in Table 1. In the majority of cases, the H2 mass sensitivity is sufficient to detect the more gas-rich galaxies in the local universe. In all cases, any FSC 10214+4724 like objects should have been detected. To improve the detection statistics, more IR-selected candidates need to be searched. Detecting the progenitors of L* galaxies probably requires an order of magnitude improvement in sensitivity. We note that neither 4C 41.17 (a z=3.8 radio galaxy with dust continuum detection - Chini & Kriigel 1994) nor B2 0902+34 (a z=3.4 "protogalaxy" - Eales et al. 1993) were detected in CO. The upper limit on the gas-to-dust ratio is 750 for 4C 41.17. We find the 650jLm (rest) continuum emission from B2 0902+34 dominated by non-thermal emission rather than by dust emission.

4. SUMMARY

Important general characteristics of the molecular gas in starburst and high infrared luminosity galactic nuclei include: (1) extremely high gas surface


TABLE 1. Summary of the OVRO CO Observations

Name z Transition SeQ ll.V MH2

(mJy) (km S-1) (h- 2 ]010 M 8 )

H]413+117 2.56 CO (7-6) ]20 ± ]2 340 23 ± 2 FSCI0214+4724 2.29 CO (3-2) 14 ± 2a 250 ]2 ± 2 3C 48 0.37 CO (1-0) 7.7 ± 1.9 b 250 3.9 ± 0.8 B2 0902+34 3.40 CO (4-3) < 4.1e (300)d < 3.3 d

4C 41.17 3.80 CO (4-3) < 4.5 (300) < 4.2 3C 368 1.13 CO (2-1) < 8.1 (300) < 4.5 TX ]243+036 3.58 CO (4-3) < 6.9 (300) < 5.5 PC 1033+4750 3.68 CO (4-3) < 6.1 (300) < 5.5 MG 10]9+0535 2.77 CO (3-2) < 6.2 (300) < 6.4 TX 0211-122 2.34 CO (3-2) < 9.] (300) < 7.3 4C 28.58 2.95 CO (3-2) < 7.4 (300) < 8.5 H0836+113 2.47 CO (3-2) < 13 (300) <11 PC 1643+463]A 3.14 CO (3-2) < ]0 (300) < 13 PC 1406+123 2.25 CO (3-2) < 12 (300) < 13

Notes: (a) Scoville et al. (]995); (b) Scoville et al. (1993); (c) all 30" upper limits; ll. V =300 km S-1 is assumed.

(d)

densities exceeding 103 M0 pc-2 ; (2) strong non-circular gas motion often accreting radially along a stellar bar potential or two merging galactic nuclei; (3) high molecular gas mass fractions in the central 500 pc with the gas constituting greater than 50% of the total mass.

The strong non-circular gas motions and high central gas concentrations may be driven by non-axisymmetric gravitational forces from stellar bars or the merging of comparable mass galactic nuclei. Large nuclear gas concentrations can then arise and set off the starburst activity which becomes non-linear due to cloud-cloud collisions and stimulated star formation effects such as high pressure HII regions and supernova blast waves.

The statistics from the IRAS survey indicate that approximately 0.2% of the present epoch spiral galaxies are undergoing high luminosity global starbursts; however, given the short duration of this activity (2 X 108 years) and the fact that standard cosmological evolution would indicate that the interaction rate was higher in the past, nearly all galaxies would be undergoing such merger induced starbursts at z=1-2 (see Scoville & Soifer 1990). Although only a small fraction of present day galaxies are highly luminous in the infrared, the star formation rates are typically enhanced by factors of 10-100, thus a significant fraction of the overall star formation may occur during such brief periods of activity.

ULTRALUMINOUS AND HIGH Z GALAXIES

Our research is supported in part by NSF Grant AST 93-14079.

References

Antonucci, J., & Miller, J.S. 1985, Ap.J., 297, 621 Barvainis et al. 1994, Nature, 371. 586 Barvainis, R., Antonucci, R., Coleman, P. 1992, ApJ, 399, L19

35

Boksenberg, A., Carswell, R.F., Allen, D.A. Fosbury, R.A.E., Penston, M.V., & Sargent, W.L.W. 1977, M.N.R.A.S., 178, 451

Brown, R.L., Vanden Bout, P.A. 1991, AJ, 102, 1956 Bryant, P.M. & Scoville, N.Z. 1995, Ap.J., in press Chini, R., Kriigel, E. 1994, A&A, 288, L33 Downes et al. 1992, ApJ, 398, L25 Eales et al. 1993, Nature, 363, 140 Falco E.E. 1993, in Gravitational Lenses in the Universe, eds. Surdej et al. (Universite

de Liege, Liege), p.127 Graham, J.R., Carico, D.P., Matthews, K., Neugebauer, G., Soifer, B.T., & Wilson, T.D.

1990, Ap.J. Letters, 354, L5 Okumura, S.K., Kawabe,R., lshiguro,M. Kasuga, T., & Morit.a, K.I. 1991 in lAU Sym

posium No. 146, 425 Planesas, P., Scoville, N.Z. & Myers, S.T. 1991, Ap.J., 369, 364 Radford, S.J.E., Delannoy, J., Downes, D. Guelin. M. & Guillot.eau, S. 1991, III lAU

Symposium, 146, 303 Rieke, G.R., Loken, K., Rieke, M.J. & Tamblyn, P. 1993, Ap.J., 412, 99 Rowan-Robinson et al. 1991, Nature, 351, 719 Rowan-Robinson et al. 1993, MNRAS, 261, 513 Sage, L., & Solomon, P.M. 1989. Ap.J., 334, 613 Sanders D.B., Scoville, N.Z., & Soifer, B.T. 1991, Ap.J., 370,158 Sanders, D.B., Soifer, B.T., Elias, J.R., Madore, B.F., Matthews, K., Neugebauer, G., &

Scoville, N.Z. 1988, Ap.J., 325,74 Scoville, N.Z., Matthews, K., Carico, D.P. & Sanders, D.B. 1988, Ap.J., 327, L61 Scoville, N.Z., Sargent, A.I., Sanders, D.B., & Soifer, B.T. 1991, Ap.J., 366, L5 Scoville, N.Z. & Soifer, B.T. 1990, in Massive Stars in Starbursts, ed. C. Leitherer (Cam-

bridge: Cambridge Univ. Press), 233 Scoville, N.Z., Yun, M.S., Brown, R., Vanden Bout, P. 1995, ApJ, in press Scoville, N.Z., Yun, M.S. & Bryant, P.M. 1995, Ap.J., (in preparation) Shen, J & Lo, K.Y. 1995, Ap. J. Letters, 445, L99 Soifer, B.T., & Wilson, T.D. 1990, Ap. J. Letters, 354, L5 Solomon, P.M., Downes, D., Radford, S.J.E. 1992, ApJ, 398, L29 Tinney, C.G., Scoville, N.Z., Sanders, D.B. & Soifer, B.T. Ap.J., 362, 473 Yun, M.S., Scoville, N.Z. & Knop, R.A. 1994, Ap. J. Letters, 430, Ll09 Yun, M.S., Scoville, N.Z., Carasco, J.J. & Blandford, R.D. 1995, Nature, in preparation

ULTRALUMINOUS INFRARED GALAXIES: DISSIPATION IN FORMING SPHEROIDAL SYSTEMS

PAUL P. VAN DER WERF Leiden Observatory P.O. Box 951:1 NL-2:100 RA Leiden The Netherlands

1. Introduction

One of the most important results from the IRAS mission was the discovery of the ultraluminous infrared galaxies (ULIRGs), defined by 8 - 1000/lm luminosities Lm > 1012 L(~h i.e., having luminosities in the range of low redshift QSOs, but outnumbering these by a factor of about three (Sanders et aZ. 1988). They are therefore the dominant population in the local universe at the highest luminosities. These large luminosities are usually attributed to an intense starburst (Rieke et ai. 1985), a dust-embedded QSO (Sanders et aZ. 1988) or a combination of these. An important clue to the nature of these objects comes from optical imaging (Sanders et al. 1988), showing that most or all of the ULIRGs in the IRAS Bright Galaxy Sample are advanced major mergers. Interferometric imaging of CO emission at arcsecond resolution shows that most of the molecular gas is concentrated in the central kiloparsec (e.g., Scoville et al. 1991). As shown by the simulations by Barnes & Hernquist (1991), the presence of such concentrations is naturally expected in the merging of gas-rich galaxies, where the gas, being dissipative, quickly sheds its angular momentum and sinks to the bottom of the potential well.

This situation has important consequences for the formation of elliptical galaxies by merging of disk galaxies, as proposed by Toomre (1977). Observational support for this scenario has been found in the approximately r 1 / 4

surface brightness law found in some advanced mergers (e.g., N GC 7252, Schweizer 1982), which results from violent relaxation during the merging process (Van Albada 1982). However, it is well known that space and phasespace densities in the centres of elliptical galaxies are much higher than in

37

M. N. Bremer et al. (eds.), Cold Gas at High RedshiJt, 37-45. © 1996 Kluwer Academic Publishers.

38 PAUL P. VAN DER WERF

disks. Hence, to form realistic ellipticals from merging disks, dissipative collapse is required. This dissipation can only take place in the gas component, resulting in the central gas concentrations observed in ULIRGs. This idea has been applied by Kormendy & Sanders (1992) who show that the central H2 mass densities in ULIRGs are of the order of 102 M8 pc-3 (assuming the same L(CO)j MH2 conversion factor as in the Galaxy), comparable to central stellar mass densities in ellipticals. Doyon et aZ. (1994) have combined stellar nuclear velocity dispersions with measured effective radii to place the luminous IRAS galaxies NGC 6240 and Arp 220 on the cooling diagram, where they are found to be located in the region occupied by elliptical galaxies. A similar result is obtained for the central surface brightness vs. core radius diagram. Since these diagrams are projections of the fundamental plane of elliptical galaxies (Kormendy & Djorgovsky 1989), it is concluded that the starburst populations in the nuclei of NGC 6240 and Arp 220 have the fundamental structural and dynamical properties of the centres of ellipticals. This result confirms the scenario in which the ULIRG phase is a crucial stage in the transformation of merging gas-rich galaxies into an elliptical. Dissipation in the gaseous component plays in this scenario a central role.

In this paper the role of dissipation in the luminous IRAS galaxies NGC 6240 and Arp 220 is studied in more detail. A general feature of ULIRGs is the presence of pronounced H2 near-infrared (NIR) vibrational line emission (Goldader et aZ. 1995). It will be shown that this emission results from shocks in the central molecular component. The essential feature of shocks is that they dissipate mechanical energy. The luminosity of the H2 v = 1-+0 S( 1) line is shown to directly measure the dissipation rate (in M8 yr-1), and the implications of this result are explored. Finally, the possibility of dust-enshrouded, forming ellipticals at high redshift is examined, and it is argued that their properties should be similar to those of low-z ULIRGs. Throughout this paper the value Ho = 75 km s-1 Mpc-1 is adopted.

2. H2 emission in NGC 6240: dissipation in shocks

Although its luminosity LIR ~ 6 X 1011 M(') places it outside the class of ULIRGs, NGC 6240 possesses all of the characteristics of its somewhat more luminous cousins. It shows a strongly disturbed morphology, with long tidal tails (e.g., Fosbury & Wall 1979) and two nuclei with a projected separation of only 1~'8 (900 pc at a distance of 100 Mpc), all indicative of an advanced merger. CO observations by Wang et oJ. (1991) reveal a pronounced concentration of molecular gas in the nuclear region. The H2 NIR emission lines in NGC 6240 are unusually bright, making this object an ideal target

DISSIPATION IN ULTRALUMINOUS INFRARED GALAXIES 39

for studying the role of the H2 NIR emission with spectral line imaging and multi-line spectroscopy.

An image of the H2 v = 1-+0 S(l) emission of NGC 6240 has been presented by Van der Werf et al. (1993). The most striking feature of the H2 distribution is the pronounced peak in the line emission between the two remnant nuclei. The lack of correspondence to the underlying stellar population strongly argues against any excitation mechanism directly or indirectly related to stars (e.g., fluorescence after ultraviolet (UV) pumping, excitation by shocks or X-rays from supernova remnants etc.). As shown in Van der Werf et al. (1993) the data point to slow shocks in dense molecular gas as the origin of the H2 v = 1-+0 S( 1) emission. The fact that the H2 emission peaks between the nuclei is significant since the gas is expected to merge more rapidly than the stars, and therefore points to the crucial role of dissipation. This phenomenon is also seen in other gas-rich merging systems (Scoville et al. 1996).

Because of the unusually high luminosity of the H2 v = 1-+0 S( 1) line in N GC 6240, several alternatives have been proposed to account for this emission. These mechanisms include X-ray excitation (Draine & Woods 1990), fluorescence after UV pumping (Tanaka et al. 1991) and H2 formation heating (Mouri & Taniguchi 1995). These proposals are based on ratios of the fainter H2 lines, most of which have only been measured at low SIN ratio. In order to improve this situation, the NIR spectrograph IRSPEC (Moorwood et at. 1986, 1991) on the New Technology Telescope (NTT) of the European Southern Observatory at La Silla, Chile, has been used to measure a large number of H2 and other lines in the NIR Hand K-bands. Resulting spectra, obtained with a slit width of 4~14, are shown in Fig. 1 and derived parameters are listed in Table 1. The flux of the H2 v = 1-+0 Q(3) line is strongly affected by telluric absorption and is less secure than that of the other lines. The H2 v = 1-+0 S(7) and S(9) lines are first detections in an extragalactic object.

From these results the following conclusions are drawn:

1. The ratios of the S(l) and S(3) lines are very similar in the v = 1-+0 band and the v = 2-+ 1 band. We thus do not confirm the faintness of H2 v = 2-+1 S(3) reported by Lester et al. (1988), and subsequently used by Draine & Woods (1990) to argue in favour of their X-ray excitation model, where this line is suppressed by selective depopulation of the v = 2, J = 5 level by resonantly scattered hydrogen Ly 0' photons.

2. We do not confirm the detection of the very high lines H2 v = 6-+4 Q(l) and v = 5-+3 0(3) reported by Elston & Maloney (1990), thus removing the argument used by Tanaka et 01. (1991) to argue for UV pumping. While the line detected at 1.64 JLm in Fig. 1 is close to the redshifted position of H2 v = 6-+4 Q(l), the non-detection of the


10-10

NGC6240

..-. i Bx 10-11

E :t

N I

56Xl0-11

[FeU] Bry

n -I III

~ JJ. CIl

,A '--' 4x 10-11

~ A >. tJt. .... ·iii r:: CIl '0

I U I I I I >< 2x 10-11 ;:l

~ Sil CO H2

0

1.6 1.8 2 2.2 Wavelength [/Lm]

Figure 1. Spectra of N GC 6240 in selected parts of the near-infrared Hand K bands, integrated over a 4'.'4 region centred on the position midway between the two nuclei.

v = 5---+3 0(3) line, which in a fluorescent spectrum should be brighter than v = 6---+4 Q( 1) (Black & Van Dishoeck 1987) argues against this identification. Since furthermore the shape of the 1.64 11m line matches very well that of the [Fe I1Jline detected at 1.68 11m, we identify the 1.64 11m line as the [Fe I1J a 4 D3 / 2 ---+ a 4 F9 / 2 line.

3. Since the H2 lines are optically thin, the observed H2 fluxes can be converted directly into upper level column densities N (v, J). The excitation temperature Tex can then be derived using the relation N(v, J) ex g( v, J) exp( -T( v, J))/Tex, where g( v, J) is the statistical weight and T( v, J) the upper level temperature corresponding to quantum numbers v and J. This analysis results in a rotational temperature Trot = 2350 ± 200 K within the v = 1 level and a vibrational temperature Tvib = 1900 ± 350 K between v = 1 and 2. Thus thermal emission at an excitation temperature of about 2000 K accounts for aU of the observed H2 line strengths, as expected in a shock-heated medium.

It is therefore concluded that shocks alone are sufficient to account for


TABLE 1. Near-infrared line fluxes in NGC6240

Line Aobs [/lm] Flux [ergs-1 cm-2 ]

[Fen] a 4 D 7 / 2 -+ a 6 D 9 / 2 1.28735 1.02 x 10-13

[Fe n] a 4 D 3 / 2 -+ a 4 F9 / 2 1.64068 1.15 x 10-14

[Feu] a 4 D 7 / 2 -+ a 4 F9 / 2 1.68409 7.05 x 10-14

H2 V = 1--0 S(9) 1. 72969 7.12 x 10-15

H2 V = 1-+0 S(7) 1.79043 6.11 x 10-14

H2 V = 1-+0 S(3) 2.00506 2.05 x 10-13

H2 V = 2-+1 S(3) 2.12408 1.39 x 10-14

H2 V = 1-+0 S(I) 2.17365 2.04 x 10-13

Br")' 2.21763 5.27 x 10-15

H2 V = 2-+1 S(I) 2.30280 1.36 x 10-14

H2 V = 1-+0 Q(I) 2.46521 1.17 x 10-13

H2 V = 1-+0 Q(3) 2.48267 1.06 x 10-13

the H2 vibrational emission from NGC 6240, and that there is no need to invoke the presence of other mechanisms. Hence, the H2 emission lines can be used as a measure of the dissipation rate in the starbursting nuclear interstellar medium, as detailed in Sect. 3.

3. Dissipation in the nuclear interstellar medium in Arp 220

At a distance of 77 Mpc, Arp 220 is the closest ULIRG, and has become the prototype of its class. While, at a luminosity LJR ~ 1.5 X 1012 L8 , emitting 2.5 times more energy than NGC 6240, it resembles this galaxy closely in its general features. Strong tidal tails (Joseph & Wright 1985) and the presence of two nuclei with a projected separation of only 0~/9 or 170 pc (Graham et al. 1990) show that it is an advanced merger system. Furthermore, a 720 pc diameter region containing 1.8 X 1010 M8 of molecular gas is found in the nuclear region (Scoville et al. 1991; Scoville et al. 1996). Like NGC 6240, Arp 220 is a bright emitter in the H2 vibrational lines. An image in this line is presented in Fig. 2, overlaid on the Hubble Space Telescope WFPCl F555W image by (Shaya et al. 1994). The H2 emission, discussed in more detail by Van der Werf & Israel (1996a) consists of two components: an extended component of faint emission following approximately the dust lane of Arp 220 and probably due to slow shocks or extended star formation in this region, and a compact component producing most of the H2 line emission. This compact component has a diameter of about 2~/5 or 470 pc and produces a luminosity LH2 ~ 107 L8 in the H2 v = 1---+0 S(I) line. Assuming an excitation temperature of 2000 K, as was found for NGC 6240


in Sect. 2, the mass of hot emitting H2 is Mhot H2 '" 3.3 X 104 Mev (cf., Van der Werf & Israel 1996a for details). The shock models by Draine et a1. (1983) show that the cooling time for these hot H2 layers is about 1000yrs. Hence, to sustain the observed H2 v = 1--...0 S(l) brightness the molecular gas must be shocked at a rate of about 1I1H2 '" 33 Mev yr- 1 . Vife note that with a total molecular mass of the central component of 1.8 x 1010 Mev (Scoville et aZ. 1991), and a typical starburst and merger timescale of 2 X 108 yrs, this implies that a large fraction of the central component will be shocked and thus undergo dissipation during the evolution of the system. This underlines the crucial role of dissipation in the centres of ULIRGs in producing the central space and phase space densities observed in ellipticals.

The more extensive analysis by Van der Werf & Israel (1996a) shows that the central gas concentration in Arp 220 can be viewed as a large collection of dense clumps, which collide frequently (thereby giving rise to the observed H2 vibrational emission), thus dissipating kinetic energy and losing angular momentum. The quantity 1I1H2 thus represents a dissipation rate in terms of mass in the nuclear gas concentration in Arp 220, that is, the mass of molecular material losing its angular momentum and subsequently falling to the centre of the potential well, per unit of time.

It is important to note that the size of the starburst region, as implied by the data presented by Larkin et aZ. (1995) and Condon et aZ. (1991) is much smaller than that of the dense molecular component (a situation also found in NGC 6240, see Van der Werf et aZ. 1993). Therefore, in order to feed the starburst, gas must be transferred inward. The dissipation rate derived above measures the transfer of gas to the starburst region, and can thus be equated to a star formation rate 111*. The implied 111* '" 33 Mev yr- 1 is remarkably close to the independently derived star forming rate of 30 - 100 Mev yr-1 based on a starburst model (Van der Werf & Israel 1996a), confirming the validity of our interpretation. If generally valid in ULIRGs, our model predicts a linear relation between LH2

and LJR in ULIRGs. This is indeed observed (Goldader et aZ. 1995). Our model of starbursting dissipative collapse not only explains this correlation, but also predicts the correct LH2 / LIR ratio.

4. High redshift ULIRGs as forming spheroids

We have argued that at low redshift the formation of elliptical galaxies by major mergers involves starbursting dissipative collapse, during which most ofthe luminosity associated with the starburst is absorbed by dust and reradiated in the far-infrared. During this phase the system is a ULIRG. The question arises whether the formation of spheroids at high redshift


.. 15

,...., 0 0 0 ~ ....... I.l G.l .. Q 10

15hs4m57.5- 57.0-R.A. [J2oo0]

Figure 2. H2 v = 1-+0 S(l) emission of Arp 220, shown ill contours at levels of 1.2, 2.4, 3.6, 4.8, 6.0, 7.5, 9.0., 10.5, 12.0 and 12.75 x 10-5 erg S-1 cm-2 sr-1 , overlaid on an F555W HST WFPC image by Shaya et al. (1994). Two crosses denote the positions of the to radio/near-infrared nuclei (from Van der Werf & Israel 1996a).

proceeds in a similar fashion. The high central densities and phase densities in speroids require dissipative collapse (e.g., Tremaine 1981; Kormendy 1989), whether a merger is involved or not. The dissipative collapse in ULIRGs is accompanied by a major starburst, which is shrouded in dust most of the time. It is, therefore, likely that the initial starburst in high-z forming ellipticals is similarly affected. The first stars must have formed in a dust-free environment, but a starburst can produce solar metallicities in about 108 yrs (Matteucci & Padovani 1993), so that the initial starburst in a forming spheroid will go through an ultraluminous, dust-enshrouded phase even if it started at low metallicity. The luminosity of an initial starburst


can be estimated from the metal abundance that needs to be produced:

(1)

where £ is the efficiency of energy generation by stellar nucleosynthesis and .6.X is the fraction of hydrogen nuclei converted into metals. It is seen that starformation rates and luminosities comparable to those in local ULIRGs are expected in the inital starburst, underlining the similarity of observational properties of these objects.

These arguments show that ULlRGs can be viewed as local analogs of forming spheroids. The detection of large amounts of dust and molecular gas in IRAS F10214+4724 and the Cloverleaf quasar, shows that galaxies with amounts of gas and dust exceeding those of local ULlRGs (even after gravitational lensing has been taken into account) do exist at high redshift, and have properties very similar to local ULlRGs (Scoville et aZ. 1996). However, to detect the CO emission, with current instrumentation still the amplification by a foreground gravitational lens is required. Unlensed objects with the same amounts of gas and dust will be detectable by future large millimeter wave arrays (Van der Werf & Israel 1996b).

The possibility of an obscured initial starburst may account for the lack of success in searches for high-z forming galaxies using deep optical imaging and searches aimed at detecting Lya emission (e.g., Thompson et aZ. 1995; Thompson & Djorgovski 1995). Such objects may be found in red shifted Ha, which is shifted into the near-IR K-band for redshifts from 2.1 to 2.6. Searches for Ha emission in the K-band are already underway and will obviously profit tremendously from the new generation of 8 m class telescopes equipped with large-format near-IR array cameras. Finally, one of the most exciting prospects is the possibility of finding high-red shift starburst galaxies through their redshifted dust emission. The order-of-magnitude increase of sensitivity that will be provided by the SCUBA instrument on the James Clerk Maxwell Telescope may enable the decisive breakthrough in searches for dusty high redshift starburst galaxies.

Acknowledgements. The research of Van der Werf has been made possible by a fellowship of the Royal Netherlands Academy of Arts and Sciences.

References

Barnes, J.E., & Hernquist, L.E. 1991, ApJ, 370, L65 Black, J.H., & Van Dishoeck, E.F. 1987, ApJ, 322, 412 Condon, J.J., Huang, Z.P., Yin, Q.F., & TIman, T.X. 1991, ApJ, 378, 65 Doyon, R., Wells, M., Wright, G.S., Joseph, R.D., Nadeau, D., & James, P.A. 1994, ApJ,

437, L23


Draine, B.T., & Woods, D.T. 1990, ApJ, 363, 464 Draine, B.T., Roberge, W.G., & Dalgarno, A. 1983, ApJ, 264, 485 Elston, R., & Maloney, P. 1990, ApJ, 357, 91 Fosbury, R.A.E., & Wall, J.V. 1979, MNRAS, 189, 79 Goldader, J.D., Joseph, R.D., Doyon, R., & Sanders, D.B. 1995, ApJ, 444, 97 Graham, J.R., Carico, D.P., Matthews, K., Neugebauer, G., Soifer, B.T., & Wilson, T.D.

1990, ApJ, 354, L5 Joseph, R.D., & Wright, G.S. 1985, MNRAS, 214, 87 Kormendy, J., & Djorgovsky, S. 1989, ARA&A, 27, 235 Kormendy, J., & Sanders, D.B. 1992, ApJ, 390, L53 Kormendy, J. 1989, ApJ, 342, L63 Larkin, J.E., Armus, L., Knop, R.A., Matthews, K., & Soifer, B.T. 1995, ApJ, 452, 599 Lester, D.F., Harvey, P.M., & Carr, J. 1988, ApJ, 329, 641 Matteucci, F., & Padovani, P. 1993, ApJ, 419, 485 Moorwood, A., Biereichel, P., Finger, G., Lizon, J.L., Meyer, M., Nees, W., & Paureau,

J. 1986, The Messenger, 44, 19 Moorwood, A., Moneti, A., & Gredel, R. 1991, The Messenger, 63, 77 Mouri, H., & Taniguchi, Y. 1995, ApJ, 449,134 Rieke, G.H., Cutri, R.M., Black, J.H., Kailey, W.F., McAlary, C.W., Lebofski, M.J., &

Elston, R. 1985, ApJ, 290, 116 Sanders, D.B., Soifer, B.T., Elias, J.H., Madore, B.F., Matthews, K., Neugebauer, G., &

Scoville, N .Z. 1988, ApJ, 325, 74 Schweizer, F. 1982, ApJ, 252, 455 Scoville, N.Z., Sargent, A.I., Sanders, D.B., & Soifer, B.T. 1991, ApJ, 366, L5 Scoville, N.Z., Yun, M.S., & Bryant, P.M. 1996, these proceedings Shaya, E.J., Dowling, D.M., Currie, D.G., Faber, S.M., & Groth, E.J. 1994, AJ, 107,

1675 Tanaka, M., Hasegawa, T., & Gatley, I. 1991, ApJ, 374, 516 Thompson, D., & Djorgovski, S.G. 1995, AJ, 110, 982 Thompson, D., Djorgovski, S., & Trauger, J. 1995, AJ, 110, 963 Toomre, A., 1977, Mergers and some consequences. In: Tinsley, B.M., & Larson, R.B.

(eds.), The evolution of galaxies and stellar populations, Yale University Observatory, New Haven, p. 401

Tremaine, S., 1981, Galaxy mergers. In: Fall, S.M., & Lynden-Bell, D. (eds.), The structure and evolution of normal galaxies, Cambridge University Press, Cambridge, p. p. 67

Van Albada, T.S. 1982, MNRAS, 201, 939 Van der Werf, P.P., & Israel, F.P., 1996a, in preparation Van der Werf, P.P., & Israel, F.P., 1996b, in Science with large millimetre arrays, ed. P.

A. Shaver, Springer, in preparation Van der Werf, P.P., Genzel, R., Krabbe, A., Blietz, M., Lutz, D., Drapatz, S., Ward,

M.J., & Forbes, D.A. 1993, ApJ, 405,522 Wang, Z., Scoville, N.Z., & Sanders, D.B. 1991, ApJ, 368, 112

THE NEUTRAL HYDROGEN DISTRIBUTION IN LUMINOUS INFRARED GALAXIES

J. E. HIBBARD Institute for Astronomy 2680 Woodlawn Dr., Honolulu, Hawai'i 96822

AND

M. S. YUN California Institute of Technology MS 105-24, Pasadena, California 91125

1. Introduction

Although there is strong evidence that the luminous (LJR > 3 X 1011 L 0 ;

H o=75) and ultraluminous (LJR > 1012 L 0 ) infrared galaxies result from the merger of two spirals (Sanders et al. 1988, Scoville et al. 1991), it is somewhat surprising that the optical morphology of these systems differs from the classical double-tailed morphology of merging disks described by Toomre & Toomre (1972) and illustrated by the sequence of 11 disk-disk systems in progressively advanced stages of merging known as the "Toomre Sequence" (Toomre 1977)1. In particular, the luminous IR galaxies as a class tend to exhibit shorter and/or more poorly defined tidal tails than the systems from Toomre's Sequence (Hibbard 1995). Similarly, optically selected interacting samples (including the Toomre Sequence) often have IR properties that are closer to those of isolated spirals than the ultraluminous IR galaxies (see Table 1).

These differences may simply reflect the short lived nature of the luminous IR phase, the rarity of long tails in disk-disk encounters, and/or a special viewing perspective for the Toomre Sequence objects. However they may also indicate necessary prerequisites for triggering luminous infrared activity in mergers, such as specific orbital geometries or Hubble

1 This sequence includes the best studied and strongest cases for disk-disk mergers, such as "The Antennae" (N4038/9), "The Mice" (N4676), and "The Atoms for Peace Galaxy" (N7252).

47

M. N. Bremer et al. (eds.). Cold Gas at High Redshift. 47-53. © 1996 Kluwer Academic Publishers.

48 J. E. HIBBARD AND M. S. YUN

TABLE 1. Comparison between IR and Optically Selected Mergers

Property

Sample Size Average Redshfit

< LIR > < LIR/L B > < MH2 > < MH2/LB > < L1R/M H2 > Avg. No. of Tails

A verage tail length: - Optical - HI

IRAS BGS

log(LIR/ L0 »12

10 15740±5770km S-1

(2±I)xl012 L0 45±21

(3±2)xl010 M0 0.9±0.6 M0Lr;} 78±53 L0Mr;;t 0.7±0.7

32±6 kpc

IRAS BGS

log(LIR/ L8 »11.5 Distance<100 Mpc

8

5520±1250km 8-1

(6±4)xl011 L8 25±24

(2±I)xl010 M0 0.6±0.5 M0Lr;} 48±21 L0Mr;;t 0.8±0.7

30±3 kpc 63±47 kpc

Toomre Sequence of Ongoing Mergers in the NGC

10 4760±2170km S-I

(2±I)xl011 L0 5±5 (1±l)xl0Io M0 0.4±0.2 M0Lr;/ 18±16 L8Mr;;t 1.7±0.4

53±23 kpc

88±47 kpc

types for the progenitors (Hibbard 1995). In an effort to clarify this matter, we have mapped the neutral hydrogen distribution in a sample of seven nearby ongoing mergers which are luminous in the infrared. We compare these observations with similar data on the less IR luminous mergers of the Toomre Sequence (Hibbard & van Gorkom AJ, in press; Mahoney, van der Hulst & Burke, in preparation; English, in preparation).

2. Observations and Results

The Infrared selected systems were chosen from the IRAS bright galaxy sample (BGS) list presented by Sanders et ai. (1991). We observed 8 systems within 150 Mpc with LJR > 3 X 1011 L0 and LJR/LB >10. The neutral hydrogen data were collected in 1994 and 1995 using the Very Large Array in its spectral line mapping mode. Each galaxy was observed in the C- and D-array configurations. Final sensitivities were a few 1019cm -2 at resolutions of ""20", with a velocity spacing of "" 11 km s-l. All but one of the systems (Mkn 231) were detected in emission. The results are presented in Figs. 1-3, and described briefly in Sect. 2.1.

There are a number of similarities between the tidal morphologies of the IR and optically selected mergers. Most of the systems (5/7) require two atomic gas-rich progenitors, and we frequently (3/7) discover massive gaseous tidal features (MHI > 109 M 0 ) which extend well beyond the end of the optical tails. In contrast to what was found optically, we find a much less significant difference in the average H I tail lengths of the two samples (see Table 1).

HI DISTRIBUTION IN LIR GALAXIES 49

The major difference is that a broader range of encounter spin orientations are represented in the IR sample. There are prograde-prograde encounters (N2623, IC 883), prograde-retrograde encounters (Arp 299 and N1614) and a retrograde-retrograde encounter (VV 114). The systems from the Toomre Sequence, on the other hand, experience prograde-prograde encounters. (The spin geometries of Arp 220 and Mkn 273 from the IR sample and N3921 from the optical sample cannot be constrained due to the lack of two distinct kinematic systems in HI).

It therefore appears that the IR luminousity criterion selects for a broader range of encounter geometries than found for the optically selected systems of the Toomre Sequence. This fact combined with a more favorable viewing geometry for the Toomre Sequence objects can probably account for the differences reported in Table 1. However, it would be reassuring to find more IR luminous mergers with long optical tails and to show that apparently short and/or poorly defined optical features are the norm in merging encounters. An imaging study is underway to address these points.

2.1. DESCRIPTION OF INDIVIDUAL OBJECTS

Figures 1-3 show the H I distribution (contours) superimposed upon an optical image (greyscales) of each of the systems in this study. The HI contours are drawn at (5,10,20,40,80,160) xl019cm-2 . All systems show central H I absorption against a radio continuum source, which is indicated by the dashed contours. Because of the central absorption, all H I mass estimates in the following are lower limits. A white bar indicates 20 kpc. Optical images are from the digitized Palomar Sky Survey (A299, N1614, VV114 and IC883), or 300 sec R-band images from the UH 88" telescope (A220 and Mk273) or the KPNO 2.1 m telescope (N2623).

Arp 220 [log(LIR/Le )=12.18]. There is 2xl09 Me of neutral hydrogen detected in emission from the outer regions of this ultraluminous IR object. It is not clear whether these features are a single continuous feature, or two separate features. The kinematics of the gas to the west of the main body are continuous with the gas directly to the north, suggesting that they are part of a continuous structure which is broken by the northern optical tidal plume. There is a clear anti-correlation between the observed H I and the optical tidal features, especially near the northern plume.

Arp 299 [log(LIR/ Le )=11.91]. The H I morphology is radically different from the starlight, with a narrow tidal filament without any apparent optical feature stretching 150 kpc to the north and containing 2x 109 Me of atomic gas. This filament suggest a very old encounter age (",750 Myr) for this very young starburst ('" 20 Myr; Augarde & Lequeux 1985, A&A, 147,273). Back in the main body, there is over 4x109 M e of HI associated


Figure 1. Arp 220 and Arp 299


Figure 2. NGC 1614 and VV 114


Figure 3. Mkn 273, NGC 2623, and IC 883

with both disks, the majority of which is associated with the NE disk. The sense of rotation derived from the plume is opposite that derived from the ionized gas in the NE disk (Hibbard et al. in prep), suggesting a prograderetrograde encounter. NGC 1614 [log(LlR/ L0 )=11.61]. Once again there is a long tidal filament (109M0 , 85 kpc) without an optical counterpart and not connecting


back in any obvious way to any of the optical tidal features. There is at least 109 M0 of H I associated with the main body. The rotation of the ionized gas in the nearly face-on disk (Neff et al. 1990, AJ, 99, 1088) gives an opposite sense of motion than for the H I tail, again suggesting a prograde-retrograde encounter. VV 114. [log(LIR/L0 )=11.62]. The HI kinematics suggest two separate components rotating in the same direction. The lack of any tail-like features and the fact that so much gas (>8.4x 109 M0!) remains close to the remnant suggests a retrograde-retrograde encounter between two gas-rich disks. There is an optical shell-like feature extending to the north anticorrelated with the gas in this region. Mkn 273 [log(LIR/ L0 )=12.14]. This system has a single gas rich tail extending to the south with smooth kinematics, suggesting at least one gas rich progenitor experiencing a prograde encounter. There is a second optical tidal feature extending to the northeast from which we detect no H I, although this may be due to its more face-on orientation. IC 883 [log(LIR/ L0 )=11.60]. The H I emission from this system is confined mostly to the crossed optical tails. Both tails have similar quantities of atomic gas (1.5x109M0 in the east, 1.1x109M0 in the west) and similar lengths (30 kpc), and move in opposite directions, suggesting a merger involving two prograde gas-rich disks. NGC 2623 [log(LIR/ L0 )=11.55]. This, the least IR luminous of the IR bright sample also happens to be the most IR luminous system in the Toomre Sequence. The tidal features are very similar to most of the galaxies in that sequence, with two gas-rich tails (109 M0 in the east, 3.4x 109 M0 in the west) moving in opposite directions, again suggesting a merger of two prograde gas-rich spirals. The H I in the western tail extends beyond the end of the optical light by a factor of three in projected distance, to a total length of 85 kpc.

References

Hibbard, J.E., 1995, Ph.D. Thesis, Columbia Universit.y. Sanders, D.B., Scoville, N.Z., Soifer, B.T., 1991, ApJ, 370,158. Sanders, D.B., Soifer, B.T., Elias, J.H., Madore, B.F., Matthews, K., Neugebauer, G.,

Scoville, N.Z., 1988, ApJ, 325, 74. Scoville, N.Z., Sargent, A.I., Sanders, D.B., Soifer, B.T., 1991, ApJ, 366, L5. Soifer, B.T., Sanders, D.B., Neugebauer, G., Danielson, G.E., Lonsdale, C.J., Madore,

B.F., Persson, S.E., 1986, ApJ (Lett.), 303, L41. Toomre, A., 1977, in "The Evolution of Galaxies and Stellar Populations", eds. B.M.

Tinsley and R.B. Larson (New Haven: Yale Univ.), p. 401. Toomre, A., Toomre, J., 1972, ApJ, 178, 623.

MOLECULAR GAS AND DUST IN INFRARED LUMINOUS GALAXIES

U. LISENFELD AND R.E. HILLS MRAO, Cavendish Laboratory, Cambrigde, UK

S.J .E. RADFORD NRAO, Tuscon, Arizona, USA

AND

P.M. SOLOMON Astronomy Program, State Universtity of New York, USA

Abstract. We have carried out observations with the James Clerk Maxwell Telescope (JCMT) of the CO(3-2) and CO(2-1) lines and of the submillimeter continuum of a sample of 7 far-infrared (FIR) luminous galaxies. All of these galaxies possess FIR luminosities higher than 1011 Lev. Together with data for the CO(1-0) intensities from the literature, we derived the ratios of the brightness temperatures R21 = n(2 - l)/n(1- 0) and R32 = n(3 - 2)/n(2 - 1). The galaxies exhibit, in spite of their similar FIR and CO luminosities, very different excitation characteristics. In some galaxies, both line ratios are around 1, consistent with thermalized, optically thick CO. In other galaxies, however, one or both line ratios are significantly lower than 1, indicating the CO is subthermally excited in regions of only moderate molecular gas density. The submillimeter continuum emission can be well explained by warm, thermally emitting dust (Tdust = 30 - 40 K ). The gas masses estimated from the 800 /lm continuum flux, Mg,800, and from the CO emission, Mg,co, are similar, although Mg,co is, especially at lower masses, systematically higher than M g,800.

1. Introd uction

Far-infrared (FIR) luminous galaxies, with luminosities of more than 1012 Lev are among the most important discoveries made with IRAS. Previous ob-

55

M. N. Bremer et al. (eds.J. Cold Gas at High Redshift. 55-59. © 1996 Kluwer Academic Publishers.

56 U. LISENFELD ET AL.

servations have shown that these galaxies contain very large quantities of molecular gas, as indicated by high CO(I-0) line intensities (Sanders et at., 1988) emitted from within the central few kpc (Scoville et at., 1991; Radford et at., 1991; Planes as et at., 1991). They are also very rich in dense molecular gas, as traced by e.g. HCN. The most luminous galaxies have LHCN / Leo ratios up to ten times higher than normal galaxies (Solomon et at., 1992). Observations of the C0(1-0) and CO(2-1) lines have shown that the ratio of the brightness temperatures of these lines is rather low, indicating that the CO is subthermally excited in regions of moderate H2 density (Radford et at., 1991).

In order to obtain more information about the physical conditions of the molecular gas in these galaxies we have performed observations of the CO(a-2) and CO(2-1) emission. Furthermore, we observed the submillimeter continuum at several wavelengths, which, together with the IRAS fluxes, allows the dust temperature and mass to be derived.

2. Observations

The observations were carried out in June 1994 using the 15-m James Clerk Maxwell Telescope (JCMT) on Mauna Kea, Hawaii. 1 A detailed presentation of the observations and the data will be given elsewhere (Lisenfeld et at., 1995). For the CO observations, single-channel SIS receivers [Receiver A2 for CO(2-1), Receiver B3 for CO(3-2)] were used together with the DAS backend spectrometer. The beam sizes (FWHM) are 21/1 at 230 GHz and 14/1 at :{45 GHz. The observations of the continuum at 450 11m, 800 lun, and 1.1 mm were performed using the UKT 14 receiver. The beam sizes are 18.7", 16.0", 17.5" at 1.1 mm, 800 11m, and 450 11m, respectively.

3. Results

:l.l. co LINE EMISSION

Table 1 summarizes the integrated CO intensities and the line ratios R21 = Tb(2 - 1)/TB(1 - 0) and R32 = Tb(a - 2)/TB(2 - 1), calculated as

(1)

IThe James Clerk Maxwell Telescope is operated by the Royal Observatory, Edinburgh (ROE) on behalf of the United Kingdom Particle Physics and Astrophysics Research Council (PPARC), the Netherlands Organisation for the Advancement of Pure Research (NWO), the Canadian National Research Council (NRC), and the University of Hawaii (UH).

MOLECULAR GAS AND DUST IN IR-LUMINOUS GALAXIES 57

TABLE 1. CO intensities and line ratios

Galaxy leop-O) leO(2-I) Ico(3-2) R21 R32

[K km S-I] [K km S-I] [K km S-I]

IRAS 1056 22(1) 6.0 < 2 0.3 <0.15

Mrk 231 22(1) 15.1 19.8 0.7 0.6

Arp 193 36(1) 36.3 63.7 1.0 0.8

Mrk 273 10(2) 20.4 15.7 1.0 0.4

Arp 220 109(1) 72.5 135.0 0.6 0.9

NGC 6240 63(1) 72.6 161.5 1.1 1.0

IRAS 1720 7(3) 35.0 19.6 1.1 0.3

The CO(1-0) line intensities are taken from the literature: (1) Downes et al., (1993), (2) Kriigel et al., (1990), both observed with the IRAM 30-m telescope, (3) Mirabel et aI., (1990), observed with the 15-m SEST.

with a corresponding expression for R21 . Here, J co is the total integrated CO line intensity (f T mbdv), z denotes the redshift of the source, ds the source diameter, and dB the beam diameter. The size of the CO-emitting region has been measured for some of the galaxies by interferometer observations or mapping. Scoville et al. (1991) have shown that tiO% of the C0(1-0) emission of Arp 220 originate in the central 1.7/1. Planesas et al. (1991) determined the size of the CO-source in IRAS 1720 (d s = 4/1). Radford et al. (1991) have mapped the CO emission from Mrk 231 and Arp 1!);{ and set an upper limit for the source diameter of 7/1. We will assume a source diameter of 4/1 for all the galaxies. (The line ratios would be less than 10% changed if ds = 0).

The ratios of the brightness temperatures show differences among the galaxies, indicating the physical conditions of the molecular gas are not the same - in spite of their similar FIR luminosities.

1. In two galaxies (NGC 6240, Arp 193) the line ratios are compatible with 1. This is consistent with optically thick, thermalized CO in regions where the densities are n(H2) 2: 5 X lO3 cm-3 and the temperahues Tex ~ Tkin 2: 50 K.

2. The other galaxies show, to different degrees, evidence for subthermal excitation. In IRAS 1720 and Mrk 273, R21 is compatible with one, but R32 is significantly less than one. In Arp 220 and Mrk n1, both R21 and R32 are smaller than 1. The most extreme case is IRAS 1056, where both R21 and R32 are much smaller than 1, indicating the density and/or temperature are too low to produce significant excitation ofthe higher levels of the CO molecule. Since the temperature is probably

58 U. LISENFELD ET AL.

TABLE 2. Dust emission and gas masses

Galaxy 8 450 8800 8 1100 log(M~,800) log(Mg,co) [Jy] [Jy] [Jy] [M8] [M8]

Mrk 231 1.02 ± 0.17 0.96 ± 0.02 < 0.03 10.1 lOA

Arp 193 0.99 ± 0040 0.08 ± 0.02 < 0.03 9.5 10.1

Mrk 273 0.89 ± 0.15 0.10 ± 0.02 0.036 ± 0.02 10.0 10.3 Arp 220 3.97 ± 0.52 0.69 ± 0.02 0.11 ± 0.02 10.2 lOA

NGC 6240 1.0 ± 0.24 0.15 ± 0.03 0.08 ± 0.03 9.8 lOA

IRAS 1720 1.04 ± 0.14 0.15 ± 0.01 0.05 ± 0.01 10.3 10.6

higher than the excitation threshold for the J = 2 and J = 3 levels (T = 30 K), the CO most likely does not trace the high density regions traced by e.g. HCN, but rather regions where the density is n(H2) < 5 X 103 cm-3 .

These observations can be understood if some of the galaxies contain large amounts of low density molecular gas in addition to the abundant dense gas traced by HCN. There is a slight tendency for R32 to decrease with increasing LFIR, which might indicate high luminosity galaxies are more likely to possess such a low density component.

3.2. DUST CONTINUUM EMISSION

The submillimeter data and the IRAS 60 and 100 !-lm were fitted by thermal dust emission, S", = Q(v)B(v, T), where B(v, T) is the Planck function and Q(v) ex: vf3 is the dust emissivity. The best fit dust temperatures were derived for {3 = 1 and {3 = 2. We did not fit for {3, but note the fits were generally better for {3 = 2. The dust temperatures derived in this case were 30 - 40 K.

The submillimeter continuum emission can be used to calculate the dust mass, and, assuming a dust-to-gas ratio, to estimate the gas mass. We calculated the gas mass from the 800 !-lm flux, with {3 = 2 and adopted the values given in Hildebrand (1983), who suggests a gas-to-dust mass ratio of 100. The gas mass was also calculated from the CO(I-0) luminosity, LCO(l-O), adopting the standard Galactic ratio Mg,co/ L CO (1-0) = 4.8 MG (K km s-l pc2)-l,

In Table 2 the data for 6 galaxies (IRAS 1056 was not observed) are presented together with the gas masses. The masses derived by the two methods are displayed in Fig. 1. The two methods agree reasonably, although Mg,co is systematically higher (factor 1.5-4) than the mass derived from the 800 !-lm flux, M g,800. This may be explained by a slight under-

MOLECULAR GAS AND DUST IN IR-LUMINOUS GALAXIES 59

~ en 9

+

9.5

+

+

+ +

+

I

10 10.5

Figure 1. The mass derived from LCO(I-O), Mg,co, and the mass derived from the 800 JIm continuum, M g,800

estimate of the dust-to-gas ratio. There is also a tendency for Mg ,800 to increase faster than M g,CO, which might indicate a larger dust- to-gas ratio in higher mass (which are at the same time higher luminosity) galaxies. This tendency should, however, be regarded with caution both because of the small number of galaxies involved and because of the different beam sizes at 800 {lm and CO(l-O).

References

Downes D., Solomon P. M., Radford S.J.E., 1993, ApJ 414, L13 Hildebrand R.H., 1983, Q.JI. R. astr. Soc. 24, 267 Krugel E., Steppe H., Chini R., 1990, A&A 229, 17 Lisenfeld U., Hills R., Radford S.J.E., Solomon P.M., 1995, in preparation Mirabel I.F., Booth R.S., Garay G., Johansson L.E.B., Sanders D.B., 1990, A&A 236,

327 Planesas P., Mirabel I.F., Sanders D.B., 1991, ApJ 370,172 Radford S.J.E., Solomon P.M., Downes D., 1991, ApJ 368, LI5 Sanders D.B., Soifer B.T., Elias J.H., Madore B.F., Matthews K., Neugebauer G., Scoville

N.Z., 1988, ApJ 325, 74 Scoville N.Z., Sargent A.I., Sanders D.B., Soifer B.T., 1991, ApJ 366, L5 Solomon P.M., Sage L.J., 1988, ApJ 334, 613 Solomon P.M., Downes D., Radford S.J.E., 1992, ApJ 387, L55

THE EVOLUTION OF THE FAR-INFRARED GALAXY POPULATION

MICHAEL ROWAN-ROBINSON

Imperial College of Science, Technology and Medicine Blackett Laboratory, Prince Consort Rd, London SW7 2BZ

Abstract. Direct estimates of the rate of evolution of IRAS galaxies from redshift surveys are reviewed. There is now good evidence for strong evolution of the starburst galaxy population and this is supported by the interpretation of the 60 11m source-counts. The sub-m.Jy radio source population at 1.4 G Hz has also been found to be essentially identical to the IRAS starburst galaxy population and faint radio source-counts support luminosity evolution models.

A new model is presented for source counts and integrated background radiation from radio to optical wavelengths. For normal spirals, only 30% of their energy, on average, is emitted in the far-infrared, so they must be optically thin. For starburst galaxies we estimate that 95% of their optical and UV radiation is absorbed by dust and reemitted at far infrared wavelengths.

The case is made that the new class of hyperluminous infrared galaxies, of which there are now more than 20 examples, represents galaxies undergoing a very major star formation episode.

1. Direct estimates of the rate of evolution of IRAS Galaxies

The evolution of the 60 11m luminosity function can be characterized by:

TJ (L, z) = </> (z) 1]0 (L, L* (z))

where, for density evolution we take

</>(z) (1 + z)P, Z < Zf

0, z > Zf,

61


(1)

(2)

62 MICHAEL ROWAN-ROBINSON

and for luminosity evolution we take

(1 + Z)Q, (l+z*)Q, 0,

Z < Z*

z*<z<zr Z > Zf.

(3)

The most direct evidence for evolution comes from large red shift surveys. Saunders et al. (1990) found that for the QDOT 1-in-6 sparsely sampled IRAS galaxy redshift survey to S(60) = 0.6 Jy:

Q = 3.1 ± 1.0 or P = 6.7 ± 2.3.

(4)

After correction for redshift errors, for the effects of Malmquist bias, and for non-linearity in the IRAS PSC flux-scale, Oliver et al. (1995) found that this should be revised to

P = 4.2 ± 2.3. (5)

From the shallower but larger 1.2 Jy survey, Fisher et al. (1992) found no clear evidence for evolution:

P = 2±3 (6)

but the uncertainty is such that the QDOT values are not inconsistent with this.

The much deeper IRAS FSS redshift survey of Oliver et al. (1995), which consists of 1400 galaxies in 700 square degrees with S(60) ~ 0.2 Jy, after correction for Malmquist bias, and with a J( -correction based on a 2-component (starburst+cirrus) fit to the 100/60 J..Lm colours, yields strong evidence for evolution:

P Q

V /Vrnax method 5.5 ± 1.8 3.7 ± 0.9

binned V /Vrnax method 4.9 ± 1.4 2.6 ± 0.8

maximum likelihood 6.3 ± 1.5 a.5 ± 0.7

average 5.6 ± 1.6 a.3 ± 0.8

2. Evidence for evolution from 60 micron counts

Several groups have found that the 60 J..Lm source counts can only be understood ifthe IRAS galaxy population is subject to strong evolution (Hacking and Houck 1987; Hacking et al. 1987, 1989; Danese et al. 1987; Lonsdale and Hacking 1989; Lonsdale et al. 1990; Hacking and Soifer 1991; Oliver et

EVOLUTION OF THE FIR GALAXY POPULATION 63

al. 1992). In Fig. 1 of Oliver et al. (1992), for example, observations from the IRAS PSC, the IRAS FSS and the pointed observations of Hacking and Houck (1987), are compared with predictions for the case of no evolution, and for strong density and luminosity evolution. The no evolution case is clearly inconsistent with the observations, while either density or luminosity evolution give consistent results. Counts with ISO will allow the latter two models to be distinguished.

3. Sub-mJy radio-sources as starburst galaxies

The bright radio source-counts (5(1.4 GHz) ~ 1 mJy) are due to radio galaxies and quasars undergoing strong evolution, which is approximately of the form of luminosity evolution (e.g., Condon 1984; Dunlop and Peacock 1990), though some models involve a small amount of density evolution also. Below 1 mJy at 1.4 GHz the slope of the counts steepens again and there is evidence of a new population of blue radio-emitting galaxies (Mitchell and Condon 1985; Windhorst et al. 1987; Thuan and Condon 1987; Franceschini et al. 1988; Condon 1989; Lonsdale and Harmon 1991).

Benn et al. (1993) have carried out spectroscopy of a sample of 112 identifications of sources with 5(1.4 GHz) ~ 0.1 mJy and shown that below 1 mJy most of the sources are starburst galaxies very similar to those seen by IRAS. The luminosity function for these galaxies agrees well with that at 60 J.Lm, shifted by the radio-FIR relation

5(60 J.Lm) = 905(1.4 GHz). (7)

Fits to the sub-mJy source-counts then show that strong evolution is required in this population, with luminosity evolution strongly favoured over density evolution (Rowan-Robinson et al. 1993a).

4. A new model for counts and background radiation from radio to X-rays

There have been a number of attempts to model the source-counts and background radiation at a wide range of wavelengths (e.g., Franceschini et al. 1991, Blain and Longair 1993). We describe here a new study by Pearson and Rowan-Robinson (1996) which gives predictions of source-counts at radio, submillimetre, far-infrared, near-infrared and optical wavelengths, and is being extended to X-rays. The galaxy populations included are: (a) normal spirals + cirrus, characterized by cool 100/60J.Lm colours, (b) starburst and ultraluminous IR galaxies, characterized by warm 100/60 11m colours, (c) the new category of hyperluminous IR galaxies, with L 60 /Lm > 1013 L8 (0 = 1, Ho = 50kms-1 Mpc-1 ), (d) Seyferts and quasars + dust


tori, and (e) elliptical galaxies. Clearly the separation of populations (a) and (b) is an oversimplification since there is some star formation proceeding in normal spirals.

For classes (b), (c), (d) we assume strong power-law luminosity evolution with Q = 3.1, Z* = 2, Zr = 5, which gives a good fit to the optical data for QSOs (Boyle et al. 1988), fits the bright radio source-counts for radio galaxies and quasars (Dunlop and Peacock 1990), is consistent with the direct evidence for evolution in IRAS galaxies (Sect. 1 above) and fits the sub-mJy source counts (Rowan-Robinson et al. 1993a). The physical motivation for such a common evolution would be that interactions simultaneously drive star formation and feed gas into black holes in active galactic nuclei. Thus the evolution represents the evolution of the gas supply from interactions with time.

The spectral energy distributions we have assumed for the different populations are shown in Table 1.

TABLE 1. Spectral energy distributions assumed for the different populations

Population FIR optical/ultraviolet

(a) normal spiral cirrus model Sbc (Coleman et al. 1980, (Rowan-Robinson 1992) Yoshii et al. 1988), 30% of

light assumed absorbed by dust and reradiated in IR

(b) starburst star burst model Sab, v < VB (Coleman (Rowan-Robinson & et al. 1980) Efstathiou 1993) HII, v > VB (Mrk36,

Neugebauer et al. 1976) 95% of light absorbed by dust and reradiated in IR

(c) hyperluminous IRAS F10214+4724 continuum (Rowan-Robinson et al. 1993b)

(d) Seyferts/quasars torus model average QSO continuum

(Rowan-Robinson 1995) (Rowan-Robinson 1995)

With these assumed spectral energy distributions and rates of evolution, we then need to define the luminosity functions for each population at only one wavelength. For normal spirals and ellipticals we use the B-band Schechter functions of Efstathiou et al. (1988). The assumption that 30% of the optical-UV light is absorbed by interstellar dust and reemitted in the far-infrared then yields a good fit to the 60/-Lm luminosity function


for galaxies with "cool" far-infrared colours (Saunders et al. 1989). For the starburst and hyperluminous galaxies we use the 60 j.Lm luminosity function for "warm" galaxies (Saunders et al. 1990). Finally for Seyferts and quasars we use the 12 j.Lm luminosity function data of Rush et ai. (1993), refitted with a 2 power-law function (Lawrence et al. 1986).

The resulting fits to the 1.4GHz, 60j.Lm, K-band and B-band counts are extremely good (Figs. 1-2). The inclusion of a strongly evolving population of starburst galaxies accounts for the excess B-band counts down to B = 22m , but not the fainter excess of blue galaxies. The increasing proportion of galaxies with emission lines in fainter optical galaxy redshift surveys finds a natural explanation in this picture. This analysis shows that the proportion of the optical-UV light from a typical starburst that is absorbed by dust and reemitted in the far-IR cannot be much less than 95%, otherwise the optical counts would be violated. On the other hand the figure can not be 100%, otherwise no emission lines or continuum colour changes would be seen in starburst galaxies. Inclusion of stellar evolution effects in the treatment of the B-band SED and evolution of normal galaxies will have only a small effect on these models.

We can now predict the far-infrared background from these models and compare these with the upper limits derived from COBE. The predicted intensity of the background is well below the latest DIRBE limits (Hauser 1995), but is interestingly close to the FIRAS limit at 500 j.Lm. It is a]so within the range calculated from absorption of TeV ,),-rays (De Jager et ai. 1994).

5. Nature of ultraluminous infrared galaxies

One of the major discoveries of the IRAS mission was the existence of ultraluminous infrared galaxies, galaxies with LFIR > 1012 h502 LtV (hso = Ho/50). The peculiar Seyfert 1 galaxy Arp 220 was recognised as having an exceptional far-infrared luminosity early in the mission (Soifer et al. 1984). Sanders et al. (1988) discussed the properties of 10 IRAS ultraluminous galaxies with 60 j.Lm fluxes > 5 J y and concluded that (a) all were interacting, merging or had peculiar morphologies (b) all had AGN line spectra. On the other hand Leech et al. (1989) found that only 2 of their sample of6 ultraluminous IRAS galaxies had an AGN line spectrum. Leech et al. (1994) found that 67% of a much larger sample of ultraluminous galaxies were interacting, merging or peculiar. Lawrence et al. (1989) had found a much lower fraction amongst galaxies of high but less extreme infrared luminosity. The incidence of interacting, merging or peculiar galaxies by IR luminosity is summarised in Fig. 1 o{Rowan-Robinson (1991). The situation on point (b) remains controversial, though, since Lawrence et al. (1995) find only


------------ ALLCOMPO~ • --------------------- EUJP11CAUl

------------ STAR BURSTS - - - - - - STARBURSTS(noevoJulion)

1&(5) (Jy)

~,--------.--------~------~-------,--------,m ,

0 HACKING &. HOUCK 1987

I AU roMPONENTS .. ROWAN-ROBINSON et aI. 1990 --------------------- NORMALGALAXIES • SAUNDERS 1990

... 3.6 ____________ STARBURSTS .t! "l -If If

5 3.2

Vl N If If CIl CIl 2. • .:g

~ o 32.4 ----- ---------------

2+-----'~._------r_----~------_r----__4 -3 .2 -I 0 1

LOG(FLUX) (Jy)

Figure 1. Normalized differential source counts with predicted contributions of different populations at (a) 1.4GHz, (b) 60 I'm.

';

'" .g 4 01) <1) -0 C'

-S <1)

.CJ

8 ::> ..s 2

8

~ -0

o g

o

·3


ALL OOMPONENTS

------- -- - - -- ------- NORMAL GALAXIES

o COWIE et Ill. 90

A GLAZEBROOK~IlI.~

oQo MOBASHER et al 86

<> OCW93HDS

v OCW93HMDS

t3 OCW 931IMWS

" OCW93HWS

r+..~.-l~-~L--'--._--'--.---'--._--r---'-T 11 12 13 14 IS 16 17 18 19 20 21 22 23

././ ./

KMAGNlTUDE

//-........_-------...., ./

ALL OOMPONENTS

------ - -- -- ---- -- ---- NORMALGALAXlES ------------ STARBIJRSfS - - - - - - HYPERLUMINOUS

W lYSON 1988

A APM

<> METCALFE et 01. 1991

.. ULlYet 01.1991 v EDSGC

~.~~~~./~-r_,,-.-r_ . .__.-.__r_.-.__r_4 ~ U 17 1B ~ ~ ~ 22 D ~ ~ u n u ~ ~

BJ MAGNITUDE

Figure 2. Normalized differential source counts with predicted contributions of different populations at (a) K-band, and (b) B-band.


a fraction 21% of 81 ultraluminous galaxies in the QDOT sample to have AGN spectra.

Rowan-Robinson and Crawford (1989) found that their standard starburst galaxy model gave an excellent fit to the far infrared spectrum of Mrk 231, an archetypal ultraluminous IR galaxy. However their models for Arp 220 appeared to require a much higher optical depth in dust than the typical starburst galaxy. Condon et al. (1991) showed that the radio properties of most ultraluminous IR galaxies were consistent with a starburst model and argued that these galaxies required an exceptionally high optical depth. This suggestion was confirmed by the detailed models of RowanRobinson and Efstathiou (1993) for the far-infrared spectra of the Condon et al. sample.

Quasars and Seyfert galaxies, on the other hand, tend to show a characteristic mid infrared continuum, broadly flat in vSlI from 3 to 30 {Lm. This component was modelled by Rowan-Robinson and Crawford (1989) as dust in the narrow-line region ofthe AGN with a density distribution n( r) <X r- 1 .

More realistic models of this component based on a toroidal geometry are given by Pier and Krolik (1992), Granato and Danese (1994), RowanRobinson (1995), and Efstathiou and Rowan-Robinson (1995). Rowan-Robinson (1995) suggests that most quasars contain both (far-IR) starbursts and (mid-IR) components due to (toroidal) dust in the narrow line region.

6. Hyperluminous infrared galaxies

In 1988 Kleinmann et al. identified P09104+4109 with a z = 0.44 galaxy, implying a total far-infrared luminosity of 1.5 X 1013 L0 , a factor 3 higher than any other ultraluminous galaxy seen to that date. In 1991, as part of a program of systematic identification and spectroscopy of a sample of 3400 IRAS FSS sources, Rowan-Robinson et al. discovered IRAS F10214+4724, an IRAS galaxy with z = 2.286 and a far-infrared luminosity of 5 X

1014 h5~ L0 . This object appeared to presage an entirely new class of infrared galaxies. The detection of a huge mass of CO by Brown and Vanden Bout (1991),1011 hS02 M0 confirmed by the detection of a wealth ofmolecular lines (Solomon et al. 1992), and of sub millimetre emission at wavelengths 450 - 1250 {Lm (Rowan-Robinson et al. 1993, Downes et al. 1992), implying a huge mass of dust, 109 h5~ M0 confirmed that this was an exceptional object. Early models suggested this might be a giant elliptical galaxy in the process of formation (Elbaz et al. 1992). Simultaneously with the growing evidence for an exceptional starburst in IRAS F10214+4724, the Seyfert 2 nature of the emission line spectrum (Rowan-Robinson et al. 1991, Elston et al. 1994a) was supported by the evidence for very strong optical po-


larisation (Lawrence et al. 1993, Elston et al. 1994b). Subsequently it has become clear that IRAS FI0214+4724 is a gravitational lens system (Graham et al. 1995, Broadhurst and Lehar 1995, Serjeant et al. 1995) with a magnification of about 10 at far-infrared wavelengths, but not much greater than that (Green and Rowan-Robinson 1995). Even when the magnification of 10 is aU owed for, IRAS FI0214+4724 is still an exceptionally luminous far- IR source.

In 1992 Barvainis et al. successfully detected submillimetre emission from the z = 2.546 "clover-leaf" gravitationally lensed QSO, HI413+117, which suggested that H1413 is of similar luminosity to IRAS F10214+4724.

The program of follow-up of IRAS FSS sources which led to the discovery of IRAS FI0214+4724 has also resulted in the discovery of a further 8 galaxies or quasars with far-IR luminosities > 1013 hs~ L8 (McMahon et ai. 1994). Cutri et al. (1994) report a search for IRAS FSS galaxies with 'warm' 25/60 J.Lm colours, which yielded the z = 0.93 Seyfert 2 galaxy, IRAS FI5307+3252. Dey and Van Breugel (1995) report a comparison of the Texas radio survey with the IRAS FSS catalogue, which yielded 5 galaxies with far-IR luminosities> 1013 hs~ L0 .

Finally, inspired by the success in finding highly redshifted sub millimetre continuum and molecular line emission in IRAS FI0214+4724, several groups have studied an ad hoc selection of very high redshift quasars and radio galaxies, with several notable successes (Andreani et ai. 1993, Dunlop et al. 1995, Isaak et al. 1994, McMahon et al. 1994, Van Ojik et al. 1995, Ivison 1995). Most ofthese detections imply far-IR luminosities> 1013 hsl L8 , assuming that the far-IR spectra are typical starbursts. In addition there are 2 PG quasars from the sample detected by IRAS and studied by RowanRobinson (1995), which also satisfy this condition.

Table 2 summarizes the properties of 13 galaxies with far-infrared luminosities estimated to be ;::: 1013 hS02 L0 ; a further four have been detected by Dey and Van Breugel (1995) and seven by McMahon et al. (1996). We define these as a new class of hyperluminous IR galaxies. The remainder of this paper is devoted to a discussion of the properties of these objects and their significance.

7. Models for hyperluminous infrared galaxies

For a small number of these galaxies we have reasonably detailed continuum spectra from radio to UV wavelengths. The continuum emission from IRAS FI0214+4724 was the subject of a detailed discussion by RowanRobinson et ai. (1993). Green and Rowan-Robinson (1995) have discussed starburst and AGN dust tori models for IRAS FI0214+4724 and for IRAS F15307 +3252. Figure 3 shows the continua. of these and several other


TABLE 2. Hyperluminous infrared galaxies

Name z

TX0211-122 2.34 4C 0647+4134 3.8 P09104+4109 0.44 IRAS F10214+4724 2.286 BR1033-0327 4.51 PG 1148+549 0.969 BR1202-0725 4.69 H1413+1l7 2.546 IRAS F1421+3845 0.99 8C 1435+635 4.26 IR.AS F15307+3252 0.93 PG 1634+706 1.334 PC 2047+0123 3.80

Van Ojik et al. 1994

2 Dunlop et al. 1995

3 Kleinmann et al. 1988

reference

1 2 3 4 5 6 7 8 9 10 11

6 10

4 Rowan-Robinson et al. 1991, 1993 5 Isaak et at. 1994

spectrum type

RG RG S2 S2

QSO QSO QSO

BALQSO QSO RG S2

QSO QSO

6 Sanders et al. 1989, Rowan-Robinson 1995 7 McMahon et al. 1994

8 Barvainis et al. 1992

9 McMahon et al. 1996

10 Ivison 1995 11 Cutri et al. 1994

log(L.b}

14.81 13.46 13.17 14.90 13.26 13.66 13.90 15.02 14.14 13.10 13.77 13.79 12.93

hyperluminous galaxies, with fits using radiative transfer models (generally the standard starburst model of Rowan-Robinson and Efstathiou (1993) or the standard QSO dust model of Rowan-Robinson (1995).

For the remaining objects in Table 2 we have only 60 [Lm or single submillimetre detections and for these we estimate their far-infrared luminosity, and other properties, using the standard starburst model of RowanRobinson and Efstathiou (1993).

8. The significance of hyperluminous infrared galaxies

In Fig. 4 we show the far-infrared luminosity against redshift for the galaxies of Table 2, with lines indicating observational constraints at 60, 800 and 1250 [Lm. Of the sources with luminosities above 5 X 1014 h5; L0' two are gravitationally lensed (IRAS FI0214+4724, M ~ 10; HI413+1l7, M ~


~

14

~ ./ ~ +

+

13

I~ 12 + +

~

IgYL. 12 i-

t ~ .. • 14 1002 + t+

13 ;::~ 12

~

~\ .. 12

13 t + ~0023

12

Figure 3. Starburst + AGN dust tori models for the far infrared emission from hyperluminous IR galaxies, log vLv in solar units versus log v in Hz.

15

13

o z

• - - - ... - - -A-

• • •

4 5

Figure 4. Bolometric star burst luminosity versus redshift for hyperluminous IR galaxies: Crosses: IRAS FSS galaxies, triangles: 800 11m detections, filled circles: 1250 11m det.ections.


7.6), and the other 3 are based on IRAS detections of low quality (from the IRAS FSS "Reject" File). If these are confirmed, they might be strong candidates for also being lensed. On the other hand there is overwhelming evidence for a population of galaxies with far-IR luminosities in the range 1- 50 X 1013 h5~ L0 . By analogy with IRAS F10214+4724, we believe that the rest-frame radiation longward of 50 J-Lm comes from a starburst component. The luminosities are such as to require star formation rates in the range 1 - 50 X 103 h5~ M0 yr-1 , which would in turn generate most of the heavy elements in a 1011 M0 galaxy in 107 - 108 yrs. Most of these galaxies can therefore be considered to be undergoing their most significant episode of star formation, i.e., to be in the process of "formation".

It appears to be significant that a large fraction of these objects are Seyferts, radio-galaxies or QSOs. For the very high-redshift (z > 3) objects, this is a selection effect in that quasars and radio galaxies are the only objects known at such redshifts. However even for the population of objects found from direct optical follow-up of IRAS samples (and omitting objects found in searches biassed to "warm" 60/25 J-Lm colours, which are biassed towards galaxies with a 3 - 30 J-Lm dust tori component), out of 10 objects, 5 are QSOs, one is Seyfert 1, 2 are Seyfert 2, and only 2 are narrow-line objects. Thus in a high proportion of cases, this phase of exceptionally high far-IR luminosity is accompanied by AGN activity at optical and UV wavelengths.

In the Sanders et al. (1989) picture, the far-infrared and submillimetre emission would simply come from the outer regions of a warped disk surrounding the AG N. However the weaknesses of this picture as an explanation of the far infrared emission from PG quasars have been highlighted by Rowan-Robinson (1995). A picture in which both a strong starburst and the AGN activity are triggered by the same interaction or merger event is far more likely to be capable of understanding all phenomena (cf., Yamada 1994). '

9. Discussion and future work

The evidence that the starburst galaxy population is undergoing strong evolution seems very strong, both from IRAS galaxy redshift surveys and from the interpretation of the faint 1.4 GHz and 60 J-Lm source counts. The rate of evolution appears to be similar to that seen in optically selected QSOs and in radio-loud quasars and radio-galaxies. A possible mechanism to account for this similarity is that both processes are driven by galaxy interactions and mergers.

We have presented a model for source-counts and integrated background radiation from radio to optical wavelengths and have been able to link the


far-infrared luminosity functions and evolution to the behaviour of the corresponding populations at optical wavelengths. For normal spiral galaxies only about 30% of the total optical and UV light is absorbed and reemitted at far-infrared wavelengths. These galaxies are therefore, on average, optically thin. This result has been clear since the work of Rowan-Robinson et al. (1987) and Rowan-Robinson and Crawford (1989; see also RowanRobinson 1992). It is only by ignoring the far-infrared evidence that claims that normal spirals are optically thick (e.g., Disney et al. 1989, Valentijn 1990, Burstein et al. 1991) can be sustained. Obviously, though, some parts of normal spirals (e.g. giant molecular clouds, the nucleus) are highly optically thick.

We are also able for the first time to give a quantative estimate for the average proportion of optical- UV light in a starburst which is absorbed by dust and reemitted in the far-infrared. We find a figure of 95%, which is consistent with the very high 60/-lm to Ha flux ratios found for starburst galaxies (Leech et al. 1989). A figure much lower than this would lead to violation of the optical source-counts for all galaxies. At this figure, the increasing proportion of emission line galaxies seen in optical galaxy redshift surveys down to B = 23m finds a natural explanation.

Our treatment does not explicitly include star formation in elliptical galaxies, which is at a low rate at present epochs, but will have been much more intense at earlier times. However the strong evolution of the starburst population back to z = 5 assumed in our evolution model must probably be interpreted as including the contribution of f;tar formation in ellipticals.

Our predicted background intensities are sirnilaT to those obtained in earlier calculations (Beichman and Helou 1991; Franceschini et al. 1991; Oliver et al. 1992; Blain and Longair 1993). However direct estimates of the rate of evolution and improved models of the spectral energy distributions of the different populations should significantly reduce the uncertainties in our estimate.

Finally we have reviewed the properties of ultraluminous and hyperluminous infrared galaxies and argued that their far-infrared luminosities are powered by starbursts. The latter probably represent galaxies undergoing their most significant episode of star formation, i.e., they are galaxies in the process of formation.

Future work:

1. We will shortly be able to test the evolution rate in the PSCZ IR.AS galaxy red shift survey, a total of 15000 galaxies with 5(60/-lm) :::: 0.6.Jy (Saunders et al. 1995).

2. We plan to greatly increase the sample size for our sub-mJy surveys using the Australia Telescope and the 2dF facility at the AAT (Mobasher,


Cram and Rowan-Robinson 1995). 3. We have been awarded 215 hours of ISO Open Time for deep surveys

of about 20 square degrees of sky at 90 and 15 ",m (Rowan-Robinson et al. 1995a).

4. We plan to carry out a survey at 850, 450 and 350 ",m with the SCUBA instrument at the JCMT (Rowan-Robinson et al. 1995b).

References

Andreani P., La Franca F., Cristiani S., 1993, MNRAS 261, L35 Barvainis R., Antonucci R., Coleman P., 1992, ApJ 399, L19 Beichman C.A., and Helou G., 1991, ApJ 370, L1 Benn C. R., Rowan-Robinson M., McMahon R.G., Broadhurst T.J., Lawrence A., 1993,

MNRAS 263, 98 Blain A.W., and Longair M.S., 1993, MNRAS 264, 509 Boyle B.J., Shanks T., and Peterson B.A., 1988, MNRAS 235, 935 Broadhurst T. and Lehar L., 1995, ApJ 450, L41 Brown R.L. and vanden Bout P.A., 1991, AJ 102, 1956 Burstein D., Haynes M., Faber S.M., 1991, Nat 353, 515 Coleman G.D., Wu C., Weedman D.W., 1980, ApJS 43, 393 Condon J.J., 1984, ApJ 284, 44 Condon J.J., 1989, ApJ 338,13 Condon J.J., Huang Z.-P., Yin Q.F., Thuan T.X., 1991, ApJ 378, 65 Cutri R.M., Huchra J.P., Low F.J., Brown R.L., Vanden Bout P.A., 1994, ApJ 424, L65 Danese L., deZotti G., Franceshini A., Toffolatti L., 1987, ApJ 318, L15 De Jager O.C., Stecker F.W., Salamon M.H., 1994, Nat 369,294 Dey A. and van Breugel W., 1995, in Mass Transfer Induced Activity in Galaxies Disney M., Davies J., Phillips S., 1989, MNRAS 239, 939 Downes D., Radford S.J.E., Greve A., Thurn C., Solomon P.M., Wink J.E., 1992, ApJ

398, L25 Dunlop J.S., and Peacock J.A., 1990, MNRAS 247,19 Dunlop J.S., Hughes D.H., Rawlings S., Eales S.A., Ward M.J., 1995, Nat 1995 Efstathiou A. and Rowan-Robinson M., 1995, MNRAS 273, 649 Efstathiou G., Ellis R.S., Peterson B.A., 1988, MNRAS 232, 431 Elbaz D., Arnaud M., Casse M., Mirabel I.F., Prantzos N., Vangioni-Flam E., 1992, AA

265, L29 Elston R., McCarthy P.J., Eisenhardt P., Dickinson M., Spinrad H., Januzzi B.T., Mal-

oney P., 1994, AJ 107, 910 Fisher K.B., Strauss M.A., Davis M., Yahil A., Huchra J.P., 1992, ApJ 389, 188 Franceschini A., Toffolatti L., Mazzei P., Danese 1., de Zotti G., 1991, AAS 89, 295 Franceschini A., Danese 1., de Zotti G., Xu C., 1988, MNRAS 233,175 Graham J.R. and Liu M.C.,1995, ApJ 449, L29 Granato G.L. and Danese L., 1994, MNRAS 268, 235 Green S., and Rowan-Robinson M., 1995, (these proceedings/ MNRAS in press) Hacking P.B., and Houck J.R., 1987, ApJS 63, 311 Hacking P.B., Condon J.J., Houck J.R., 1987, ApJ 316, L15 Hacking P.B., Condon J.J., Houck J.R., Beichman C.A., 1989, ApJ 339, 12 Hacking P.B. and Soifer B.T., 1991, ApJ 367, L49 Hauser M.G., 1995, in IAU Symposium 168, Examining the Big Bang and Diffuse Back

ground Radiations Isaak K.G., McMahon R.G., Hills R.E., Withington S., 1994, MNRAS 269, L28 Ivison R.J., 1995, MNRAS 275, L33


Januuzzi B.T., Elston R., Schmidt G.D., Smith P.S., Stockman H.S., 1994, ApJ 429, L49 Kleinmann S.G., Hamilton D., Keel W.C., Wynn-Williams C.G., Eales S.A., Becklin

E.E., Kuntz K.D., 1988, ApJ 328, 161 Lawrence A., Rowan-Robinson M., Leech K.J., Jones D.H.P., Wall J.V., 1989, MNRAS

240, 329 Lawrence A., Walker D., Rowan-Robinson M., Leech K.J., Penston M.V., 1986, MNRAS

219, 687 Lawrence A. et al., 1993, MNRAS 260, 28 Leech K.J., Penston M.V., Terlevich R.J., Lawrence A., Rowan-Robinson M., Crawford

J., 1989, 240,349 Leech K.J., Rowan-Robinson M., Lawrence A., Hughes J.D., 1994, MNRAS 267, 253 Lonsdale C.J. and Hacking P.B., 1989, ApJ 338, 712 Lonsdale C.J.,and Harmon R.T., 1991, Adv.Space Res. 11, 255 Lonsdale C.J., Hacking P.B., Conrow T.P., Rowan-Robinson M., 1990, ApJ 358, 60 McMahon R.G., Omont A., Bergeron J., Kreysa E., Haslam C.G.T., 1994, MNRAS 267,

L9 McMahon et al. 1996 (in prep) Mitchell K.J., and Condon J.J., 1985, AJ 90, 1957 Neugebauer G., Becklin E.E., Oke J.B., Searle L., 1976, ApJ 205, 29 Oliver S.J., Rowan-Robinson M., Saunders W., 1992, MNRAS 256, 15p Oliver S.J. et al. 1995, in Wide-Field Spectroscopy and the Distant Universe, eds.

S.J .Maddox, A. Aragon-Salamanca, World Scientific (Singapore) Pearson C. and Rowan-Robinson M., 1996, MNRAS (in press) Pier E., and Krolik J., 1992, ApJ 401, 99 Rowan-Robinson M. and Crawford J., 1989, MNRAS 238, 523 Rowan-Robinson M. et al. 1991, Nat 351,719 Rowan-Robinson M. and Efstathiou A., 1993, MNRAS 263, 675 Rowan-Robinson M. et al., 1993, MNRAS 261, 513 Rowan-Robinson, M., 1991, in Dynamics of Galaxies and Molecular Cloud Distributions,

eds. F.Combes and F.Fasoli (Kluwer) p.211 Rowan-Robinson M., 1995, MNRAS 272, 737 Rowan-Robinson M., Helou G., Walker D., 1987, MNRAS 227, 589 Rowan-Robinson M., 1992, MNRAS 258, 787 Rowan-Robinson M., Benn C. R., Lawrence A., McMahon R.G., Broadhurst T.J., 1993a,

MNRAS 263, 123 Rowan-Robinson M. et al. 1993b, MNRAS 261, 513 Rowan-Robinson M., 1995, MNRAS 272, 737 Rowan-Robinson M., and Efstathiou A., 1993, MNRAS 263, 675 Rowan-Robinson M., Cesarsky C., Danese 1., Franceschini L., Genzel R., Lawrence A.,

Lemke D., McMahon R., Miley G., Oliver S., Puget J.-1., Rocca-Volmerange B., 1995a

Rowan-Robinson M., Pearson C., Mobasher B., Griffin M., Gear W., Lawrence A., Dunlop J., Longair M.S., Blain A.W., 1995b

Rush B., Malkan M., Spinoglio 1., 1993, ApJS 89, 1 Sanders D.B., Soifer B.T., Elias J.H., Madore B.F., Matthews K., Neugebauer G., Scoville

N.Z., 1988, ApJ 325, 74 Sanders D.B., Phinney E.S., Neugebauer G., Soifer B.T., Matthews K., 1989, ApJ 347,

29 Saunders W.S., Rowan-Robinson M., Lawrence A., Efstathiou G., Kaiser N., Ellis R.S.,

Frenk C.S., 1990, MNRAS 248, 318 Saunders et al. 1995, in Wide-Field Spectroscopy and the Distant Universe, eds.

S.l.Maddox, A. Aragon-Salamanca, World Scientific (Singapore) Serjeant S., Lacy M., Rawlings S., King L.J., Clements D.1., 1995, MNRAS 276, L31 Soifer B.T., Helou G., Lonsdale C.J., Neugebauer G., Hacking G., Houck 1.R., Low F.J.,

Rice W., Rowan-Robinson M., 1984, ApJ 283, L1


Solomon P.M., Downes D. and Radford S.J.E., 1992, ApJ 398, L29 TllIlan T.X., and Condon J.J., 1987, ApJ 322, L9 Valentijn E., 1990, Nat 346, 153 Van Ojik R. , Riitt,gering H.J.A., Miley G.K., Bremer M.N., Macchetto F., Chambers

K.C., 1994, AA 289, 54 Windhorst R.A., Dressler A., Koo D.C., 1987, in Observational Cosmology, eds A.Hewit,

G.Burbidge, L.Z.Fang, Reidel (Dordrecht), p.573 Yamada T" 1994, ApJ 423, L27 Yoshii Y., Takahara F., 1988, ApJ 326, 1

THE EUROPEAN LARGE AREA ISO SURVEY: ELAIS

S.J. OLIVER Imperial College of Science Technology and Medicine Astrophysics Group Blackett Laboratory Prince Consort Rd. London SW21BZ [email protected]

Abstract. I describe a European collaborative project to survey'" 20 square degrees of the sky at 15j.lm and 90j.lm with ISO. This is the largest open time project being undertaken by ISO. The depth and areal coverage were designed to complement the various Guaranteed Time surveys. The main science thrust is to explore star formation in galaxies to a much higher redshift than was probed by IRAS. We expect to detect around 8000 extragalactic objects and a similar number of Galactic sources. The maps and source catalogues will represent a major legacy from ISO, inspiring follow up work for many years to come.

1. Introduction

The Infrared Space Observatory (ISO) will be the only major infrared mission for the next decade. Although the satellite was principally designed as an observatory the case for devoting a substantial amount of the mission time to surveys was overwhelming.

The Infrared Astronomical Satellite (IRAS) had enormous success arising principally from its survey products (particularly the Point Source Catalog and the Faint Source Catalog). Perhaps most significant was the discovery of a whole new class of objects with enormously high far infrared luminosity [notably F10214+4724 (Rowan-Robinson et at., 1991) and P09104+4109 (Kleinmann et at., 1988)]. As well as discovering new

77

M. N. Bremer et al. (eds.). Cold Gas at High Redshijt. 77-83. © 1996 Kluwer Academic Publishers.

78 S.J. OLIVER

objects, IRAS demonstrated the benefit of selecting objects in the far infrared. This wave-band is not sensitive to dust obscuration which biases optically selected samples. The emission arises from thermally heated dust and thus complements studies of emission directly from star-light, gas, or AGN engines.

The sensitivity of ISO is orders of magnitude better than IRAS. Using it as a survey instrument will thus allow us to explore IRAS-like populations to higher redshift and possibly unveil new classes of objects or unexpected phenomena.

This paper outlines the open time survey which is a collaborative venture between fifteen European Institutes; the PI being M. Rowan-Robinson, Co-Is being: C. Cesarsky, L. Danese, A. Franceschini, R. Genzel, A. Lawrence, D. Lemke, R. McMahon, G. Miley, S. Oliver, J-L. Puget and B. Rocca-Volmerange. Many other people are also heavily involved.

2. Science Goals

While it is impossible to predict all the scientific benefits of such a large project, I outline some of the key issues that we hope to address. A major theme is the detection of high redshift galaxies.

2.1. EPOCH OF GALAXY FORMATION

The search for galaxies at high redshift to uncover the formation epoch is one of the holy grails of cosmology. The failure to detect high redshift objects in optical surveys, particularly using Lya, has two competing explanations. The first is that early galaxies contain a large dust component which obscures the optical emission. E.g. if elliptical galaxies underwent a massive burst of star-formation between 2 < z < 5, they would be observable in the far infrared since massive stars produce both dust and the UV to heat it, and may look like F10214 (Elbaz et at., 1992). Alternatively galaxies may have been formed by the assembly of constituents which are individually too faint to detect. This survey will provide a powerful discrimination between these two hypotheses, since we would detect optically obscured galaxies but not low luminosity proto-galaxies.

2.2. STAR FORMATION IN SPIRAL GALAXIES AT HIGH REDSHIFT

The main extra-galactic population detected by IRAS was galaxies with high rates of star formation. Their far infrared emission arises from dust heated by young stellar populations. These objects are now known to evolve with a strength comparable to AGN (Oliver et at., 1995). The distance to which these objects were visible by IRAS wa.s, however, insufficient to

THE EUROPEAN LARGE AREA ISO SURVEY: ELAIS 79

determine the nature of their evolution. The sensitivity of ISO will allow us to detect these objects at much higher redshifts and thus obtain greater understanding of the cosmological evolution of star formation.

2.3. ULTRA AND HYPER-LUMINOUS IR GALAXIES AT HIGH Z

IRAS uncovered a population with enormous far infrared luminosities, LFIR

> 1012 L8' This far infrared emission represents the bulk of the bolometric luminosity of these objects which is comparable to that of AGN. The local space density of these objects, however, exceeds that of optically selected AGN, implying this population is a more energetically significant component of the Universe. For these reasons this population has been carefully studied. The energy source in these objects is still disputed. While most of these objects appear to have an AGN, it is argued that star formation could provide most ofthe energy. Interestingly, most of these objects appear to be in interacting or merging systems, suggesting a triggering mechanism. Exploration of these objects at higher red shift will have particular significance for models of AGN Jgalaxy evolution.

2.4. EMISSION FROM DUSTY TORI AROUND AGN

Unified models of AGN suggest that the central engine is surrounded by a dusty torus. Optical properties are then dependent on the inclination angle of this torus. The far infrared emission from the torus will be less sensitive to the viewing angle. Thus a far infrared selected sample of AGN will be more uniform than an optically selected sample and the far infrared properties of these will place important constraints on unification schemes. AGN are known to be strongly evolving and this sample will tell us about the evolution of the tori. Also, we will be able to detect dust emission from tori in 'face-on' AGN which would not be detected in the optical.

2.5. DUST IN NORMAL GALAXIES TO COSMOLOGICAL DISTANCES

Faint optical redshift surveys find surprisingly few galaxies beyond z = 0.5. One possible explanation for this is a dust fraction that increases with z. Emission from the cool interstellar 'cirrus' dust in normal galaxies will be detectable in our survey to much greater distances than were accessible with IRAS, so we will be able to examine the dust content to higher z.

2.6. CIRCUMSTELLAR DUST EMISSION FROM GALACTIC HALO STARS

The deep stellar number counts provided by this survey will be relatively unaffected by Galactic extinction and may provide, amongst other things,

80 S.J. OLIVER

improved estimates of the halo/disk population ratios.

2.7. NEW CLASSES OF GALACTIC AND EXTRA-GALACTIC OBJECTS

F10214 was at the limit of IRAS sensitivity and new classes of objects may well be discovered at the limit of the ISO sensitivity. The lensing phenomenon which made F10214 detectable by IRAS may become more prevalent at fainter fluxes, increasing the proportion of interesting objects. Current predictions suggest we would not expect to detect Galactic Brown Dwarfs, but unexpected Galactic objects may be discovered.

2.8. CLUSTERING PROPERTIES

The volume of this survey is comparable to that surveyed by the entire IRAS Point Source Catalog. The median red shift will be much higher. We will thus be in a position to examine the evolution of clustering strength, giving perhaps the most direct test of the gravitational instability picture of structure formation.

3. Survey Definition

As with any time-constrained survey we had to balance factors such as depth, wavelength and areal coverage. To complement Guaranteed Time deep ISO CAM surveys (Franceschini et at., 1995) we decided to sacrifice depth at the shorter wavelength for increased areal coverage. This section describes the rationale behind the choices we made for: wavelengths, depths and areas.

3.1. WAVELENGTH AND SENSITIVITIES

We initially proposed to survey at three wavelengths to give useful colour information over a long wavelength baseline, but were required by the OTAC to restrict ourselves to two. At the longer ISO wavelengths we pick up star forming galaxies. Consideration of the SED of these galaxies together with the capabilities of the ISO PHOT instrument suggested that the optimal sensitivity to these objects would be obtained using the ClOO detector with 90j.tm filter. At shorter wavelengths ISO is more sensitive to AGN emission. Consideration of the ISO CAM sensitivities, AGN SEDs and avoidance of frequencies in atmospheric windows lead us to select the CAM LW-3 filter centred at 15j.tm .

The limited resolution but high sensitivity of ISO at long wavelengths means that the Galactic Cirrus confusion limit is reached with very short integration times. This confusion limit thus defined our PH aT integration.


TABLE 1. Survey parameters for a single raster

Instrument CAM PHOT

Filter LW-3 90

>'o/p,m 15 ± 3 95.1 ± 26 Detector LW-l SiC +Ga) CI00 Ge:Ga AOT CAMOI PHOT22 Pixel Size 5.6" 43.5" Pixels/frame 32 x 32 3x3 Frame size 180" 135" 8x,8y 90",180" 130" ,130" Raster Points 28 x 14 20 x 20 Raster size ( 42')2 ( 43.3')2

50" Sensitivity 1.7 mJy 15 mJy

We decided to use a similar total observation time for both instruments. Table 1 summarises the two observing modes used.

3.2. AREAS

The allocated observing time allowed 37 rasters as described above. The choice of where to distribute these on the sky was governed by a number of factors. Firstly we decided not to group these all in a single contiguous region of the sky. Had we done so we may have had difficulty distinguishing evolutionary effects from local large scale structures. Distributing the survey areas across the sky also has advantages for follow up work. Cirrus confusion is a particular problem, so we selected regions with low IRAS lOOj.lm intensities (l1O0 < 1.5MJy/sr), using IRAS 100j.lm maps (RowanRobinson et ai., 1991 b). To avoid conflict with other ISO observations we further restricted ourselves to regions of high visibility over the mission lifetime (> 25%). To avoid unnecessarily high Zodiacal backgrounds we only selected regions with high Ecliptic latitudes (1,81 > 40°). Finally it was essential to avoid saturation of the CAM detectors so we had to avoid any bright IRAS 12j.lm sources. These requirements led us to selecting the four areas detailed in Table 2. A further 6 areas were selected as being of particular interest to warrant a single small (24' X 24') raster. These were chosen either because of existing survey data or because the field contained a high red shift object and were thus more likely to contain high redshift ISO sources. These 6 regions are also described in Table 2.

82 S.J. OLIVER

TABLE 2. Summary of Areas. The first four areas comprise the main survey made up from 43' x 43' rasters. One raster in N3 will be repeated. The final 6 areas are single smaller rasters 24' x 24'

Area Rasters N aminal Coordinates (1100 ) Visibility {3

J2000 /MJysr- 1 /%

N1 3 x 3 16h08m44 s +56°26'30" 1.2 98.0 73 N2 4x2-1 16h39"'34s +41°15'34" 1.1 58.7 62 N3 3 x 3 14h28m26s +32°25'13" 0.9 26.9 45

Sl 4 x 3 00h38m24s -43°32'02" 1.1 32.4 -43

Lock. 3 13h34m36s +37°54'36" 0.9 17.3 44 Sculptor 1 00h22m48 s -30°06'30" 1.3 27.5 -30

TX1436 14h36m43 s +15°44'13" 1.7 22.2 29 4C24.28 1 13h48m15s +24°15'50" 1.4 16.8 33

VLA 8 1 17h 14"'14s +50°15'24" 2.0 99.8 73

Phoenix 1 01 h 13"'13s -45°14'07" 1.4 36

4. Expectations

IRAS luminosity functions and model SED of star-bursts, AGN and normal galaxies and F10214 like objects together with simple pure luminosity evolution models have been used to predict the number of extra-galactic object we expect to see (Pearson & Rowan-Robinson, 1996). This simple model predicts 5000 star-bursts (20% detected in both bands, 30% z> 1), 650 AGN (5% detected in both bands, 23% z > 1), 2300 normal galaxies (30% detected in both bands) and 4 F10214 like objects. Models including a dusty phase in elliptical galaxy formation would predict higher numbers. We would also expect of order 10000 stars.

5. Science Products

The products we will provide to the community are source catalogues together with catalogue associations and maps at both ISO wavelengths. We anticipate these will be available a year after the end of the ISO mission (i.e. May 1998). A WWW page will be on line in the near future to keep the community abreast of the progress of the survey, a link to this will be found on http://icstar5.ph.ic.ac.uk/

References

Elbaz, D. et al. {1992} Astr. Astrophys.,265, pp. L29-L32


Franceschini A., Cesarsky,C Rowan-Robinson, M., (1995) In 'Near-IR Sky Survey' San Maniato (Pisa) Memorie della Societa Astronomica Italiana (in press)

Kleinmann, S.G. et al. (1988) Astophys. J. 328, pp. 161-169 Oliver, S., et al. (1995), In Wide-Field Spectroscopy and the Distant Universe, Maddox,

S.J., Aragon-Salamanca, A. eds, Proceedings of the 35th Herstmonceux Conference, World Scientific. p. 264

Pearson, C., Rowan-Robinson, M. (1996) Mon. Not. R. Astro. Soc., (in press) Rowan-Robinson, M. et al. (1991) Nature, 351, pp. 719-721 Rowan-Robinson, M. et al. (1991) Mon. Not. R. Astra. Soc., 249, pp. 729-741

KECK OBSERVATIONS OF MICROJANSKY RADIO SOURCES

Hints to Galaxy Evolution

JAMES D. LOWENTHAL AND DAVID C. KOO

Lick Observatory Kerr Hall University of California Santa Cruz, CA 95064

Abstract. Millions of times fainter than classical double-lobed radio monsters, /-LJy sources cover the sky at densities approaching faint optical galaxy counts. To investigate their redshifts, star formation characteristics, optical morphologies, and evolution, we have used the 10-meter Keck telescope to obtain BRI broad-band images and optical spectra of a complete sample of /-LJy sources detected in three fields at the VLA. In one field, only 1 of 16 sources remains unidentified down to I '" 26.5, and in another field only 3 of 14 sources lack redshifts. The /-LJy sources appear to be dominated ('" 50% of the sample) by blue star-forming galaxies with median red shift z '" 0.5, with a large fraction of interacting and close-pair systems. About 25% are high-redshift QSOs, including one with strong Mg II absorption at Zabs = Zem = 1.8, and the remaining 25% show signs of both starformation-induced emission and absorption lines characteristic of evolved stellar populations ("S+A" galaxies). The colors of the sources are consistent with those of field galaxies, which are dominated by Sbc galaxies, though the /-LJy radio galaxies are at least 1 mag brighter than field galaxies. It appears that the typical /-LJy source is similar in most respects to M82, the canonical starburst galaxy, moved out to Z '" 0.5. Evolution of the sources, which could have been avoided only if Zmed < 0.1, now appears inevitable.

1. The Faintest Radio Galaxies

Just as number counts of optical galaxies and IRAS sources show excesses at faint levels that imply evolution, so too do counts of the faintest radio

85

M. N. Bremer et al. (eds.), Cold Gas at High Redshi/t, 85-90. © 1996 Kluwer Academic Publishers.

86 JAMES D. LOWENTHAL AND DAVID C. KOO

sources greatly exceed simple no-evolution models (d. Rowan-Robinson, this volume; also Condon 1989; Windhorst 1990; Wall 1994). Below 1 mJy, radio sources are dominated by blue galaxies generally identified with starbursts, similar to the ones dominating the IRAS Faint Source Survey, in contrast to the old red ellipticals that host the classical double-lobed sources at the bright end of the radio luminosity function (Benn et al. 1993; RowanRobinson et al. 1993). But what are the pJy sources? Are they a new population of low-luminosity local sources, are they just more distant versions of the sub-mJy galaxies, or do they trace activity of a different sort altogether? Only recently have optical identifications and redshifts become available for a handful of pJy sources (Hammer et al. 1995; Windhorst et al. 1995), but the samples are tiny and the results less than conclusive.

2. Observations

To understand the nature and evolution of the pJy sources and their relation to the more general family of galaxies, we have obtained deep optical images and spectra with the 10-meter Keck telescope and the Low Resolution Imaging Spectrograph (LRIS) of a complete sample of p.Jy sources discovered with the VLA. We studied three fields familiar to many of you from deep optical and radio studies over the years: Lynx.2 and SA68 from the Leiden-Berkeley Deep Survey (Windhorst et al. 1984) and a field studied with the VLA and HST in the course of the HST Medium Deep Survey (Windhorst et al. 1995 - the "Lilly field"). With 10" levels of 3, 60, and 2 pJy, the complete radio catalogs comprised 14, 9, and 16 sources, respectively; additional sources fell below the completeness level, usually 4.50".

U sing both multi-object slit masks and long-slit pointings, we obtained 14 new redshifts from the complete samples (as well as several more for objects in the incomplete samples), bringing the total numbers of redshifts in the Lynx.2, SA68, and Lilly fields to 11 (80%),2 (22%), and 10 (63%), respectively.

Optical identifications had been essentially completed for the SA68 and Lilly fields but only 64% completed for the Lynx field. With this in mind, we obtained deep B RI images of the same part of the Lynx field containing the objects we studied spectroscopically, and summed all the images to create a deep identification image. The 30" levels were B '" 26.2, R '" 26.9, and I '" 25.4 in 3" apertures. In most cases there is a fairly bright object within 3" of the VLA coordinates of the pJy source. In many cases the VLA coordinates are centered not on but near a close pair or group of galaxies that appear to be interacting, or in the outskirts of an extended disk. In two cases the VLA position corresponds to a faint bridge of emission between galaxies that make up a short chain of objects. In only two cases in the Lynx.2

KECK OBSERVATIONS OF MICROJANSKY RADIO SOURCES 87

complete sample is there no optical object visible within 3" of the VLA coordinates, and no obvious candidate nearby.

To test for contamination from field galaxies, we randomized the positions and repeated the identification excercize. The results indicate that at most about 15-20% of our identifications could be due to field galaxies unrelated to the j.LJy radio source - and as we will see below, there is probably significant overlap between faint field galaxies and j.LJy radio sources anyway, so in fact the identifications may be not "contaminated" at all.

3. What Are the j.LJy Radio Sources?

Morphologically, as noted above, many of the optically-identified sources correspond to galaxies in close pairs or groups, often with clear signs of interactions such as tidal tails; this is consistent with the predictions and preliminary results of Windhorst et al. (1993) and Windhorst (1995). Most of the sources show the strong H,B and [0 III] >..5007 emission lines characteristic of star formation; the source 16V36 at z = 0.408 in Lynx.2 (Fig. 1) is a good example. Several sources show only a single emission line that is most likely [0 II] >..3727.

About 25% of the sources also show Calcium H&K, Fe, and/or Balmer series absorption lines characteristic of evolved stellar populations. These appear to be similar to the "S+A" galaxies identified as mJy sources by Benn et al. (1993), although we should keep in mind that most spiral and starburst galaxies with any evolved underlying stellar population will also show such emission plus absorption spectra (d. Kennicutt 1992).

Only one spectrum has a featureless continuum that yields no red shift or spectral type.

Due to our short spectral coverage, we lack sufficient emission line data to discriminate between star formation and AGN as the cause of the emission. Hammer et al. (1995) have attempted this with their sample of j.LJy sources from the Canada France Redshift Survey (CFRS), and have concluded that AGN dominate the j.LJy sources. However, we emphasize that 40% of the j.LJy sources are extended in the radio at () >5" while only 20-30% are variable (Windhorst et al. 1995), which argues against their being AGN, and also that discriminating among emission processes with even the line fluxes used by Hammer et al. is difficult and less than reliable.

Furthermore, the CFRS group finds a significant number of elliptical type galaxies in their j.LJy sample; we find none. As Hammer et al. point out, their field may be affected by a large structure at z '" 1, which could weight the tally heavily towards ellipticals.

Approximately 25% of the sample consists of high-redshift (z > 1.5) QSOs. One of these, at z = 1.8, shows strong Mg II and Fe II absorption


7000 7600 6000 6600 9000

leV2S 20=0.729

9600

6500 7000 7500 6000 8500

BOO

600

400

200

6000 6500 7000 7500 8000 8600

Figure 1. Samples of B + R + I-band images and spectra of three JLJy sources from the Lynx.2 field. Each image is 30"on a side, and centered on the source's VLA coordinates. 16V13 is a QSO at Z = 1.835 with extraordinarily strong MgII absorption lines at Zabs = Zem. The source is extended in both the radio (8", 13") and the optical. 16V25 at Z = 0.729 has intrinsic radio and optical properties remarkably similar to those of M82. 16V36 is associated with an interacting pair of galaxies; note the tidal tails.

at the emission redshift of the QSO (see Fig. 1). In fact, the absorption is so strong - > 5 A in each of the Mg II lines - that we could find only one example of stronger absorption in the literature. The QSO is extended in both the radio and the optical, where it is surrounded by an asymmetrical clump of faint emission.

The median redshift of the sample is z = 0.5, with no chance of bringing it higher than z '" 0.7 even if all the missing redshifts are filled in at z > 1. This is in contrast to the suggestion by Hammer et al. that 40% of the J-LJy sources are at z > 1.

It has been shown (Wall 1986) that only if the median redshift of the J-LJy sources were less than 0.1 could we avoid the conclusion from the number counts that the sources are an evolving population. Our measured median z '" 0.5 forces us to infer that evolution has been detected. Perhaps we

KECK OBSERVATIONS OF MICROJANSKY RADIO SOURCES 89

should not be surprised by this, given the strong evolution derived from IRAS Faint Source Survey counts (e.g. Lonsdale et al. 1990), which are presumably sampling a parent source population similar or identical to the one seen by the VLA (except for the VLA's much more sensitive detection limits, which therefore probes much deeper into cosmological space).

4. Field Galaxies and p,J y Radio Sources

We performed photometry on field galaxies in the Lynx.2 field to compare to the p,Jy sources. The B - Rand R - I colors of the radio-selected sources are consistent with those of the field galaxies: B - R rv 1 and R - I rv 0.7. However, the p,Jy sources, with Imed rv 20, are 1-2 magnitudes brighter than the typical field galaxy down to our photometric limits. We note that the field galaxy population is generally dominated by moderate-luminosity late- type spiral galaxies (e.g. Driver et al. 1995).

With redshifts, of course, we can calculate intrinsic colors and luminosities of our sample of galaxies. These also turn out to be consistent with Sbc galaxies over the observed range of redshifts of our sample, with a typicalluminosity close to L* (ME rv -20.1 for h = 0.75) - again, somewhat brighter than a typical field galaxy. The median radio power is Pmed rv 1022W HZ-I, similar to M82, the prototypical starburst galaxy. This echoes the results of Windhorst et al. (1995).

It is interesting that the surface density of radio counts extrapolated down to 300 nJy matches that of optical field galaxies to V < 28 (1.5 - 3 105 deg-2 ; Windhorst et al. 1993). Unless some radically new population of bizarre sources has appeared in one or the other spectral band, we should expect the bulk of sources to appear in both ultra-faint radio and optical surveys on an almost one-to-one basis. Supposedly, star formation provides the common link, producing faint blue galaxies that dominate the optical counts and thermal bremsstrahlung and non-thermal synchrotron radiation (via supernovae) to power the radio emission. There may also be strong implications for merging models of field galaxy evolution (e.g. Broadhurst et al. 1992), since it is mergers that drive the starbursts that figure so prominently in the IRAS and p,Jy samples (Rowan-Robinson et al. 1993).

Carrying this possible connection between optical field galaxies and J-LJy sources to its logical conclusion, we see that evolution observed in one spectral band implies evolution in the other. Faint blue galaxies have often been thought to occur in excess of no-evolution models (see Koo & Kron 1992 for a review), though uncertainties pertaining mostly to the faint end of the assumed local luminosity function remain. New field galaxy redshift surveys with the Keck telescope and deep high-resolution surface brightness, size, morphology, and number count studies with HST should resolve many of


these outstanding issues in the near future.

5. The Near Future

If the VLA is upgraded (see other pleas elsewhere in this proceeding!), those nJy flux levels will be attainable in reasonable integration times, allowing us to see down the radio luminosity function to dwarf galaxies at moderate redshifts forming stars at fairly modest rates.

Furthermore, upcoming observations with Keck in the near-IR of the current J-LJy sample may reveal additional clues to the nature of the sources through morphology and broad-band colors, which can tell us about underlying stellar populations. We also hope to obtain additional hints to the importance of mergers and interactions in this class of faint radio galaxies that is so different from the monsters dominating the bright end of the radio scale.

References

Benn, C.R., Rowan-Robinson, M., McMahon, R.G., Broadhurst, T.J. & Lawrence, A. 1993, MNRAS, 263, 98

Broadhurst, T.J., Ellis, R.S. & Glazebrook, K. 1992, Nature, 355, 55 Condon, J.J. 1989, Ap.J., 338, 13 Driver, S.P., Windhorst, R.A., Ostrander, E.J., Keel, W.C., Griffiths, R.E. & Ratnatunga,

K.U. 1995, Ap.J.Lett., 449L, 23 Hammer, F., Crampton, D., Lilly, S.J., Le Fevre, 0., & Kenet, T. 1995, MNRAS, 276,

1085 Kennicutt, R.C.J. 1992, Ap.J.Supp., 79, 255 Koo, D.C. & Kron, R.G. 1992, Ann. Rev. Astron. Astroph., 30, 613 Lonsdale, C.J., Hacking, P.B., Conrow, T.P. & Rowan-Robinson, M. 1990, ApJ, 358,60 Rowan-Robinson, M., Benn, C.R., Lawrence, A., McMahon, R.G. & Broadhurst, T.J.

1993, MNRAS, 263, 123 Wall, J.V. 1990, Austr.J.Phys., 47,625 Wall, J.V., Benn, C.R., Grueff, G. & Vigotti, M. 1986, in Highlights of Astronomy, Vol.

7 (Dordrecht: D. Reidel), p. 345 Windhorst, R.A., Fomalont, E.B., Kellermann, K.I., Partridge, R.B., Richards, E.,

Franklin, B.E., Pascarelle, S.M., & Griffiths, R.E. 1995, Nature, 375, 471 Windhorst, R.A., Mathis, D. & Neuschaefer, L. 1990, ill Evolution of the Universe of

Galaxies; Proceedings of the Edwin Hubble Centennwl Symposium (Astr. Soc. Pac.), p.389

Windhorst, R.A., van Heerde, B.M. & Katgert, P. 1984, AASupp, 58, 1 Windhorst, R.A., Fomalont, E.B., Partridge, R.B., & Lowenthal, J.D. 1993, Ap.J., 405,

498

THEORETICAL ASPECTS

SMALL SCALE STRUCTURE AND HIGH REDSHIFT HI

DAVID H. WEINBERG Ohio State University Dept. of Astronomy 174 W. 18th Ave. Columbus, OH 43210 USA

LARS HERNQUIST U. C. Santa Cruz Dept. of Astronomy Santa Cruz, CA 95064 USA

NEAL S. KATZ University of Washin9ton Dept. of Astronomy Seattle, WA 98195 USA

AND

JORDI MIRALDA-ESCUD:E Institute for Advanced Study Olden Lane Princeton, NJ 08540 USA

1. Introduction

Galaxy red shift surveys reveal the presence of large scale structure in the local universe, a network of sheets and filaments interlaced with voids and tunnels. Zel'dovich (1970) showed that gravitational instability in an ex-

93

M. N. Bremer etal. (eds.), Cold Gas at High RedshiJt, 93-107. © 1996 Kluwer Academic Publishers.

94 DAVID H. WEINBERG ET AL.

panding universe can create such structures from generic random initial conditions. Zeldovich's analysis was originally used to describe the first collapse in "top-down" scenarios like the hot dark matter model, which have a cutoff in the primordial fluctuation power spectrum at small scales. The theories of structure formation that are most popular today have no intrinsic cutoff in the power spectrum. Structures in such a theory grow by hierarchical clustering -low mass perturbations collapse early, then merge into progressively larger objects.

The Zel'dovich analysis does not apply directly to hierarchical clustering models, but numerical and analytic studies show that they tend to develop the same types of structure (e.g., Shandarin and Zel'dovich, 1989; Weinberg and Gunn, 1990; Melott and Shandarin, 1993). The smooth "pancakes" of the top-down theory are replaced by "second generation pancakes" that are themselves made up of smaller clumps. The characteristic scale of voids, sheets, and filaments grows with time, as larger scales reach the non-linear regime. In a hierarchical scenario, one naturally expects the high red shift universe to contain "small scale structure" that is qualitatively similar to today's large scale structure, but reduced in size by a factor that depends on the specifics of the cosmological model.

Observations of absorption and emission by neutral hydrogen can trace this small scale structure over a wide range of redshifts. Such observations probe the evolution of the intergalactic medium and the condensation of gas into galaxies, filling in the gap between cosmic microwave background anisotropies and maps of present day structure. On the theoretical side, an important recent development is the use of hydrodynamic simulations to work out the predictions of a priori cosmological models for observable high redshift structure. This talk is based primarily on the results of a numerical simulation ofthe cold dark matter (CDM) model using TreeSPH, a combined N-body Ihydrodynamics code. The simulation methods and some applications to galaxy formation are discussed in Katz et al. (1995a), and some early results on Lya absorbers are described in Katz et al. (1995b, hereafter KWHM) and Hernquist et al. (1995, hereafter HKWM).

2. Lya Absorption in the CDM Model

Figure 1 shows the distribution of gas (SPH) particles at z = 2 in a simulation of the "standard" CDM model, with parameters n = 1, nb = 0.05, and h == HoI100 km s-1 Mpc1 = 0.5. The simulation volume is a periodic cube of comoving size 22.222 Mpc, so its physical size at z = 2 is 7.4 Mpc, with a corresponding Hubble flow of 1925 km S-1. There are 643 SPH particles to represent the baryon component and 643 collisionless particles (not shown) to represent the cold dark matter component; individual particle masses are

6

4

2

o o

SMALL SCALE STRUCTURE AND HIGH REDSHIFT HI 95

2 4 6

Figure 1. The distribution of gas particles in a hydrodynamic simulation of the CDM model, at z = 2. The simulation volume is a cube 22.222 comoving Mpc on a side (for h = 0.5), making the physical size at this redshift 7.4 Mpc.

1.5 X lOs M0 and 2.8 X 109 M 0 , respectively. The simulation incorporates radiative cooling for a gas of primordial composition (76% hydrogen, 24% helium) in ionization equilibrium with an ultraviolet (UV) radiation background of intensity J(v) = 10-22 F(z)(vL/v) ergs-1 cm-2 sr-1 Hz-I, where VL is the Lyman limit frequency and F(z) = 0 for z > 6, 4/(1 + z) for 6 > z > 3, and 1 for 3 > z > 2.

We normalize the CDM power spectrum so that, if it were linearly extrapolated to z = 0, the rms mass fluctuation in spheres of radius 16 Mpc would be (1Sh-1Mpc = 0.7. This normalization is roughly that required to


match the observed abundance of massive galaxy clusters (White et ai., 1993). However, with this normalization and the other parameters we have adopted, the CDM model predicts large scale microwave background fluctuations nearly a factor of two lower than those observed by COBE (Bunn et al., 1995). An n = 1 model dominated by cold dark matter must involve some additional complication (e.g. a "tilted" or "broken" primeval power spectrum, a lower Hubble constant, an admixture of massive neutrinos) in order to account for COBE fluctuations and galaxy clusters simultaneously. The version of CDM that we have simulated might be a useful approximation to such a model on the scales considered here. We plan to examine alternative scenarios - in particular low-n CDM models - in the near future.

The spatial structure in Figure 1 has the filamentary character seen in typical simulations (and observations) of large scale structure. However, the size of the structures is relatively small - the largest low density regions, for instance, have a diameter of 5 - 10 comoving Mpc. This scale would be somewhat larger if the simulation box were itself large enough to accommodate longer wavelength modes, but primarily the reduced scale of structures reflects the lower amplitude of fluctuations at z = 2 relative to z = O. Only at later times do larger scale fluctuations reach the amplitude required to produce non-linear gravitational collapse. At the level of detail discernible in Figure 1, the dark matter distribution would look very similar to the depicted gas distribution.

Figure 2 shows the distribution of gas in the density-temperature plane. Each point represents a single SPH particle, and histograms at the edges of the Figure show marginal distributions. This representation reveals four main components. One is low density, low temperature gas, which occupies a well defined locus along which adiabatic cooling balances heating by photoionization. A second is overdense, shock heated gas; at this redshift, 10% of the gas has T > 105 K and 5% has T > 106 K. The third component consists of very overdense gas that has radiatively cooled to the equilibrium temperature, T ~ 104 K, where heat input and radiative cooling balance. The fourth component is warm gas at moderate overdensity. While this category is to some extent a "catch-all" for gas that does not fit into one of the other, more distinct components, it accounts for an appreciable fraction of the baryonic mass.

Figure 3 shows the spatial distribution of the gas in different regimes of density and temperature. Gas with T < 30,000 K and overdensity p / p < 1000 (upper left panel) mostly occupies the low density regions, though hints of the filaments in Figure 1 can be seen here as well. The filaments stand out dramatically in the warm gas component, with 30,000 K < T < 106 K (upper right panel). This temperature cut selects gas that has been


0.5

6

E-bI) 4 o -

2

OL...L-......................................................................................................... ........ -2 o 2 4 6

log (p/Pb)

8 0.5

Figure 2. The distribution of gas in the density-temperature plane at z = 2. Each point represents a single SPH particle; temperatures are in degrees Kelvin and densities are scaled to the mean baryon density. Histograms show the I-d marginal distributions, i.e. the fraction of particles in each decade of density and of temperature.

heated by adiabatic compression and mild shocks as it falls into moderate overdensity structures. The hottest gas (T > 106 K, lower left panel) is confined to fully virialized dark matter potential wells, and its spatial distribution is more clumpy. The gas with T < 30, 000 K and p / fi > 1000 (lower right panel) occupies radiatively cooled knots inside these hot gas halos. The larger halos may contain several such knots. The most massive knots contain several hundred particles (merged into a single extended dot at the resolution of Figure 3), while the least massive, which are generally the ones that have started to cool and condense most recently, contain only a handful of cold gas particles. The gravitational softening of the simulation, 7 kpc at z = 2, prevents us from resolving the detailed internal structure of these knots, but the physical conditions imply that they are likely to fragment and form stars. It is plausible to identify these knots as young - in some cases just forming - galaxies.

Knowing the density and temperature of each gas particle and the intensity of the model UV background, we can compute the corresponding neutral hydrogen fractions assuming ionization equilibrium. Figure 4 shows


pIli < 1000, T < 30,000 K

6 • 4

.'~ ..

-.. -. 0

~.~ ,. . .. jt ~

2 ~.> ~

t:'4r ~ :t

0 '" 0 2 4 6

30,000 K < T < 10sK .....,=,......,..,.-,......,,,..,... ..........

6

4

2

2 6

pIp:> 1000, T < 30,000 K r " , I. I

'", ~ .,.' ~ ; l ~ ~ '> . .

o " """, ~: ~><jII +~\ '<~ ,~, :: ~". ~ -~<~ :{ ~~<:: ~

..... 'l",

. . 2 ~ , .... ,

o " o

I

4:

, o. . ' ....

I

6

. ,'-

Figure 3. The spatial distribution of gas in different regimes of density and temperature, as indicated above each panel.

a 2-d map of the neutral hydrogen column density projected through the simulation cube. There is a close correspondence between the prominent structures in this map and the gas distributions in the right hand panels of Figure 3. The cold gas knots nearly always produce absorption at a column density of NHI = 1017 cm-2 or greater, features that appear white in the grey scale representation of Figure 4 (these regions are corrected for self-shielding using a procedure described in KWHM). The warm gas filaments produce the extended, lacy structures at lower column density. It is important to note that the faintest visible structures in Figure 4 have NHI ~ 1014.5 cm-2, so much of the absorption at column densities typical


Figure 4. A map of the projected neutral hydrogen column density in the simulation at z = 2. At this redshift in the n = 1 cosmology, the depth of the 7.4 Mpc simulation box would be ctl.z = 1924 kms-1 and the angular size would be 15.1 arc-minutes. In this representation, the saturated (white) regions have NHI ~ 1016 .5 cm-2 • and the faintest visible structures have NHI '" 1014 .5 cm-2 • Even regions that are black in this map can give rise to absorption at column densities typical of the Lya forest.

of the Lya forest arises in regions that are black in this Figure.

Figure 5 shows artificial QSO absorption spectra along four randomly chosen lines of sight through the simulation box at z = 2. In each panel, the solid line shows the transmission T = e-r , where T is the Lya optical depth. The dashed line shows a spectrum along a line of sight 100 kpc away from this primary spectrum. The dotted line shows a spectrum at 300 kpc separation (200 kpc from the dashed spectrum). Many absorption features appear in both of the first two spectra, and there are significant matches

100

a

DAVID H. WEINBERG ET AL.

500 1000 1500 v (km/sec)

a 500 1000 1500 v (km/sec)

Figure 5. Examples of artificial spectra at z = 2. Solid lines show transmission against velocity along four random lines of sight. At this redshift, the physical size of the periodic simulation box is 7.41 Mpc, corresponding to a Hubble flow of 1924.5 km/s. Dashed and dotted lines show spectra along lines of sight displaced arbitrarily from that of the primary spectrum by physical separations of 100 kpc and 300 kpc, respectively. The translation from velocity v to wavelength ,\ is ,\ = 1216 x (1 + z) x (1 + vic) A, where z = 2.

-10 ,.......,

S (J

'-' -15 53 z

'0 N '0

~ -20 N '0

tlD o --25

14 16 18 20 22

Figure 6. Distribution of neutral hydrogen column densities. The solid line shows the simulation results at z = 2. Points with error bars are taken from the observational compilation by Petitjean et al. (1993).

even for lines of sight separated by 300 kpc, though these are often accompanied by substantial changes in the features' depth or shape. Figure 4 shows that the low column density absorbing structures are extended and coherent, so the correlation of features along neighboring lines of sight is


not surprising. Qualitatively, it appears that our results can account for the large coherence scale found in absorption studies of QSO pairs (e.g. Bechtold et al. 1994; Dinshaw et al. 1994, 1995), though quantitative tests (e.g. Charlton et al. 1995) are needed to assess the agreement or lack of agreement with recent observations.

Figure 6 displays the distribution of neutral hydrogen column densities, d2 N / dN HI dz, at z = 2. The procedure adopted to identify lines at column densities NHI < 1015.5 cm-2 is described in HKWM. Above this column density, we determine the distribution by measuring the fractional area above each column density in the projected HI map of Figure 4. High column density lines are rare enough that the absorption along any line of sight that has NHI > 1015.5 cm-2 through our 7.4 Mpc box is always dominated by a single absorber. Observational data and error bars in Figure 6 are taken from Table 2 of Petitjean et al. (1993). There is a significant (factor of ten) discrepancy with the Petitjean et al. data for column densities near 1017 cm -2, which could reflect either a failure of standard CDM or the presence in the real universe of an additional population of Lyman-limit systems that are not resolved by the simulation. Nonetheless, given that CDM is an a priori theoretical model that was not "designed" or adjusted to fit these observations, the overall level of agreement across eight orders of magnitude in neutral hydrogen column density is rather remarkable. Other analyses of absorption in this simulation are presented by HKWM and KWHM.

Cen et al. (1994), who were the first to use these sorts of simulations to model the Lya forest, report similar agreement between observations and a 10w-11 CDM model with a cosmological constant. Zhang et al. (1995) find similar agreement for an 11 = 1 CDM model with a higher normalization (U8h-1Mpc = 1) than used here. The qualitative success of three different models (and numerical methods) suggests that the Lya forest arises naturally, and at least somewhat generically, in a hierarchical theory of structure formation with a photoionizing background. The comparison between simulations and the extraordinary data emerging from high resolution QSO spectra can clearly be carried out much more carefully than has been done so far. There is every reason to hope that detailed comparisons will reveal discrepancies that restrict the pool of acceptable theoretical models. For now, the agreement between simulated and observed line populations suggests that the simulation described here is worth taking seriously as a realistic general picture for the origin of Lya absorbers.

In this simulation, the structures that produce low column density absorption (NHI '" 1013 - 1015cm-2 ) are physically diverse: they include filaments of warm gas, caustics in frequency space produced by converging velocity flows (McGill, 1990), high density halos of hot, collisionally ion-


ized gas, layers of cool gas sandwiched between shocks (Cen et ai., 1994), and modest local undulations in undistinguished regions of the intergalactic medium. Temperatures of the absorbing gas range from below 104 K to above 106 K. The "typical" low column density absorbers - to the extent that we can identify such a class - are flattened structures of rather low overdensity (pi 11 f'V 1 - 10), and their line widths are often set by peculiar motions or Hubble flow rather than thermal broadening. Because of their low overdensities, most of the absorbers are far from dynamical or thermal equilibrium, and many are still expanding with residual Hubble flow. The simulation reveals a smoothly fluctuating intergalactic medium, with no sharp distinction between "background" and "Lya clouds". Indeed, in this picture one might say that the Lya forest is the absorption by diffuse intergalactic hydrogen known as the Gunn-Peterson (1965) effect. There is also absorption by low density gas outside of the "lines," but it is relatively weak. At higher redshifts, increasing neutral fractions make the absorbing gas more opaque, and the distinction between lines and background becomes even harder to draw.

A comparison between Figures 3 and 4 shows a clear association between high column density absorbers and the knots of radiatively cooled gas that represent forming galaxies. Damped Lya absorption (NHI 2:: 1020.2 cm-2 )

occurs along lines of sight that pass through the denser, more massive protogalaxies. The column density correlates inversely with the projected distance from the protogalaxy center, and the maximum projected separation that yields damped absorption is about 20 kpc (see KWHM). Lyman limit absorption (1017 cm-2 ~ NHI ~ 1020.2 cm-2 ) occurs on lines of sight that pass either through the outer parts (20-100 kpc projected separation) of the more massive protogalaxies or near the centers of younger, lower density systems. By definition the stronger Lyman limit systems are selfshielded against the ionizing background, so they are mainly neutral, in constrast to the Lya forest systems, which usually have neutral fractions of 10-6 - 10-4 •

3. Prospects for 21cm Observations

Figure 4 can be regarded as a (highly) idealized 21cm observation of a CDM universe at z = 2, covering a region 15.1 arc-minutes on a side and 1924 km s-1 in velocity. The contributions of Ingram and Braun to these proceedings show examples of what the Square-Kilometer Array (SKA) might see if it observed a universe like this, with realistic assumptions about resolution and noise. Here we will stick to more generic comments, inspired by the simulation but not directly tied to it. Some of the issues that affect the observability of high redshift HI in a variety of cosmological models are


discussed by Scott and Rees (1990) and references therein. In a typical hierarchical model, the "re" combination of hydrogen at

z = 1100 is followed by an era of linear growth, with the density contrast of fluctuations 181 ~ 1 on all cosmologically relevant scales. To the extent that pressure and Compton drag against the microwave background can be ignored, density contrasts grow in proportion to the expansion factor, 6' ex: a ex: t 2/ 3 • Once the strongest fluctuations (which occur on small scales) reach 6' '" 1, they collapse, eventually giving rise to luminous objects that reionize the remaining diffuse gas. It is not at all clear what objects actually cause reionization (globular clusters? quasars? supermassive stars? dwarf galaxies?), but the visibility of high red shift quasars at wavelengths shorter than Lya implies that the reionization occurred before z ~ 5. It could plausibly have happened much earlier (Tegmark et al., 1994; Fukugita and Kawasaki, 1994), though the detection of microwave background anisotropies at degree scales suggests that it did not occur at z much above 50 (Scott et al., 1995).

Reionization heats the diffuse medium to T '" 104 K, introducing a Jeans mass - thermal pressure prevents baryons from collapsing into objects with circular velocities below about 35 km s-l(Quinn et al., 1995; Thoul and Weinberg, 1995). Depending on the relative timing of reionization and fluctuation growth, the universe may enter a quiet phase during which little collapse occurs. Eventually, however, fluctuations larger than the Jeans mass reach the nonlinear regime, and the condensation of gas into protogalaxies and consequent star formation can begin in earnest. As time goes on and larger scale fluctuations become nonlinear, galaxies pull themselves into groups, clusters, and superclusters.

Reionization may have occurred at a red shift beyond the reach of current radio telescopes. However, it could have occurred at a redshift only slightly greater than 5, since at z = 5 the density of quasars is plummeting and the opacity of the diffuse medium is rising. In this case, there might be interesting observables at 5 < z < 8. Structure should be present on Mpc scales, and much of the gas should be cool enough to have a high neutral fraction. More work, including analysis of simulations, is needed to investigate whether the predicted structure is bright enough to detect with existing or plausible future instruments. If reionization takes place by the growth of "Stromgren spheres" around quasars or other rare objects, then there might be strong structure in the neutral hydrogen even where the underlying gas distribution is fairly uniform.

The prospects for HI emission searches above z = 5 are intriguing but highly uncertain. Below this redshift, observations give more guidance about what to expect. In particular, we know that the universe is reionized, and the statistics of Lya absorption give an estimate of the neutral


hydrogen density parameter OHl.

In the simulation of §2, most of the gas at z rv 2 - 4 is either in the low density, photoionized medium (roughly speaking, the Lya forest), or in the high density, hot, collisionally ionized medium. However, nearly all of the neutral gas is in high density, radiatively cooled objects - Lyman limit and damped Lya absorbers. This theoretical result is consistent with observational inferences, which show that the value of OHl is dominated by contributions from the highest column density systems.

Single objects containing 1014 Me of HI do not appear in the simulation, and they seem unlikely on generic grounds. The collapse of an object with 1014 Me of baryons would shock-heat and collision ally ionize the gas, and before collapse the gas would be low density and photoionized. One could in principle have an object in which 1014 Me of baryons collapsed and cooled - a sort of super damped Lya system - but if such objects were common we would expect to see many galaxies with 1014 Me of stars today.

The simulated observations of Ingram and Braun in these proceedings show that it is difficult to detect the faint caustics seen in Figure 4 even with an instrument as ambitious as the SKAI. What the SKAI can detect rather easily are the damped Lya systems, which are more numerous than present day, L * galaxies by a factor of several. A single pointing covers a small angular field, but it is sensitive to a gigantic range in redshift. A program of multiple long exposures could thus provide a "galaxy" (damped Lya) redshift survey over a large cosmological volume at high redshift. This would be an extraordinary feat, and observing the history of galaxy clustering would doubtless teach us a great deal about the underlying physics of structure formation.

At lower sensitivity, the first objects to be detected in 21cm emission at high red shift are likely to be clusters of damped Lya systems, the high red shift analogs oftoday's rich galaxy clusters. The statistics of these can in principle be calculated from simulations, but very large simulation volumes are needed to predict the abundance of the rarest, most massive systems. We will therefore attempt a more phenomenological calculation, based on scaling the abundance of present galaxy clusters back to high Z.

Bahcall and Cen (1993) find that the abundance of observed rich clusters with total mass M or greater can be described by the cumulative mass function

n(> M) = n~(M/M~)-le-MIMo, with Mn = 1.8 x 1014h-1 Me and no = 4 X 1O-5h3 Mpc3 • Suppose that this mass function has emerged from a hierarchical model of structure formation in which the power spectrum on the scales of interest can be approximated as P( k) oc kn. The characteristic mass Mn corresponds to the scale on which the rms linear theory mass fluctuations have some value (1* rv 1/2,


so that only the rare ('" 2 - 30') fluctuations on this scale have collapsed to form virialized clusters. At high redshift we might expect a mass function of similar form, but the characteristic mass M; will correspond to the (smaller) scale on which the rms fluctuation is 0'* at this redshift. For a k n

power spectrum,

O'(M, z) = 0'* (MjM;)-(3+n )/6 D(z),

where a useful approximation to the growth factor D(z) is (Shandarin et ai., 1983)

D(z) ~ (1 + 2.5!!oz )-1 1 + 1.5!!0

The condition 0'( M;, z) = 0'* implies that the characteristic mass at redshift z is M; = MoD 6/(3+n). The abundance n;, in co moving units, can be determl·ned from the condition n* M* = n* M* ---'- n* = n* D-6/(3+n) zz 00----"- z 0 , since objects of mass M > M* contain a constant fraction of the total mass of the universe in this sort of scale free model.

For our purposes, we are interested in a cluster's HI mass rather than its total mass. To go from one to the other, we can make the adventurous assumption that the ratio of MHI to Mtot is the same as the universal ratio !!HI(Z)j!!(z). This assumption is almost certainly incorrect today because the galaxies in clusters tend to be gas poor, early types. It may be more plausible at high redshift, when galaxies are primarily gaseous; indeed, the error may be in the opposite sense (underestimating MHI instead of overestimating) if galaxies at high z formed preferentially in the densest regions. Putting all of this together, we arrive at an expression for the cumulative, comoving number density of clusters with HI mass greater than MHI:

F == D-6/(3+n ) , R = !!HI(Z) _ !!HI(Z) (1 + !!oz) - !!( z) - !!o (1 + z) .

There are many ways that this calculation could depart from reality, but it illustrates how the power spectrum, the cosmological model, and the history of the neutral gas density might interact in determining the abundance of observable high redshift objects. Unfortunately, the numbers that it yields are not particularly encouraging because even today the HI mass of an M* object is only several x 101lh-1 M~j, and at higher redshift the tendency of galaxies to be more gas rich is countered by the lower masses of the largest collapsed clusters. As a specific example, if we adopt z = 3,


n = -1, no = 0.3, and nHI(Z) = 0.004, then F = 16.61, R = 6.3 X 10-3 ,

the HI mass of an M* object is RM~F-l = 6.9 X 1010h-1 M0, and the abundance of such objects is noFe-1 = 2.4 X 10-4 h3 comoving Mpc3 . For an HI mass of 3 X 1011h-1 M0' 4.37 times higher, the abundance is down by a factor of e-4 .37 /4.37 to 6.9 X 1O-7 h3 co moving Mpc-3•

Of course, the fact that this argument leads to a low abundance of massive HI concentrations means that detection of such a concentration would be all the more interesting. The most dubious elements of the argument (if one is looking for order of magnitude gains, not factors of 2) are probably the assumption that the HI to total mass ratio R is universal and the assumption that the objects easiest to detect are indeed collapsed clusters as opposed to, e.g., lower overdensity structures that are just detaching from the Hubble flow.

4. Conclusions

According to conventional theories of cosmic structure formation, the large scale structure that we observe today should be mirrored in a scaled down form at high redshift. Absorption and emission measurements of high redshift HI can trace out the elements of this small scale structure. The agreement between hydrodynamic simulations and observed quasar spectra suggests that the Lya forest is produced largely by the moderately overdense (pi Ii '" 1 - 10) components of this structure, especially the collapsing filaments and sheets of warm, photoionized gas. Lyman limit and damped Lya absorption probably arises in the radiatively cooled gas of forming galaxies. Detecting 21cm emission from high redshifts is an ambitious goal, but simulated observations and inferences from absorption suggest that the radio arrays of the future could map the youthful universe in the way that today's galaxy redshift surveys have mapped the local large scale structure. The opportunity to watch galaxies, clusters, voids, and superclusters grow through time should take us a long way towards understanding their origin in the physics of the big bang.

DW acknowledges research and travel support from NASA grant NAG5-2882.

References

Bechtold, J., Crotts, A.P.S., Duncan, R.C. and Fang, Y. (1994), ApJ, 437, L83 Bahcall, N. A. and Cen, R. (1993), ApJ, 407, L49 Bunn, E. F., Scott, D., and White, M. (1995), ApJ, 441, L9 Cen, R., Miralda-Escude, J.,Ostriker, J.P. and Rauch, M. (1994), ApJ, 437, L9 Charlton, J.C., Churchill, C.W., and Linder, S.M. (1995), ApJ, 452, L81 Dinshaw, N., Foltz, C.B., Impey, C.D., Weymann, R.J. and Morris, S.1. (1995), Nature,

373, 223


Dinshaw, N., Impey, C.D., Foltz, C.B., Weymann, R.J. and Chaffee, F.H. (1994), ApJ, 437, L87

Fukugita, M. and Kawasaki, M. (1994), MNRAS, 269, 563 Gunn, J.E. and Peterson, B.A. (1965), ApJ, 142, 1633 Hernquist, L., Katz, N., Weinberg, D. H. and Miralda-Escude, J. (1995), submitted to

ApJ Letters (HKWM) Katz, N., Weinberg, D. H. and Hernquist, L. (1995), submitted to ApJS Katz, N., Weinberg, D. H., Hernquist, L. and Miralda-EscudC, J. (1995), submitted to

ApJ Letters (KWHM) McGill, C. (1990), MNRAS, 242, 544 Melott, A. L. and Shandarin, S. F. (1993), ApJ, 410, 469 Petitjean, P., Webb, J.K., Rauch, M., Carswell, R.F. and Lanzetta, K. 1993, MNRAS,

262,499 Quinn, T., Katz, N., and Efstathiou, G. (1995), MNRAS, in press Scott, D. and Rees, M. J. (1990), MNRAS, 247, 510 Scott, D., Silk, J., and White, M. (1995), Science, 268, 829 Shandarin, S. F., Doroshkevich, A. G., and Zel'dovich, Ya. B (1983), SOy Phys Usp, 26,

46 Shandarin, S. F. and Zel'dovich, Va. B (1993), Rev Mod Phys, 61, 185 Tegmark, M., Silk, J., and Blanchard, A. (1994), ApJ, 420, 2 Thoul, A. A. and Weinberg, D. H. (1995), submitted to ApJ Weinberg, D. H. and Gunn, J. E. (1990), MNRAS, 247, 260 White, S. D. M., Efstathiou, G., and Frenk, C. S. (1993), MNRAS, 262, 1023 Zel'dovich, Y. B. 1970, A&A, 5, 84 Zhang, Y., Anninos, P. & Norman, M.L. 1995, preprint astro-ph/9508133

ARE THE LYMAN ALPHA FOREST "CLOUDS" EXPANDING PANCAKES?

Some theoretical implications of the recent size determinations of Lya absorbers

M.G. HAEHNELT

Max-Planck-Institut fur Astrophysik K arl-Schwarzschild-Strape 1 85740 Garching, Germany

Abstract. The large sizes of Lya "clouds" inferred from coincident absorption in the spectrum of close quasar pairs suggests that these are transient flattened structures of small over density. It is argued that the observed absorbers should be preferentially located in underdense regions of the universe and should typically expand faster than the Hubble flow.

1. Introduction

The use of quasar absorption spectra to study the distribution of neutral hydrogen at high red shift is a well established cosmological tool. However, the physical nature of Lya forest absorbers itself had long remained unclear (see also the contribution by Michael Rauch, these proceedings). This is mainly due to the fact that basic properties like typical size, density and mass of the absorbers successfully eluded meaningful observational constraints. This has changed with the recent detection of coincident absorption lines in the two close quasar pairs Q0107-025AB and Q1343+266AB with proper separations of 360h-1 kpc and 40h-1 kpc at redshifts z r'V 1 and z r'V 2 (Dinshaw et al. 1994, Dinshaw et al. 1995, Bechthold et al. 1994). The observed fraction of coincident lines of about 50-80% implies that the absorbing structures coherently cover (with covering factor close to unity) an area which is up to a Mpc across. The inferred size depends somewhat on the shape and internal column density distribution of the absorbing structure and also on the cosmological parameter, but is in any case considerably larger than predicted by most of the models discussed earlier.

109


110 M.G. HAEHNELT

2. Fiducial parameters of Lyo: forest absorbers

Knowing the transverse size of the absorbers renders it possible to infer a variety of otherwise poorly constrained physical parameters. The typical neutral hydrogen density is nHI""'" 1.6 X 1O-10hf1-1 N141'10~cm-3, where N14 and 1'100 are the column density of neutral hydrogen and the transverse "radius" of the absorber scaled to 1014 cm-2 and 100h-1 kpc. There is still some uncertainty left parametrized as it which is due to the a priori unknown ratio between the measured transverse size of the absorber and its extent along the line of sight. It has been known for a long time that the absorbers must be highly ionized by the UV background. Taking the fiducial value for its intensity Iv = hI X 10-21 erg s-l cm- 2 Hz-1 sr-1

(Bechthold 1994) gives a typical neutral hydrogen fraction of x "-' 4 X

10-6 h1/ 2 f 1-O.S Ii.io.5 Nf4s 1'10or/. This corresponds to a typical total hydrogen density and baryonic mass of n 4 X 10-s h1/ 2 f-o. s 10.s NO . .') 1'-0.5 cm-3

H "-' 1 21 14 100 and M 8 X 109 h- S/ 2 fO. S 1°.5 N°·S 1' 2 .S M We can further use the bar "-' 1 21 14 100 8· ,

observed column density distribution f( N) to obtain an estimate of the overall fraction of the critical density contained in the Lyo: forest

n - j1 mH H o J x-1(N) N feN) dN 008 h-3/ 2 f O.S 10.5 1'0.5 HLyc> - "-' . 1 21 100, C POcrit

(1)

where j1mH is the mean mass per hydrogen atom and the other symbols have their usual meaning. Equation (1) shows that the recent size estimates make the inferred OLyc> uncomfortably large in comparison to the nucleosynthesis constraint on the baryonic matter content of the universe. As argued by Rauch & Haehnelt (1995) this suggests a flattened geometry of the absorbers with it ;G 0.1.

3. What is the nature of the Lyo: forest absorbers?

The main properties of the absorbers can be summarized as follows:

• A significant fraction of all baryons (of order unity) is contained III

them. • The baryonic mass of the individual absorber is similar to that of a L*

galaxy, but the inferred number density exceeds this of L* galaxies by a factor of about 10 - 30.

• They are overdense compared to the mean baryonic density by a factor of about a few to ten.

• They are likely to have a flattened geometry.

All of these points strongly argue against them being virialized objects. As realized by Cen et al. (1994) and others the Lya forest absorbers are most likely some modest transient density fluctuations of the intergalactic

LYMAN ALPHA ABSORBERS AS EXPANDING PANCAKES 111

medium caused by the large scale flows and density fluctua.tions of the dynamically dominant dark matter component of the universe.

However, the question remains what sort of underlying structures are causing these density fluctuations and in which dynamical state they are. The standard paradigm for the origin of large-scale structure is the growth of small primordial density fluctuations which can be described by a Gaussian random field. In the following I will use the Zeldovich approximation (Zeldovich 1970) to describe the dynamical evolution of such a density fluctuation. The Zeldovich approximation is an astonishingly good description even in the mildly non-linear regime and should give a good qualitative understanding. The trajectory of a particle in an Einstein-de-Sitter universe is then given by

ri(q, t) = a(t) [qi + (a(t) - 1)] cJ>,.(q), (2)

where a(t) is the global growth factor, q is the initial Lagrangian coordinate and cJ> is the gravitational potential. The distribution of the eigenvalues of the Zeldovich tensor cJ> ,.j was first derived by Doroskhevich (1974) and can be written as

153 p(Ab A2, A3) = h (A3 - A2) (A3 - AI) (A2 - AI) (3)

81l'v5

X exp { -~[2(Ai + A~ + A~) - (AIA2 + AIA3 + A2A3)]} d3 A,

(4)

where Al :S A2 :S A3 (Steinmetz & Bartelmann 1995). Negative eigenvalues correspond to a contraction along the corresponding axis with respect to the Hubble flow, while positive eigenvalues describe expansion. The density contrast in the linear regime is determined by the sum of all eigenvalues {j = (p - (p)) / (p) = -0' (AI + A2 + A3) and the rms amplitude of the density fluctuations 0'. Completely collapsed and virialized objects generally originate from regions, where {j is initia.lly positive. However, collapse along one axis should be sufficient to produce low-column density absorption lines. This is considerably more likely to occur than a full collapse along all three axis. The solid curve in Fig.la shows the fraction of the mass in regions where collapse along one axis has occurred according to the Zeldovich approximation (AI 0' < -1) while the other two eigenvalues are positive. For the dashed curve (AI + A2 + A3) > 0 was imposed as second constraint. Both fractions are about 50 % and depend only weakly on 0'.

To estimate the dynamical state of a typical absorber we can use again the Zeldovich approximation. The velocity between two points in the plane of the pancake will then be linearly proportional to their distance (just as in

112 M.G. HAEHNELT

O.B

-------

0~0~.~5~~1~~1~.5~~ a

1.5

1.4

~ 1.3

1.2

1.1

o

--.,-:'"

/

" .......... -_. : ............... . a= 1.5

a= 1.0

a= 0.5

0.5 1 1.5 D/R

Figure 1. The solid curve of the left diagram shows the fraction of the mass in regions where collapse along one axis has occured according to the Zeldovich approximation (J\J IT < -1) while the other two eigenvalues are positive. For the dashed curve P'l + ).2 + ).3) > 0 was imposed as second constraint. The right diagram shows the ratio Q' between the expansion velocity in the plane of a randomly orientated Zeldovich pancake giving rise to coincident absorption as a function of the ratio between line-of-sight separation and radius of the pancake. The three curves are for different values of the rms amplitude of the density fluctuations IT as indicated in the plot.

Hubble's law). Figure 1b shows the ratio 0' between the expansion velocity in the plane of the pancake and the Hubble velocity for randomly orientated pancakes. The typical pancake giving rise to coincident absorption is expanding about 30% faster than the Hubble flow. This is due to the fact that positive and negative eigenvalues of the Zeldovich tensor are equally likely and that the pancakes expanding fastest have the largest cross section for absorption. For the same reason low-column-density absorption lines should be preferentially embedded in underdense region of the universe.

4. Observational tests

The model described above predicts rather small velocity differences between coincident absorption lines in adjacent lines of sight

6.V= O'cotBH(z)D.-v500'cotB (1+Z)3/2 ( hDlk ) 3 100 - pc

where B is the angle between the plane of the pancake and the line of sight and D is the proper distance between the two lines of sight. The probability distribution ofthe orientation angle of Zeldovich-pancakes giving rise to coincident absorption is plotted in Fig.2. Face-on pancakes are more likely to be responsible for coincident absorption. This is mainly due to their larger

LYMAN ALPHA ABSORBERS AS EXPANDING PANCAKES 113

2.5 ,

2

, -, -, , , I

D/R = 0.5

D/R = 1.0

D/R = 1.5

2 D/R = 0.5

......... 1.5 CJ:> '-'

0.. 1

0.5

L •

I ·"-t - • .'.1..[

"'-i

D/R = 1.0 ......... CJ:> 1.5 +-'

D/R = 1.5 0 0

1 '-' 0..

0.5

00 0.5 1 cot (J

Figure 2. The left diagram shows the probability distribution of the orientation angle f) between the plane of a Zeldovich-pancake giving rise to coincident absorption and the line of sight. The three curves are for different values of the ratio between line-of-sight separation and radius of the pancake as indicated in the plot. The right diagram shows the probability distribution of cot f}.

cross section. The observational results are so far inconclusive. The velocity differences observed in the spectra of Q0107-025AB and Q1343+266AB are smaller than would be expected for virialized objects of this size, but the quoted values are hardly larger than the errors. Furthermore, the lineof sight differences are uncomfortably small and the column densities of the coincident lines rather large for this kind of test. However, the model should become testable in the near fu ture with quasar pairs of somewhat wider separation and improved signal-to-noise.

5. Conclusions

The recent measurement of the sizes of Lya forest absorbers has changed our understanding of their nature. They are now believed to be transient density fluctuations of the intergalactic medium reflecting the evolution of pancake-like structures in the dark-matter component of the universe. These pancakes are typically expanding about 30% faster than the Hubble flow and in most cases they will never collapse, but rather be incorporated into larger structures.

References

Bechtold,J., Crotts, A.P.S., Duncan, R.S., Fang, Y., 1994, ApJ, 437, L83 Bechtold,J., 1994, ApJS, 91,1 Cen, R., Miralda-Escude, Ostriker, J.P., Rauch, M., 1994 ApJ, 437, L9

114 M.G. HAEHNELT

Dinshaw, N.,Foltz, C.B., Impey, C.D., WeymanIl,R.J., Morris, S.L., 1995, Nature, 373, 232

Dinshaw, N., Impey, C.D."Foltz, C.B., Weymann,R.J., Chaffee, F.H., 1994, ApJ, 437, L87

Doroskhevich, A.D., 1970, Astrofislka, 6, 581 Rauch M., Haehnelt M.G., 1995, MNRAS, 275, L76 Steinmetz, M., Bartelmaun, M., 1995, MNRAS, 272, 570

ON THE DISTRIBUTION OF INTERGALACTIC CLOUDS

STANISLAW BAJTLIK Copernicus Astronomical Center ul. Bartycka 18, 00-716 Warsaw, Poland

1. Introduction

The spatial correlation function of Ly-a clouds should be similar to that of galaxies if both populations have the same mass distributions and if both formed by the process of gravitational instability. The 1-D correlation function (along the line-of-sight) is equal to the 3-D function, so long as the spatial separation of clouds is much greater than their size. Observations of close quasar pairs suggest that this may not be the case. We present an explanation for why the spatial correlation of these clouds may be negligibly small. This explanation takes into account the large cloud size to cloud separation ratio, the redshift evolution of the correlations and mixing in the same catalog objects located at a very wide range of redshift. The implications for observing strategies are discussed.

Thirty years after the first theoretical predictions (BahcaU and Salpeter 1965, Wagoner 1967) and observational detections of the Ly-a clouds (Burbidge et al. 1966, Stockton and Lynds 1966) their physical properties, such as size, shape, density, ionization level, temperature and confinement mechanism are still unknown. Similarly, their epoch and method of formation is not clear. Models of Lya clouds range from pressure confined clouds (Ikeuchi and Ostriker 1986), to clouds in mini-halos of dark matter (Rees 1986), and slabs of intergalactic material (Charlton, Hogan, Slapeter 1993). Recent numerical simulations show intergalactic absorbers are mostly very elongated, thin filaments (Cen and Ostriker 1994, Katz et al. 1995, Hernquist et al. 1995). The thickness of these filaments (10 kpc) seems to be in contradiction with observations of close pairs of quasars (Bajtlik and Smette, 1995). Low red shift observations suggest that Ly-a absorption lines arise in the extended gaseous halos of the most luminous galaxies (Lanzetta et ai. 1995). For a review of the proposed models and basic observations of Ly-a forest see Bajtlik (1993).

115


116 STANISLAW BAJTLIK

One of the most intriguing properties of the population of Ly-a clouds is the apparent lack of spatial correlations in their distribution. Since the first study of this problem, made by Sargent, Young, Boksenberg and Tytler (1980) (hereafter SYBT) most research has resulted in no detections of any spatial correlations. Usually 1-D correlation function analysis has been used (Carswell et al. 1984, Rees and Carswell 1987, Lu et al. 1991). Crotts (1989) and Ostriker, Bajtlik and Duncan (1988) used nearest neighbor statistics, which are more suitable than the correlation function for detection of completely empty voids (rather then over-dense and under-dense regions). Dobrzycki and Bechtold (1991) have detected a single void in the spectrum of a quasar they observed. Duncan, Bajtlik and Ostriker (1988) developed a model for the description of the line blending effect. Other classes of intergalactic absorbers (metal line systems) do show a strong correlation on a velocity scales of a few hundred km/s (SYBT, Sargent, Boksenberg and Steidel 1988).

Cristiani et al. (1994) announced a detection of significant clustering between strong lines with ~ ~ 1 at .6.v = 100 km s-1. As this result is controversial, based on weak detected signals and recent observations imply an opposite conclusion (Carswell, 1994), we should wait for further confirmation of a possible detection of correlations.

Models of the formation of Ly-a clouds view this process as a part of the process of structure formation in the Universe in general (Bond, Szalay and Silk 1988, Miralda-Escude 1995) and predict cloud masses to be comparable to the galactic masses. We should therefore expect clustering properties of clouds to be similar to the clustering of galaxies.

2. Correlation Function: Time Evolution and I-D Properties

The correlation function for galaxies and clusters is:

~(r) = (ro/r)'Y, (1)

where I ~ 1.8. The correlation scale for galaxies is ro ~ 5h101o Mpc and for clusters is ro ~ 24hlo~ Mpc (Peebles 1993). This correlation is equivalent to an excess probability of finding a second object at a distance r = lrul from the first one. A differential probability is defined as:

(2)

where n is a comoving mean number density of objects, and dV1 and dVz are volume elements.

There is a crucial difference between 3-D or 2-D catalogs and a I-D survey which QSO spectra represent. For galaxies, we correlate positions of

ON THE DISTRIBUTION OF INTERGALACTIC CLOUDS 117

, ,

..,#<4; .... _-- ...

, I

I , I I \ \ ,

"

I " I " , ,

.... ,~ ,

,

""""""" ",

"

, , ,

, ,

Figure 1. Clouds ofradii R are intercepted by line ofsight (LoS). 0 1 A = Fi', O 2 '8 = B, 010~ = r>, AB = s.

objects regardless their physical sizes. In one-dimensional catalog however physical sizes of objects - their cross sections are crucial.

SYBT discussed this problem. The excess differential probability in a one dimensional survey is:

dp = <Poa(1 + ((s))ds, (3)

where <Po is comoving number density of objects, a their cross section and s separation along the line of sight. SYBT concluded that provided cloud sizes are small compared with their separations the relation holds: (( s) = ~(s).

The evidence for cloud sizes comes from of spectra of lensed QSO or close pairs of quasars. Foltz et al. (1984) and Bajtlik and Duncan (1991) studied spectra of Q2345+007 A and B (at Zem = 2.14) and concluded that the radii of clouds are of the order of 10 - 80 kpc. Smette et al. (1992), from the analysis of UM 673 A and B (at Zem = 2.727) obtained a lower limit of 6 kpc for clouds radii. Recently Dinshaw et al. (1995) have obtained much larger limits from the spectra of Q0107-025 A and B (at Zem = 0.956 and 0.952). According to them, clouds can be as big as several hundred kps. This means that their sizes could be comparable to their mean separations.

The situation discussued here is visualized on Fig. 1. The definition of the correlation function in the 1-D case is equivalent

to the 3-D case, with the volume of integration corresponding to a narrow cylinder (marked with the dashed lines on Fig. 1).


We assume all clouds are of the same size R (this can represent pressure confined clouds in an inhomogeneous IGM). The relation between the true correlation function ~(r), where r is a physical distance between clouds centers and the observed I-D function ((s), where s is a projected distance along the line of sight is:

(4)

where

(5)

Observations of Ly-a absorption in the optical regime can cover wide red shift intervals (typically ~z ~ 0.6 for a high redshift quasar). The correlation function can vary quite substantially in this range. We can factorize time and space dependence by:

For a flat Universe (n = 1):

D(t)ex{t~/3, for~~I; t, for ~ > 1.

(6)

(7)

This is a very rapid evolution in redshift space. Changing variable from t to z, again for the flat Universe we have:

D(z) ex {(I + z)-l, for (~ 1; (1 + Z)-3, for ( > 1.

From equations (4) and (6) we get:

(8)

(9)

For a specific epoch we can calculate the numerical factor by which the "true" correlation function ~ is smoothed out by the fact that a given physical separation can correspond to a range of projected line of sight separations (and vice versa):

((s) = f(u)~(s). (8)

Figure 2 presents this numerical factor f( u) as a function of the ratio 11 = siR.

ON THE DISTRIBUTION OF INTERGALACTIC CLOUDS 119

0.1 10 100 u

Figure 2. Numerical factor f(u) relating ~ and (: ((s) = f(lI)~(S), where 11 = siR

3. Conclusions

For clouds of sizes comparable to their separations, the amplitude of the correlation function can be diminished by up to 50 percent. As the expected amplitude of the correlation function should not be very large, this may be an important effect and should be taken into account in the analysis of the data.

Evolution of the correlation function makes searching for correlations in optically (i.e. z > 1.6) collected data difficult. A standard procedure of combining data from very different regions in redshift space also reduces the signal. More strongly correlated signal from part of the sample at lower z is diluted by more the abundant and less correlated data from higher z. This effect can be quantified for each specific sample (Eqn. (9)). In Bajtlik, Juszkiewicz & Bergeron (1995) detailed analysis of this effect will be presented. We will also present conclusions about the ratio of sizes of Ly-o: and metal systems. Metal systems show a clear correlation with redshift. If the large sizes of the Ly-o: clouds (relative to their mean separations) are the cause of the apparent lack of the correlations between, them we can hope to use this to learn about the ratio of sizes of both type of absorbers.

It is clear that the best observing strategy must be a compromise between the two trends: it is favorable to look for the correlations in a low redshift sample, as the correlation function is larger there, but there are fewer clouds at lower z. The optimal observational program should be to observe many lines of sight, each covering the same narrow red shift interval. Results of such an observing program will be presented in Bajtlik et al. (1995). The conclusions from this analysis will have applications to one


dimensional (pencil beam) survey of galaxies as well.

Acknowledgments. The author thanks J. Bergeron and R. Juszkiewicz for allowing the presentation of the results prior to publication. This research was partially supported by Polish grant KBN No 2 P304 016 07, by PICSjCNRS No. 198 Astronomie Pologne Program and by Kapteyn Institute, Groningen.

References

Bahcall, J.N. (1965). ApJ, 142,1677 Bajtlik, S., (1993). The Evolution of the Intergalactic Medium and Ly-O' Clouds, in Pro

ceedings, Tetons Conference on Environment and Evolution of Galaxies, J .M. Shull, M.A. Thornson, Eds, Kluwer, p. 191

Bajtlik, S., Duncan, R.C. (1991). Proceedings of the ESO Mini- Workshop on Quasar Absorption Lines, ed. P.Shaver, E.J. Wampler, A.M. Wolfe ESO Scientific Report No.9, February 1991, 35

Bajtlik, S., Juszkiewicz, R., Bergeron, J. (1995). To be published Bajtlik, S., Smette, A. (1995) in preparation Bond, J.R., Szalay, A.S., Silk, J. (1988). ApJ, 324, 627 Burbidge, E.M., Lynds, C.R., Burbidge, G.R .. (1966). ApJ, 144, 447 Carswell, R.F. (1995). This volume Carswell, R.F., Morton, D.C., Smith, M.G., Stockton, A.N., Turnshek, D.A., Weymann,

R.J. (1984). ApJ, 278,486 Carswell, R.F., Rees, M.J. (1987). MNRAS, 224, 13P Cen, R., Ostriker, J.P. (1994). ApJ, 431,451 Charlton, J.C., Salpeter, E.E., Hogan, C.J. (1993). ApJ, 402, 493 Cristiani, S., D'Odorico, S., Fontana, A., Giallongo, E., Savaglio, S. (1995). In QSO

Absorption Lines, Meylan, G., Ed., Springer-Verlag, p. 357 Crotts, A.P.S. (1989). ApJ, 336, 550 Dinshaw, N., Foltz, C.B., Impey, C.D., Weymann, R.J., Morris, S.L. (1995). Nature, 373,

223 Dobrzycki, A., Bechtold, J. (1991). ApJ, 377, L69 Duncan, R.C., Ostriker, J.P., Bajtlik, S. (1988). ApJ, 345, 39 Foltz, C.B., Weymann, R.J., Roser, H.J., Chaffee, F.H. (1984). ApJ, 281, L1 Hernquist, L., Katz, N., Weinberg, D.H., Miralda-Escude, J. (1995). ApJ, submitted Ikeuchi, S., Ostriker, J.P. (1986). ApJ, 301,522 Katz, N., Weinberg, D.H., Hernquist, 1. (1995). ApJ, submitted Lanzetta, K.M., Bowen, D.V., Tytler, D., Webb, J.K. (1995). ApJ, submitted Lu, L., Wolfe, A.M., Turnshek, D.A. (1991). ApJ 367,19 Miralda-Escude, M., Cen, R., Ostriker, J.P., Rauch, M. (1995). In QSO Absorption Lines,

G. Meylan (Ed.), Springer-Verlag, p. 427 Ostriker, J.P., Bajtlik, S., Duncan, R.C. (1988). ApJ, 327, L35 Peebles, P.J.E. (1993). Principles of Physical Cosmology, Princeton University Press Rees, M.J. (1986). MNRAS, 218, 25P Sargent, W.1.W., Young, P.J., Boksenberg, A., Tytler, D. (1980). ApJS, 42, 41 (SYBT) Sargent, W.1.W., Boksenberg, A., Steidel, C.C. (1988). ApJS, 68, 539 Smette, A., Surdej, J., Shaver, P.A., Foltz, C.B., Chaffee, F.H., Weymann, R.J., Williams,

R.E., Magian, P. (1992). ApJ, 389, 39 Stockton, Lynds, R. (1966). ApJ, 144, 451 Wagoner, R.V. (1967). ApJ, 149, 465

DISK GALAXIES AT Z=O AND AT HIGH REDSHIFT

Explanation of the Observed Evolution of Damped Lya A bsorption Systems

G. KAUFFMANN

Max-Planck-Institut fur Astrophysik Karl-Schwarzschild-StrafJe 1 85740 Garching, Germany

Abstract. The analysis of disk formation in this paper is based on the White & Rees (1978) picture, in which disk galaxies form by continuous cooling and accretion of gas within a merging hierarchy of dark matter halos. A simple Kennicutt law of star formation for disks, based on a singlefluid gravitational stability model, is introduced. Since the gas supply in the disk is regulated by infall from the surrounding halo, the gas is always maintained at a critical threshold surface density l:e, where l:c (X Vel R. Chemical enrichment of the disks occurs when the surrounding hot halo gas is enriched with heavy elements ejected during supernova explosions. This gas then cools onto the disk producing a new generation of metal-rich stars.

I first show that models of this type can reproduce many of the observed properties of a typical spiral galaxy like the Milky Way, including its gas and stellar surface density profiles and the observed relationship between the ages and metallicities of solar neighbourhood stars. I then use the models to make inferences about the properties of disk galaxies at high redshift. The total neutral hydrogen density n(RI) increases at higher z. The predicted increase is mild, but is roughly consistent with the latest derivation of n(RI) as a function of z by Storrie-Lombardi & McMahon (1995). The models are also able to account for some of the other trends seen in the high-redshift data, including the increase in the number of high column-density systems at high redshift, as well as the metallicity distribution of damped Lya systems at z '" 2 - 3.

121


122 G. KAUFFMANN

1. Introduction

The damped LyQ systems are the rarest of the different classes of quasar absorption lines, yet they have attracted considerable attention because of indications that they are the high-redshift progenitors of present-day galaxies. The most persuasive evidence comes from the integrated mass of neutral hydrogen contained in these systems at redshifts between 2 and 3, which is comparable to the total mass in stars in galaxies today. In addition, the neutral hydrogen mass density is seen to decrease with redshift, and at the lowest redshifts surveyed so far, is in good agreement with the global density of neutral gas determined from 21 em studies of nearby galaxies (Rao & Briggs 1993, Fall & Pei 1993). The conventional theoretical picture is one in which pure HI disks are assembled at some high redshift. The gas is then transformed into stars until roughly 90 percent has been "used up" by z = o. Models of this type have been explored in detail by Lanzetta and coworkers (Lanzetta, Wolfe & Turnshek 1995; Wolfe et al. 1995). Models taking into account the effects of dust on the inferred column density distribution of the damped Lya systems have been explored by Fall & Pei (1993) and Pei & Fall (1995).

Although models of this type are certainly illustrative, they are not realistic representations of the formation histories of real disk galaxies. According to hierarchical clustering theories, which currently constitute the standard paradigm of structure formation in the universe, galaxies, groups and clusters form continuously through the merging of small subunits to form larger and larger systems. In this scenario, the evolution of the dark matter component of the universe is roughly self-similar; only the scale of the collapsed structures changes with time. If it is indeed the evolution in the clustering of the dark matter component that regulates the formation and evolution of galaxies, one might expect that to first order, galaxies at high redshift be rather similar to galaxies today. The main difference would be that the "typical" galaxy be of lower mass and luminosity. Thus if one takes hierarchical clustering theories seriously, the suggestions of Pettini, Boksenberg & Hunstead (1990), York (1988) and Tyson(1988) that the damped systems may be dominated by a population of dwarf galaxies with properties similar to dwarf galaxies today may warrant re-examination.

In this paper, we explore predictions for the gas properties of high redshift galaxies in cold dark matter (CDM) models, in which disk galaxies form as gas cools and forms stars at the centres of dark matter halos. In the model, disk gas is continuously replenished as a result of infall from the surrounding hot halo. This type of model requires a substantial source of energy input to keep all the gas from cooling off and forming dense lumps in small halos at high redshift, where the cooling times are short. Possibilities

DISK GALAXIES AT Z=O AND AT HIGH REDSHIFT 123

include energy injection by supernova explosions that occur soon after the first generation of stars begin to form in a galaxy. Alternatively, gas may be prevented from cooling off at all in small systems at high redshift by the presence of a photo-ionizing background of UV-radiation produced by quasars (Efstathiou 1992). Continuous infall models appear to be necessary in order to explain the observed sizes of disk galaxies today. Simulations that do not incorporate any heating processes produce cold gas disks with scale radii much too small to be compatible with observations (Navarro & White 1994; Steinmetz 1995). These authors find that a substantial amount of angular momentum is lost to the dark matter during merging of the small dense lumps of gas that are able to cool off at high redshift. On the other hand, hot gas halos around ordinary spiral galaxies have yet to be observed in the X-ray, where they should be most easily visible.

2. Description of the Model

2.1. FORMATION AND MERGING OF DARK MATTER HALOS

The formation, evolution and merging histories of the dark matter halos in which gas will cool and condense, is specified using a semi-analytic technique developed by Kauffmann & White (1993), based on an extension of the Press-Schechter theory due to Bower (1993) and Bond et al. (1993). The original Press-Schechter theory gave the mass distribution of collapsed, virialized objects in the universe as a function of redshift. The formalism was applicable to any set of cosmological initial conditions resulting in the hierarchical buildup of structure. With the extended theory, one is able to specify the probability that a halo of mass Ml at redshift Zl will later be incorporated into a halo of mass Mo at Zoo Extensive tests of the theory using numerical simulations of gravitational clustering have been carried out and remarkably good agreement has been found (Kauffmann & White 1993, Lacey & Cole 1994).

In our models, we construct Monte Carlo realizations of the merging history of present-day halos of given mass. In this way, we follow not only the evolution in the global mass distribution of halos as a function of redshift, but also the formation and growth of individual halos with time. The interested reader is referred to Kauffmann & White (1993) for further details.

2.2. COOLING OF GAS

The treatment of gas cooling is based on a model by White & Frenk (1991). A dark matter halo is modelled as an isothermal sphere that is truncated at its virial radius, defined as the radius within which the mean overdensity

124 G. KAUFFMANN

is 200. In the standard spherical collapse model, each mass shell is assumed to virialize at one half its maximum expansion radius. The virial radius is then related to the circular velocity of the clump and to redshift by

(1)

The virial mass may then be written

(2)

We assume that when a halo forms, the gas initially relaxes to a distribution which exactly parallels that of the dark matter. The gas temperature is then related to the circular velocity of the halo through the equation of hydrostatic equilibrium. The cooling radius is defined as the radius within the halo at which the cooling time is equal to the Hubble time. If the cooling radius lies outside the virial radius, we are in the accretion-limited case where all infalling gas cools immediately. If the cooling radius lies inside the virial radius, we model the cooling rate by a simple inflow equation:

• () 2 dreool Meool = 4rrpg reool reool~ (3)

The gas that is able to cool will collapse to form a rotationally-supported disk at the centre of the halo. For gas that collapses within the potential of a massive halo while conserving angular momentum, the collapse factor fdiss may be written

(4)

where AD is the spin parameter of the disk, observed to lie in the range 0.4-0.5 for real spiral galaxies, and AH is the spin parameter of the dark halo. N-body simulations show that halos typically have AH '" 0.05 ± 0.03 as a result of tidal torquing (Barnes & Efstathiou 1987), so for a disk/halo mass ratio of 0.1, one obtains a collapse factor of 0.1. In practice, because AH has substantial scatter, one expects a spread in collapse factors. For simplicity, we will assume that gas will collapse to a constant fraction fdiss = 0.1 of its radius within its parent halo. Gas distributed isothermally within the halo thus forms a disk with surface mass density E ex r-1 and outer radius RD = O.lRH.

As seen from equation 1, the size of a dark matter halo is proportional to its circular velocity and scales with red shift as (1 + z )-3/2. Disks are thus built up from the inside out as the mass of their surrounding dark halos grows with time and gas falls in from larger and larger radii.


2.3. STAR FORMATION AND FEEDBACK IN GALACTIC DISKS

In a seminal paper in 1989, Kennicutt showed that the star formation rates and radial profiles of a sample of nearby spiral galaxies could be explained very simply by a combination of a Schmidt power-law rate of star formation at high gas densities and a cut-off in star formation below a certain critical threshold surface density. In this paper, Kennicutt noted that an abrupt decrease in star formation at low gas densities is expected from simple gravitational stability considerations, as first discussed by Toomre (1964). In the case of a thin isothermal gas disk, instability is only expected if the surface density exceeds the critical value

OK,C ~ --

c - 3.36G' (5)

where 0 is a dimensionless constant near unity, c is the velocity dispersion of the gas, and K, is the epicyclic frequency given by,

V( RdV)1/2 K, = 1.41- 1 +--

R VdR (6)

For his sample of 15 galaxies, Kennicutt showed that the outer radii of the HII regions in these galaxies corresponded extremely well to the radii at which their gas surface densities dropped below the critical density. In addition, within the star-forming disks of the galaxies, the ratio of the gas surface density to the critical density was always close to unity, indicating that the gas disks tended to lie near their gravitational stability limit.

In our models, we adopt the simple form of a Kennicutt star formation law. We adopt a constant gas velocity dispersion c = 6 km s-l and assume that all disks have flat rotation curves and a rotational velocity equal to the circular velocity of their surrounding dark matter halos. The epicyclic frequency K, is then simply given by VIR and the stability condition takes the form

(7)

At densities greater than ~crit, we adopt a star formation law of the form

(8)

where f3 is a free parameter controlling the efficiency of star formation and tdyn is the dynamical time of the disk tdyn = (VI RD)-l . In practice, the resulting gas profiles of the disks in our model depend rather little on f3 because the supply of gas in the disk is regulated on a short timescale by infall of gas from the surrounding halo. The surface density of gas in the disk always tracks the critical density, except at the outer edges where the disk is only just beginning to form (see section 3.1).

126 G. KAUFFMANN

To maintain a supply of hot gas in the halo so that continuous infall is able to take place, we postulate that supernova explosions can release enough energy to drive cold gas back into the hot intergalactic medium. For a standard Scalo IMF, the number of supernovae expected per solar mass of stars formed is 1JSN = 4 X 10-3 Mr;/. The kinetic energy of the ejecta from each supernova is 1051 ergs. If a fraction f of this energy is used to reheat cold gas to the virial temperature of the halo, the amount of cold gas lost to the intergalactic medium in time !::.t may be estimated from simple energy balance arguments as

(9)

Here f is a free parameter controlling the efficiency of the feedback process. In practice, the value of f will determine the total amount of gas that is transformed into stars and hence the luminosity of the galaxy, but the gas profiles in the disks are insensitive to changes in this parameter.

2.4. A SIMPLE MODEL FOR CHEMICAL ENRICHMENT

The continuous infall models described above require substantial heating of halo gas by star formation activity, so it seems plausible that a large amount of processed material could be mixed into the gas out to large radii in the halo. There are at least two observational indications that this kind of process does indeed occur. The [Mg II] absorption-line systems observed at low redshift in quasar spectra are almost always associated with starforming galaxies (Bergeron 1988; Steidel, Dickinson & Persson 1994). These systems are often seen at several optical diameters away from the galaxy centre, suggesting that halos are enriched to large radius as a consequence of star formation. The second indication is the mean metallicity of the xray emitting gas in rich clusters. The total metal contents of the cluster gas and the stars of the cluster galaxies are comparable. This would indicate that a substantial fraction of heavy elements produced by the stars was not retained by the galaxies.

White & Frenk (1991) have explored chemical evolution models of this type. For every solar mass of stars formed in a galaxy, an effective yield y of heavy elements is assumed to be ejected and uniformly mixed with the hot halo gas. The mass of metals in stars thus increases as metals are incorporated from the surrounding hot gas as a result of cooling and infalI. The metallicity of the gas is increased by stellar ejecta, but decreased by metals lost to stars and accretion of primordial material as the halos grow in mass. Following the techniques outlined in White & Frenk (1991), the effective yield y is taken as a free parameter and its value is constrained by


requiring that L* galaxies be enriched to roughly solar metallicity by the present day. This approach is adopted because of the large uncertainties associated with modelling the ejection and mixing processes. The main purpose of the model is to explore relative changes in the metallicity of galaxies as a result of differing accretion or star formation histories. We do not claim that it in any way constitutes a detailed or accurate description of the physical processes affecting the intergalactic medium.

2.5. FIXING THE FREE PARAMETERS IN THE MODEL

The parameters 0', f and y, which respectively control the star formation efficiency, the feedback efficiency and the heavy element yield, are constrained by requiring that on average, disk galaxies that form within halos of circular velocity 220 km s-1 have properties that match those of our own Milky Way galaxy. As discussed previously, the gas surface density profiles of the disks in our model track the critical density. The total gas mass thus does not depend very much on our choice of parameters. For a Milky Way-type galaxy, Mtot(gas) "-' 1010M0 , which is roughly comparable to our Galaxy's measured HI mass of 8 X 109 M 0 . The parameter f fixes the luminosity of the galaxy and we set its value by requiring that the B-band luminosity LB have a value of"-' 2 X 1010 L0 for a Milky Way-type galaxy. Finally, the yield is fixed by requiring that the mean metallicity of the stars in the Milky Way be close to solar.

We restrict ourselves to cold dark matter (CDM) initial conditions with n = 1, Ho = 50 km S-l Mpc-1 and nbaryon = 0.1. Whenever not specified, we adopt a normalization with 0"8 = 0.67 (b = 1.5).

3. Results of the Model

3.1. THE PROPERTIES OF DISK GALAXIES AT Z = 0

We will first explore to what extent our models can reproduce the properties of disk galaxies at the present day. The left-hand panel of figure 1 shows the HI and stellar surface density profiles of a disk galaxy residing in a halo with circular velocity Vc = 220 km s-l. The dotted line is the critical surface density defined in equation 7. As can be seen, the HI surface density tracks the critical density out to a radius of 20 kpc. Beyond this radius, gas is falling in for the first time and there has not yet been time for the surface density to reach the critical density and for star formation to "switch on". This is reflected in the stellar mass density profile, which is truncated abruptly at a radius of 20 kpc. The HI column densities measured through a face-on disk range from a few X 1021 cm-2 in the central few kiloparsecs, to 1020 cm-2 at a distance of"-' 30-40 kpc. This agrees rather well with what is

128 G. KAUFFMANN

22 22

21 21

fP fP ! ! ~ 20 ~ 20 ,.. ,..

19 19

5 5

k ,;: ;.! 0

1 :s -5

;.! 0

1 .s -5

o 10 20 30 o 5 10 Radius (kpc) RadIus (kpc)

Figure 1. The gas and stellar surface density profiles of disk galaxies halos with circular velocity Vc = 220 km S-1 and 75 km S-1 at z = o. The gas surface density is plotted in units of the HI column density seen through a disk with face-on orientation. The dotted line is the critical surface density. The stellar surface density is plotted ill M 8/PC2.

observed for real spiral galaxies (Bosma 1981), where N(HI) typically falls below 1020 cm-2 at about 1.5 Holmberg radii. The stellar mass falls off much more steeply than the gas. It is roughly exponential over much of the disk, with a scale length "-' 4 kpc. We thus conclude that our model galaxy is in reasonable agreement with the properties of the Milky Way disk component. In the right-hand panel of figure 1, we show the HI and stellar profiles of a "dwarf" disk galaxy contained in a halo of circular velocity 75 km s-1. This galaxy is simply a scaled-down version of its brighter counterpart. The stellar scale length is about 1 kpc, typical of fainter disk systems such as the Magellanic Clouds.

The total HI mass of a galaxy is obtained by integrating the gas surface density over the area of the disk,

(Rlim Mgas = 21l" Jo Egas(r)rdr, (10)

where Rlim is the outer limit of the disk. Since Egas( r) "-' Eerit( r) ex: Vel r and Rlim ex Ve , we find that Mgas ex: Ve2 • Therefore, using equation 2, Mgasl Mhalo ex: Ve-1, i.e. smaller halos have higher gas mass fra.ctions. If


"

O~~~~~~LL~~-w

10 B B " 2 o z (redshiCl)

Figure 2. The star formation history of a disk galaxy that is in a halo with Vc = 220 km S-1 at the present day. The different lines show various realizations of the formation history of the disk.

galaxies obey the B-band Tully-Fisher relation (Ls lX Vc2 .7 ), we obtain a gas-mass/ B-band luminosity relation of the form Mgas lX L Bo.7 . As was first realized by Quirk (1972), this offers a natural explanation for why dwarf galaxies appear substantially more gas-rich than bright galaxies.

Star formation histories of Milky Way-type disk galaxies are plotted in figure 2. The star formation rate typically increases from a few tenths of a solar mass per year at redshifts around 10, to values of between 1 and 2 solar masses per year at redshifts around 2-3, by which time a substantial fraction of the mass of the final halo has already been assembled. the star formation rate then remains roughly constant until the present day. It should be noted that at no time does a disk galaxy have very high rates of star formation.

The simple chemical evolution model outlined in section 2.4 enables us to determine the metallicity distribution of stars in a Milky Way-type disk at the present day. Recall that in our model, metals ejected in supernova explosions are mixed into the hot halo gas. Chemical enrichment occurs when metal-rich gas from the halo cools and forms new disk stars. Note also that enrichment occurs more readily in lower mass galaxies, since the mass of gas that is returned to the intergalactic medium per solar mass of stars formed scales as Vc-2 in equation 9, i.e gas can more easily escape the potential wells of less massive galaxies. As a result, disk galaxies in ollr model undergo rapid, early enrichment while they are still dwarf systems. One of the advantages of this chemical evolution scheme is that the disks then do not suffer from the classic "G-dwarf problem" that plagues simple closed-box models. This is illustrated in figure 3, where we plot the cumulative fraction of stars in the disk with metallicities smaller than a given value. As can be seen, only a few percent of the stars have metallicities less than 0.25 solar, in contrast to the closed box models where this fraction is

130

0.8

0.6

,,0.4

:3 ~O.2

G. KAUFFMANN

Figure 3. The percentage of stars in a Milky Way-type disk with metallicity less than Z

Or ~

r'c~/ -1

8.5 9 9.5 10 10.5 11 loa (LiLa)

Figure 4. The mean metallicity of a galaxy (relative to solar) is plot.ted versus its B-band luminosity. The error bars indicate the scatter obtained for different disk formation histories. The dotted line is a fit to the observed relation from a compilation of data presented in a review article by Roberts & Haynes (1994)

almost always much higher.

As noted by White & Frenk (1991), the chemical evolution scheme described above naturally results in a metallicity-luminosity relation in the sense that more luminous galaxies are more metal rich. There are two reasons for this effect. One is that large galaxies form stars for a longer period of time. The most important effect, however, is that a larger fraction of the gas reservoir in larger galaxies is turned into stars by the present day. The metallicity-luminosity relation that we obtain is shown in figure 4. It is in reasonable agreement with a fit to the observed relation from a compilation of data from different sources (Oey & Kennicutt (1993); Garnett & Shields (1987); Skillman et at. (1989)) presented in a review article by Roberts & Haynes (1994).


3.2. EVOLUTION OF THE PROPERTIES OF DISK GALAXIES AT HIGH REDSHIFT

In the previous section, we demonstrated that our model is able to account successfully for many of the observed properties of disk galaxies at the present day. In this section, we extend our analysis to probe the properties of disk galaxies at high redshift.

The gas profiles of typical disk galaxies at z = 2.5 are plotted in figure 5. The upper panel shows the profile of a disk with circular velocity 220 km s-1 and the lower panel is for a disk with circular velocity 100 km s-1. Since we assume that the disk formation process is the same at all redshifts, it is no surprise that the gas surface densities once again track the critical density over most of the disk. The main difference is that disks formed at higher redshift are smaller and more concentrated, since their virial radii scale as (1 + z)-3/2. The SFRs of high-red shift disks again range from a few tenths to a few solar masses per year. However, as will be seen later, at z = 2.5 disks with circular velocities less than 100 km S-1 dominate the total absorption crossection. These galaxies are inferred to have star formation rates of only a few tenths of a solar mass per year. It is therefore not a surprise that it has proved difficult to detect these objects in emission (see for example Hu et al. 1993).

In order to calculate the total neutral hydrogen density f2(HI) contributed by damped Lya systems at a given redshift, one must know both the total amount of gas in each galaxy that contributes to the absorption, and the mass function of galaxies at that redshift. We will assume that the galaxies that contribute to damped Lya absorption form at the centres of dark matter halos with circular velocities in the range 35 to 300 km s-1. Figure 6 shows the contribution of galaxies of different circular velocities to the total crossectional area of gas at column densities greater than 2 x 1020

cm- 2 • Results have been plotted at four different redshifts. At z = 0, galaxies with circular velocities greater than 100 km s-1 make up about 50% of the total crossection; at z=2.5, this has decreased to 30% and by z=;{.8, only 20% of the absorption area is produced by these more massive galaxies. In principle, this is something that can now be tested with kinematical data derived from high-resolution spectra (Wolfe et al1994).

In figure 7, we show the evolution of f2(HI) with redshift for a series of CDM models with different normalizations. The data points are from Storrie-Lombardi & McMahon (1995) and are derived from a compilation of data from several different surveys. Their data includes 12 new highredshift damped systems discovered in the APM QSO survey, together with all other existing lower redshift samples (Wolfe et al. 1986; Lanzetta et al. 1991; Lanzetta, Wolfe & Turnshek 1(95). It should be noted that the new values of f2(HI) do not rise as steeply with increasing redshift as indicated

132 G.KAUFFMANN

22

21

19~~~~~~~~~~~~~~~~

024 Radius (kpc)

6 0

RadluB (kpc)

Figure 5. The gas surface density profiles of disk galaxies in halos of circular velocities 100 and 220 km S-1 at a redshift z = 2.5. The dotted line shows the critical surface density.

by previous data. There appears to be evidence for a flattening at z "" 2 and possibly even a turnover at z "" 3. It is interesting that the same qualitative trends are apparent in the evolution of n(HI) derived from the models. The most general conclusion is that the evolution predicted all the models is mild: n(HI) increases by at most a factor 3 from z=O to z=3. This increase comes about because of the increase in the number of halos of galactic mass, and because of the shift in the distribution of galaxies to less luminous systems that are also more gas-rich. The red shift at which n(HI) peaks depends on the normalization; models with high values of b have late structure formation and n(HI) peaks at low redshift. It is clear, however, that much more data is needed before any constraints can be placed on cosmology.

Finally in figure 8, we show how the metallicity distribution of damped Lya systems changes at high redshift. At z = 0 the metallicity distribution is sharply peaked at values just under solar. By a red shift of 2.5. the metallicity distribution is much more evenly spread. The mean value is about 0.1 solar, but values as low as 0.01 solar and as high as 0.7 solar are expected. This accords rather well with the data of Pettini et at. (1994), who find that the zinc abundances of damped systems at redshifts between 2 and 3 span a wide range. We also predict that at all redshifts, there should be a strong


0.4

g 0.3 z=o

~ ./:: 0.2

0.1

0.4

10.3

~ - 0.2

0.1

o 100 200 o 100 200 Vc (km/s) Vc (krnls)

Figure 6. The contribution of galaxies of different circular velocities t.o the t.otal crossectional area of HI gas in the universe with column densities greater than 2 x 1020 cm-2

Results are shown at four different redshifts.

0.004

-.::- 0.002 C-ol !>II ., Ei o o

o 2 Redshifl

4 6

Figure 7. The evolution of r2(HI) with redshift in COM models with different normalization. The solid line is for b = 1, the dotted line b = 1.5, the short. dashed line b = 2 and the long dashed line b = 2.5. The data points are taken from Storrie-Lombardi & McMahon (1995). The value adopted for Ho is 50 km S-l Mpc-1 .

134

0.4

0.3 .:: o :;:l o

~ 0.2

0.1

G. KAUFFMANN

z=O z=2.5

O .......... .!..I...JL..L.J...l..L..I..u...L..L..L..u.. ............................ .J....L.JU-L.a...w..w...L..L..L..J..J

-1.5 -1 -0.5 log(Z/Z.)

-1.5 -1 -0.5 log(Z/Zoo)

Figure 8. The metallicity distribution (relative to solar) of damped Lya systems at z = 0 and at z = 2.5.

correlation between the metallicity of a damped systems and its circular velocity, with more metal-rich systems having higher rotation speeds. This is again something that can be checked using high resolution spectra.

4. Discussion and Conclusions

The classical "closed box" approach to galaxy evolution is motivated by an Eggen, Lynden-Bell & Sandage-type picture of galaxy formation (1962), in which initially overdense regions in the early universe break away from the uniform Hubble expansion and then collapse monolithically into centrifugally supported gaseous disks. These disks then form stars over a Hubble time, becoming enriched in heavy elements in the process. The analysis in this paper is based on the White & Rees (1977) picture of galaxy formation, in which disk galaxies form by continuous cooling and accretion of gas within a merging hierarchy of dark matter halos. The Kennicutt law of star formation combined with this assumption of continuous infall results in gas in galactic disks being maintained at a critical threshold density E e, where Ec ex Vel R. Chemical enrichment of the disks takes place when the surrounding hot halo gas is enriched in heavy elements ejected during supernova explosions. This gas then cools onto the disk producing a new generation of metal-rich stars.


We have shown that models of this type can reproduce many of the observed properties of present-day galactic disks like the Milky Way. These include gas and stellar surface density profiles, metallicities and the distribution of stars as a function of age and metal content. In particular, we find that the rapid early enrichment predicted by our model solves the classic G-dwarf problem.

We then use the models to make some inferences about the properties of disk galaxies at high redshift. Because the overall mass distribution in the universe shifts to smaller halos at higher redshifts, and these smaller halos contain less luminous, more gas-rich galaxies, we find that the total neutral hydrogen density !1(HI) increases at higher z. The predicted increase, however, is rather mild, but is roughly consistent with the latest derivation of !1(HI) as a function of z by Storrie-Lombardi & McMahon (1995). Time will tell whether this model will still hold up when more data is accumulated. More extreme evolution would indica.te the Kennicutt law does not hold for galaxies at high redshift and that some extra physical process must cause star formation to be less efficient at high z than at present. It is encouraging, however, that our models are also able to account for some of the other trends seen in the high-redshift data, including the increase in the number of high column-density systems at high redshifts, as well as the metallicity distribution of damped Lya systems at z rv 2 - 3.

Finally, one rather general prediction of all hierarchical models is that the galaxies that give rise to the damped Lya absorption become progressively less luminous and more compact at higher redshift. This prediction will no doubt soon be tested by a new generation of telecopes and instruments capable of imaging galaxies as they were when the universe was young.

References

Barnes, J., Efstathiou, G.P., 1987, ApJ, 319, 575 Bergeron, J., 1988, in Kaiser,N., Lasenby, A., eds., The Post Recombination Universe,

Dordrecht:Kluwer, p201 Bond, J.R., Cole, S., Efstathiou, G., Kaiser, N., 1991, ApJ, 379, 440 Bosma, A., 1981, Astron J., 86, 1825 Bower, H.., 1991, MNRAS, 248, 332 Efstathiou, G.P., 1992, MNRAS, 256, p43 Eggen, O.J., Lynden-Bell, D., Sandage, A.R., 1962, ApJ, 136, 748 Fall, S.M., Pei, Y.C., 1993, ApJ, 402, 479 Garnett, D.R., Shields, G.A., 1987, ApJ, 317, 82 Hu, E.M., Songaila, A., Cowie, L.L., Hodapp, K.W., 1993, ApJ, 419, L13 Kauffmann, G., White, S.D.M., 1993, MNRAS, 261,921 Kennicutt, R.C., 1989, ApJ, 344, 685 Lacey, C.G. , Cole, S., 1994, MNRAS, 271, 676 Lanzetta, K.M., Wolfe, A.M., Turnshek, D.A., 1995, ApJ, 440, 435 Navarro, J.F., White, S.D.M., 1994, MNRAS, 267, 401

136 G. KAUFFMANN

Oey, M.S., Kennicutt, R.C., 1993, ApJ, 411, 137 Pei, Y.C. , Fall, S.M., 1995, ApJ, in press Pettini, M., Smith, L.J., Hunstead, R.W., King, D.L., 1994, ApJ, 426, 79 Pettini, M., Boksenberg, A., Hunstead, R.W., 1990, ApJ, 348, 48 Roberts, M.S., Haynes, M.P., 1994, Ann.Rev.Astron.Astr., 32, 115 Quirk, W.J., 1972, ApJ, 176, L9 Rao, S., Briggs, F.H., 1993, ApJ, 419, 515 Skillman, E.D., Kennicutt, R.C., Hodge, P.W., 1989, ApJ, 347, 875 Steidel, C.C., Dickinson, M., Persson, S.E., 1994, ApJ, 437, L75 Steinmetz, M., 1995, in Bender, R., Davies, R., eds., IAU Symposium 171: New Light on

Galaxy Evolution, Dordrecht: Kluwer, in press Storrie-Lombardi, L.S. , McMahon, R.G., 1995, preprint Toomre, A., 1964, ApJ, 139, 1217 Tyson, N.D., 1988, ApJ, 329, L57 White, S.D.M., Frenk, C.S., 1991, ApJ, 379, 52 White, S.D.M., Rees, M.J., 1978, MNRAS, 183, 341 Wolfe, A.M., Lanzetta, K.M., Foltz, C.B., Chaffee, F.H., 1995, preprint Wolfe, A.M., Fan, X.M., Tytler, D., Vogt, S.S., Keane, M.J., Lanzetta, K.M. 1994, ApJ,

435, LI01 Wolfe, A.M., Turnshek, D.A., Smith, H.E., Cohen, R.D., 1986, ApJ Supp, 61, 249 York, D.G., 1988 in Blades, J.C., Turnshek, A., Norman, C.A., eds., QSO Absorption

Lines: Probing the Universe, Cambridge University Press, p227

WARM GAS AT HIGH REDSHIFT

Clues to Gravitational Structure Formation from Optical Spectroscopy of Lyman a A bsorption Systems

MICHAEL RAUCH Astronomy Department 105-24, California Institute of Technology Pasadena, CA 91125, USA

Abstract. We discuss the effects of gravitational collapse on the shape of absorption line profiles for low column density (N(HI) < 1014 cm-2 ) Lyman a forest clouds and argue by comparison with cosmological simulations that Lyman a forest observations show the signs of ongoing gravitational structure formation at high redshift. The departures of observed line profiles from thermal Voigt profiles (caused by bulk motion of infalling gas and compressional heating) are evident from the results of profile fitting as a correlation in velocity space among pairs of components with discrepant Doppler parameters. This correlation also allows us to qualitatively understand the meaning of the Doppler parameter - column density (b-N HI)

diagram for intergalactic gas.

A part of this conference was devoted to the prospects of detecting neutral hydrogen at high redshift as tracers of the gaseous large scale structure (contributions by Braun, de Bruyn, Ingram, Swarup, and Weinberg). A successful detection with radio-astronomical techniques depends on the gas being in a state of high HI column density (N(HI) > 1018cm-2 ) and low temperature. We have reason to believe that gas condensations where such conditions prevail are rather rare in terms of geometric cross section, volume filling factor, and probably also in terms of the total fraction of baryonic matter they represent. Here we take a complementary point of view and ask what we can learn by observing more typical low column density, warm (T ...... a few X 104 K) gas at high redshift. To study such tenuous gas condensations (the so-called Lyman a forest) we need to look for the absorption imprinted by intervening gas onto the spectrum of a strong background source, an AGN or QSO. Typically, with a large optical telescope like the 10m Keck neutral hydrogen column densities down to 1012cm -2 can be

137

M. N. Bremer etal. (eds.), Cold Gas at High Redshift, 137-142. © 1996 Kluwer Academic Publishers.

138 MICHAEL RAUCH

measured in a few hours of observing time over a simultaneous redshift range of .6.z ",1, in front of a 17-18th magnitude QSO.

__ 3HI Jl7t! ... 72 l174 J81. IN'

.... ,...

Figure 1. Spectral regions from lines-of-sight through the z=2 simulation of Miralda-Escude et ai., showing absorption lines departing from single Voigt profiles. Several of the clouds show asymmetries and are better fitted by very close pairs of a broad and a narrow component.

What causes the Lyman a forest phenomenon? Recent numerical simulations of gravitational structure formation within the cold dark matter scenario (including the effects of gasdynamics and ionizing background radiation) have brought us closer to an answer to this question (Cen et al. 1994; Weinberg, this volume; Zhang et al., 1995; Miralda-Escude et aL 1995). According to these experiments gravitational collapse of baryonic matter produces extended (length scale of order 1 Mpc) condensations of gas giving rise to absorption phenomena very similar to the observed Lyman a forest. Typical low column density Lyman a clouds appear to be sheet-like or filamentary structures with relative overdensities of"" 1 - 10. During the epoch accessible to observation collapse of gas and accretion continue - we are watching gravitational structure formation in situ.

WARM GAS AT HIGH REDSHIFT 139

If this picture is right, what is the observational signature of this formation process? Assuming the clouds are really objects in a phase of collapse, then bulk motion and compressional heating of the infalling gas should affect the absorption line profiles of these clouds, producing departures from a Maxwellian velocity distribution. Line shapes are then no longer well represented by a Voigt profile (which would characterize a static, homogeneous temperature phase). Nevertheless, fits with multiple, pure Voigt profile models to entire high resolution Lyman C\' forest spectra (e.g. Carswell et at. 1991) have yielded excellent results statistically indistinguishable from the data, a statement which remains true even at the very high signalto-noise ratios (up to 100 or more) achievable with the Keck telescope (e.g. Tytler et ai. 1995). Thus, if a decomposition in terms of Voigt profiles works but the real absorption line shapes individually depart from such profiles, all the information about the physics must be contained in the correlations among various parameters (redshift, Doppler parameter, HI column density), analogous to a Fourier decomposition of a periodic function (though the analogy is limited, as Voigt profiles are not orthogonal base functions).

Figure 1 shows four regions from artificial spectra created from the simulation of Miralda-Escude et at. (1995) for redshift 2. Close inspection reveals a number of lines of intermediate strength (column densities below 1Q14cm-2) consisting of a narrow central core surrounded by broad, often asymmetric wings. Comparison with the physical parameters of the simulation shows that the wings are caused partly by the bulk motion of infalling gas and partly by temperature gradients due to compressional heating, although in individual cases it is difficult to disentangle the contributions to the width from temperature and bulk flow.

Modelling the simulated lines by Voigt profiles at an assumed signalto-noise ratio of 50 often requires two (occasionally more) components, close together in velocity space, a narrow one for the core and a broad one representing the non-maxwellian wings. Thus, one spectral signature of gravitational structure formation would be an anticorrelation of Doppler parameters for profile pairs at very small (0-30 kms- l ) separation. In Fig. 2 we plot the fraction of line profile pairs with Doppler parameters discrepant by more than a certain factor R, i.e. Ilog(bI/b2 )1 > log(R), as a function of the velocity splitting between the components of the pair. Data in the upper diagram are from the Miralda et at. z=2 simulation. As expected, there is a strong excess of pairs with very discrepant Doppler parameters at the smallest splittings. Dividing the sample into subsets with different column densities it can be shown that this signal is dominated by lines with column density N(HI) <1013.5 cm- 2 . If the combination of narrow and broad components were an artifact in that we would be substituting a broad component for a number of unresolved normally narrow ones in a blend, then

140

"

MICHAEL RAUCH

I.. .,

I-

1 -1

I

I I I -I

L _-I I-I .. ., - 1- I I .. ., I I I I L_ I __ ~"- _1- L"'-I

I I ~ r'" I I

- ~.J ~J 1.."

I_ I --1 1_ I--L~ _~ I ,_ ...... _ ~I 1- I I I .. ...

~I _J _J-I.._-

I I -1 _I I

Figure 2. Fraction of absorption line pairs with Doppler parameters different by more than a factor 3 (solid line) or a factor 1.5 (dashed line), versus the velocity splitting between the components, for absorption lines at redshift 2 from the Miralda-Escude et al. model. Apparently, at small splittings an absorption line can be decomposed into two close components with very discrepant Doppler parameters, indicative of a narrow line core with broad wings. Below, the same statistic for the real data from the Hu et al. (1995) Keck sample. There is indeed a signal similar to the one from the simulation.

we would expect to find the higher column density, more clustered absorption systems to provide the bulk of the correlation signal, contrary to what we are seeing here. The lower half of Fig.2 shows the same statistic for the actual observations from the work of Hu et al. (1995). The mean red shift here is higher «z> '" 2.86), and the line fitting was done somewhat differently (the deficit in the zero velocity bin is likely to be an artifact), but the overall picture, the discrepancy among Doppler parameters, increasing with decreasing velocity separation, as a signature of non-Voigt profiles, is

WARM GAS AT HIGH REDSHIFT

.' ."

.' . .. . . .. . :- ::

• +. t. :' :.: .

. ~,. ::. ,: .. ,.~::.. .. ~ .. . . . . : .... .:~ .. ::- .. '.: .. ::.: .. .. .. , 4'.:.,:~( '.,:" ., .. :'" .. ~ ......

•• .o¥ .... l.I-4 A "., .. ".... .. '. , ..... ~ " .... i.· .. S ........ ..... : •• ,...... .' • . ;:!\:,,~ • ., .... ,,"t.~ ~ ....•.. ::.~.':;4· ~~~.~~~~Ht,:.;,!~ .. ; .. : .

.. ,~~~ .:lTo • ~::;r;" .. .. ... .. .. .. .. ::~1· ::,'" .' ., ......... . . ,

log N(HI)

log N(HI)

141

Figure 3. Above: Doppler parameter b versus HI column density N(HI) for LyO' from 100 lines of sight through the Miralda et al. 10 Mpc box at z = 2. For clarity only the lower 100km s-1 are shown. Main features also seen in real data are: (1) the absence of lines below a minimum Doppler parameter for a given N(HI), a weak positive correlation between band N (HI), and a significant population of large Doppler parameter components at lower column densities (b's ranging up to several hundred km S-I). Below: b-N diagram of pairs of lines with velocity splitting < 20 km 13- 1 corresponding to the first two bins in Fig. 2. Each pair (i,j) is represent by a straight line, with [(b" N.),(bJ , NJ )) as its endpoints. At a velocity separation this close, the nearest neighbour of a narrow line close to the photoionization cutoff (lower envelope) is usually a broad companion (often with Doppler parameter larger than 100 km 8-1 ) rather than a component of random Doppler parameter.

similar. Obviously, the observed absorption line shapes are consistent with the expected line profiles from gravita,tional structure formation scenarios. Of course, this is not a proof that the structure formation scenario for intergalactic gas clouds is correct. We could postulate the (ad hoc) existence

142 MICHAEL RAUCH

of unresolved clustering of weak lines on very small scales to explain the presence of broad wings, but to emulate the large width of the broad components would require a rather contrived column density distribution for the cluster components - currently there is no physical justification for such a picture. The existence of galactic winds could be an alternative physical model which cannot be discussed on the basis of line profiles alone as we cannot distinguish between outflow or infall, but the presence of absorption coherent over Mpc scales (e.g. Dinshaw et al. 1995) makes this possibility less likely. In any case our above result is probably the first direct observational evidence we have of these objects being in a dynamic state.

The above result also helps to clarify the longstanding issue of the meaning ofthe supposedly "supra-thermal" Doppler parameters in the b - N(HI) diagram (Fig. 3). From what was said above it is clear that many of the broad low column density lines in this distribution are actually not independent systems but the bulk motion- and temperature-broadened wings of absorption lines from gas clouds with more quiescent gas in the center, the temperatures of which are consistent with photoionization heating plus some contribution from the potential well in which the gas has collapsed. Together with a second population of low column density clouds in voids with line profiles dominated by bulk motion (Miralda-Escude et al. 1995) these broad wings account for the high Doppler parameters in the b-N diagram. It is hoped that the future analysis of large Lyman a forest data samples at different redshifts, in connection with gasdynamical simulations will teach us more about the thermal history of cosmologically distributed gas.

I thank my collaborators on various projects, Bob Carswell, Renyue Cen, Len Cowie, Esther Hu,Tae-Sun Kim, Jordi Miralda-Escude, Jerry Ostriker, Wal Sargent, Tony Songaila for permission to quote from some of the results prior to publication.

References

Carswell, R.F., Lanzetta, K.M., Parnell, H.C., Webb, J.K., 1991, Astrophys. J.,371, 36. Cen, R., Miralda-Escude, J., Ostriker, J.P., Rauch, M., 1994, Astrophys. J.,437, L9. Dinshaw, N., Foltz, C.B., Impey, C.D., Weymann, R.J., Morris, S.L., 1995, Nature, 373,

223. Hu, E.M., Kim, T.-S., Cowie, L.L., Songaila, A., Rauch, M. 1994, Astrophys. J., 110,

1526. Miralda-Escude, J., Cen, R., Ostriker, J.P., Rauch, M., 1995, Astrophys. J., submitted. Tytler, D., Fan, X.-M., Buries, S., Cottrell, L., Davis, C., Kirkman, D., Zuo, L. 1995, in

QSO Absorption Lines, Proc. ESO Workshop, ed. G.Meylan (Heidelberg: Springer), p.289.

Zhang, Y., Anllinos, P., Norman, M.L., Astrophys. J., submitted.

G AS IN CLUSTERS

HI IMAGING OF CLUSTERS

JACQUELINE VAN GORKOM Columbia University Astronomy department 538 W 120th Street New York, NY10027

Abstract. With steadily increasing sensitivity and correlator capabilities the major synthesis instruments are now beginning to live up to their promise as H I surveying instruments. Pointed observations of galaxies in the nearest clusters of galaxies, such as Virgo, Coma and A1367, reveal shrunken H I disks in the cluster cores and detailed studies of the H I surface density distribution make it possible to identify the various gas removal processes at work. The Hydra cluster, 3 times as distant as Virgo, is distant enough that many galaxies fall within a single pointing of the VLA. This is the first cluster for which the entire volume has been searched for H I. The combined information on velocity, position and (lack of) H I deficiency of large numbers of spirals seen toward the core make it plausible that we see a chance super position of a group of spirals along the sightline toward the center of the cluster. The most distant cluster imaged in H I so far is Abell 2670 at z = 0.0767. The preliminary VLA results detect many gas rich spirals, which are exclusively located in the outer parts of the cluster. The results strengthen the suggestion (Bird 1994) that substructure is present and there is a hint of galaxies falling into the cluster along filaments.

1. Introduction

To study cold gas at high redshifts one could either be very macho, tune the LO's to as Iowa frequency as possible and start integrating or one could slowly move out, starting at z = 0 and pushing out as far as one can get. The story of H I imaging of clusters of galaxies is a story about the latter approach and it will be my story today. What I would like to concentrate on is the use of synthesis imaging instruments to do H I survey work.

145

M. N. Bremer et al. (eds.), Cold Gas at High Redshift, 145-157. © 1996 Kluwer Academic Publishers.

146 JACQUELINE VAN GORKOM

One of the best documented and least understood relations is the densitymorphology relation, the observation that ellipticals tend to inhibit the regions of high galaxian density, while spirals mostly populate the low density regions (Hubble and Humason 1933; Dressler 1980). Postman and Geller (1984) show that the relation extends over six orders of magnitude in space density: at densities less than 5 galaxies Mpc3 all population fractions show little density dependence, with typical values of 65% spirals, 23% SO's and 12% ellipticals, at densities between 5 and 600 galaxies Mpc-3 there is a dramatic decrease of the spiral population fraction, at densities greater than 600 galaxies Mpc-3 SO's dominate the galaxy population, while at densities greater than 3000 galaxies Mpc-3 the fraction of ellipticals rises steeply. Attempts to explain the observed relation range from purely nature, where environmental effects during galaxy formation, in particular local density, influence morphological type (e.g. Evrard, Silk and Szalay 1990) to purely nurture, where it is the environment, particularly the presence of hot intergalactic gas, a high merger rate and the tidal field of the cluster potential that affects the evolution of galaxies (e.g. the review by Dressler 1984). HI imaging studies can contribute to the resolution of this ongoing debate in several ways: detailed studies of galaxies of the same morphological type in a wide range of environments might identify the processes at work that affect the evolution of galaxies differently in different environments (e.g. Cayatte et ai. 1993), while the large field of view of synthesis imaging instruments, especially at higher z, can be used to explore the larger scale distribution of gas rich systems irrespective of their optical luminosity thus giving a complementary perspective on the density morphology relation.

Here I shall first discuss what we have learned from H I studies of well defined samples of individual galaxies in l nearby clusters (Section 2), then I will discuss some efforts, in which a significant fraction of a cluster volume was probed for H I (Section 3), followed by a presentation of the first preliminary results on the most distant cluster imaged so far, Abell 2670 at a redshift of 0.08 (Section 4).

2. The most nearby clusters

H I studies of nearby clusters started in the early seventies (Davies and Lewis 1973; Bottinelli and Gouguenheim 1974). These studies were performed with single dish telescopes and aimed at comparing the neutral hydrogen content of galaxies in clusters with that of galaxies in the field. By 1980 overwhelming evidence had been presented that at least the spirals in the Virgo cluster were hydrogen deficient (Chamaraux, Balkowski and Gerard 1980). The single dish work culminated in a paper by Giovanelli and Haynes (1985), who analyzed high quality H I data on a sample of nine

H I IMAGING OF CLUSTERS 147

clusters. A significant H I deficiency was found in 5 out of 9 clusters. The three clusters containing normal H I contents are loosely organized, contain a higher proportion of spirals and are not X-ray sources. In a series of imaginative displays these authors illustrate that the fraction of H I deficient galaxies is correlated with the X-ray luminosity of the cluster and that the H I deficient galaxies are strongly clustered toward the centers of the clusters. The interpretation of these data is not straightforward. A further analysis of this sample by Dressler (1986) shows that it is almost exclusively the earlier spiral types that are H I deficient and that these galaxies are on radial orbits through the cluster. The later types are not deficient and most likely on circular orbits which never go through the center of the cluster. The simplest interpretation of these data is that ram pressure stripping of spiral galaxies on radial orbits is responsible for the H I deficiency. The data are consistent with a larger fraction of early-type galaxies on radial orbits than for late-types.

It was around this time that the first H I images appeared in the literature. The first H I image of Virgo, a composite of the 10 optically most luminous spirals, showed strikingly that the H I disks of the galaxies within 3° of M87 (the center of the cluster) were much smaller than the optical disks, beyond 3° they had almost normal extents (van Gorkom, Balkowski and Kotanyi, 1984). More comprehensive surveys done with the VLA and the WSRT fully confirmed those first VLA results (Warmels 1988a, 1988b, Cayatte et al. 1990, Cayatte et al. 1994). Similarly shrunken H I disks have also been found in Coma (Sullivan 1989) and in A1367 (Dickey and Gavazzi 1991).

Figure 1 (taken from Cayatte et al. 1990) shows images of the H I emission from the 25 optically brightest spirals near the center of Virgo. The image shows the galaxies at their proper location in the cluster, but each H I image has been magnified by a factor five. It illustrates that the HI disks close to M87 are tiny compared to those further out, while optically the galaxies have quite similar sizes. Although this had been inferred indirectly from single dish work (where an H I model distribution had to be assumed (Giovanelli and Haynes 1983)) it was the isophotal images of the H I distribution such as the one shown in Fig. 1 a.nd especially as overlaid on optical images (Warmels 1988a, Cayatte et al. 1990) that emphasized how dramatic the H I disks get affected by the cluster environment. It was that picture that made Cayatte et al. 1993 realize that the single most important signature of environmental impact was the size of the H I disks compared to the optical size. The galaxies can be divided into four groups, according to the ratio of the H I to optical diameter. Group I has a ratio larger than 1, group II has a ratio close to 1, group III has a ratio less than 0.75 and group IV consists of optically selected galaxies, the so called anemics (van

148

17'

16'

15'

N ~689

1~' ~ N4639

~ @jPt - lr N4654

c: 0 -0 c:

12' N4641 @ u

" c N4694 ... ~ 11'

10'

9'

12h 4510 40 m

JACQUELINE VAN GORKOM

.N454B ~

~ N4569

(-11 N4519

Nmo~

N4321 @

~~SOl

I D5S N41J8 ,'b e_ N 4402

N4388

M 81

N ~237 <V

N4192 1 N42S4~

N 4222 ~\\

N4216 VI

\?) N 4568

N 4178 ~

i N4535

35,. 30" 25m 20 .. 15m 10 .. R.lght AscenSIon [1950'

Figure 1. Integrated neutral hydrogen contour images of the brightest spirals in the Virgo cluster center. Each image has been drawn at the galaxy position, indicated by a cross, and magnified by a factor 5 compared with the coordinate scale. The first contour in each image corresponds to about 1020 cm-2

den Bergh 1976). As it turns out the groups have several other common characteristics that set them apart. The galaxies in group I are in the outer parts of the cluster, and have a normal radial H I surface density distribution (Fig. 2), for group II the surface density is low across the entire face of the galaxy, while the small H I disks have a normal to slightly elevated central H I surface density and sharply truncated outer disks. Group I is not H I deficient, in group II mildly and group III strongly. Figure 2 shows the radial surface density distribution arranged by group and normalized by surface density profiles for field spirals of the same morphological type.

Objects in group III are found very close to the center only and have large radial velocities, thus they probably are pa.ssing through the center

" \

H I IMAGING OF CLUSTERS

--.. -_0

fig" .. 5

1: ., f I~ ,.

,-,'''''-', - ".' -...... '\" I' ...... ., \,

I ,. \

!",/ " ~ -_ .. _-

149

Figure 2. H I surface density profiles normalized by an averaged surface density according to the morphological type of each group. The radius is normalized by the optical corrected radius at the 25th magnitude arcsec-2 isophote. The solid line represents the normalized profile of a typical field galaxy

on radial orbits. Cayatte et al. (1993) infer that group III is affected by ram pressure sweeping, where gas gets removed from the outer parts of the disks, but if anything gets enhanced in the central part. These galaxies also have a normal CO content in the inner parts. The fact that the mildly H I deficient galaxies in group II have been affected across the entire face of the disks suggests that whatever process was at work, the local gravity of the galaxy doesnot play much of a role in counteracting the removal process. This suggests that transport processes, such as turbulent viscosity and thermal conduction may have removed some of the gas. The majority of the galaxies are in group I, a group which is hardly if at all affected by the cluster environment. The anemic galaxies may be completely stripped galaxies that currently rebuilding their disks from gas shed by the stars. Thus the imaging observations are useful to identify the different removal processes at work. Figures likes Fig. 1 also make it abundantly clear that only a tiny fraction of the galaxies is affected, ie the picture of Tully and Shaya (1984) in which the majority of the ellipticals in Virgo have been there for a long time, while most of the spirals have entered the cluster in


the last one third to one half of the age of the universe is probably correct.

A quite different result has recently been obtained on the Ursa Major cluster, in a survey similar to the ones mentioned above and currently being carried with the WSRT (Verheijen, this conference). Ursa Major is at a similar distance as Virgo, but contrary to Virgo it has no central Xray source and hardly any ellipticals. Thus it will form ideal comparison material to the Virgo cluster. The results indicate that the galaxies in Ursa Major are not H I deficient at all (Verheijen, this conference), which is perhaps not too surprising considering the lack of a hot IGM (Giovanelli and Haynes 1985). The data will provide a unique data set to study the effects of a high galaxy density per se on the properties of spiral galaxies and to study, what is perhaps a young cluster.

3. Clusters out to z=O.035

The large field of view combined with the high angular resolution makes synthesis imaging instruments ideally suited to survey clusters of galaxies. However due to the poor sensitivity and the fact that galaxies in clusters tend to be H I deficient little was achieved in terms of detections until the VLA came on line. The first imaging survey that succesfully used the large field of view to simultaneously detect many galaxies within the primary beam were the observations of the Hercules cluster at z = 0.035 by Salpeter and Dickey (1985). Using only three pointings the entire core of this spiral-rich unrelaxed cluster was covered. These observations, though suffering from poor sensitivity (the 50" H I mass sensitivity limit is 1.4 X 109 h-2 Mev) and poor velocity resolution (175 km s-1), showed the potential for H Imaging survey work of clusters. In total 31 spirals were detected,though most of them only at the 2 or 30" level. Most importantly though, it was shown that the VLA was sensitive enough to detect large numbers of spiral galaxies at z = 0.035 and that even the largest single dishes (ie Arecibo) were beginning to suffer from H I confusion within the beam for clusters at those distances. Dickey (1994) is currently using the much improved sensitivity of the VLA to redo the project and the (preliminary) results look very impressive indeed.

The first survey that searched the entire volume of a cluster for H I is the VLA survey of the Hydra I cluster of galaxies (McMahon 1993). Hydra I (Abell 1060) is at a redshift of 0.011 (v = 3400kms-1). It is classified as irregular, has a richness class of 1 and is a Bautz-Morgan type III, it has a large cD-like elliptical, NGC 3311, at its dynamical center. Hydra I is remarkably similar to the Virgo cluster in size, shape, population and kinematics, with the important difference that Hydra appears more relaxed and dynamically isolated. It has a central X-ray source, but its luminosity


of Lx = 2.37 X 1043 erg s-l is low compared to most X-ray clusters(Jones and Forman 1984). Its similarity to Virgo makes comparison of the HI properties of the two clusters particularly interesting. Contrary to Virgo, Hydra is distant enough that its area can be surveyed in only 14 VLA pointings. McMahon (1993) surveyed almost the entire volume of Hydra down to a 5a mass limit of 4 X 107 M0 of H I with a velocity resolution of 42 km s-l. A total of 49 galaxies were detected, Figure 3 shows the total H I emission of the Hydra cluster. In comparing this with Fig. 1, one should keep in mind that Fig. 3 shows all the H I that is present in the entire volume of Hydra, while Fig. 1 shows the H I present in small volumes around the optically most luminous galaxies in Virgo. The distance to Hydra is roughly ;{ times the distance to Virgo, so while it was found that the radius of the stripping zone in Virgo was 30 , we may expect in Hydra to see severely stripped galaxies out to a distance of 10 from NGC 3311, which is indicated by a cross.

Figure 3 shows a number of interesting things. The radial di!:itribution of detections appears to be flat out to a degree from the center. A careful analysis (McMahon 1993) confirms what a first look suggests, there is no trend of H I disk size or deficiency with distance from the center. In fact, NGC 3312, the bright spiral closest to NGC 3:311 has an H I extent, which is similar to its optical size, quite unlike the bright spirals in the center of Virgo. Considering the presence of a hot IGM in the center of Hydra it seems likely that NGC 3312 is in projected distance only close to NGC 331l. These first results led McMahon et aZ. (1992) to suggest that NGC 3312 and NGC 3314a are part of a foreground group to the Hydra cluster. The more general result now emerging from Fig. ~{ is that in Hydra currently no spiral galaxies seem to be crossing the center.

The distribution of detections is not homogeneous, but rather the galaxies appear to be clumped in groups of 3 to 5 (spatially and in velocity). Outside the center 50 arcmin dwarfs seem to be associated with big spirals, with a hint that the large spirals furthest from the center have more companions. Eight uncatalogued dwarf galaxies are detected in the central field, but at extreme velocities. Although a more careful analysis of these data are still underway, the results seem to suggest that considerable substructure exists in Hydra.

The velocity structure of Hydra is particularly intriguing. Kurtz et aZ. (1985) comparing the peculiar properties of the X-ray cluster Abell 744 with those of Hydra note that both have inside their core radii higher velocity dispersions and a distinctly non gaussian velocity distribution, while the velocity distribution outside the core radii is gaussian. The discrepant velocity distribution in the core of Hydra, turns ont to be entirely due to the II I detected spirals, which have in fact a bimodal distribution. This


Hydra I Cluster Total HI Emission

-25 00 a~ -~ Q

30

\9

-26 00 ~~

....... 0 II) 01 ~

30 c 0 • ;:: i ca .!: 0

U • • GI 0 C

-27 00

<Ii' II Cl +

(j

f) 0 o 0

tI 0

30 • !I 0

<&> 0

0c. • -28 00 P

10 39 38 37 36 35 34 33 32 31

Right Ascension (1950)

Figure 3. Tbe total H I emission from tbe entire volume of tbe Hydra cluster. Contours begin at 3 x 1019 cm-2 • Tbe cross indicates tbe position oft.be central cD galaxy, NGC 3311


combined with the fact that the spirals are not H I deficient suggests as possible interpretations that the spirals either form a subgroup falling in from behind or a foreground group coincidentally centered on the cluster.

4. Imaging of Abell 2670 at z=0.0767

By far the best evidence for evolutionary effects in clusters (ie nurture comes from observations of clusters at moderately high red shift , typically z = 0.5. Butcher and Oemler (1978, 1984) have shown that the fraction of blue galaxies in clusters increases with increasing redshift. In addition a larger number of post starburst (E+A) galaxies is seen (Dressler and Gunn 1983, Couch and Sharpless 1987), while recent HST imaging of these clusters (Dressler et al. 1994, Couch et al. 1994) suggest that interactions or mergers may be responsible for some of the activity. Thus it would be particularly exciting to image clusters in H lover a range of redshifts, from o to 0.5, to see if the gas rich component of clusters evolves over these time scales. Although still at very modest distances a first data point is currently being obtained with the VLA, where a deep H I image has been made of Abell 2670 (van Gorkom, Dwarakanath and Guhathakurta, 1996).

Why Abell 2670'1 It is one of the most distant clusters for which photometry and spectroscopy of a large number (221) galaxies is available (Sharpless, Ellis and Gray 1988). It is the nearest Abell cluster of richness class 3, a BM I cluster (Bautz and Morgan 1970) with a supergiant cD galaxy in its center and an extended X-ray source. The velocity dispersion is 1038 km s-l. The velocity of the cD galaxy, 23282 =f 100 km s-l, differs significantly from the cluster mean of 22843 =f 64 km s-l, indicating that the cD may not be at the dynamical center (Sharpless et al. 1988). Most authors (e.g. Sharpless et al. 1988, Zabludoff, Franx and Geller 1993 and references therein) conclude nevertheless that there is no statistical evidence for sub clustering in A2670. However Bird (1994), analyzing velocity and position data, finds evidence for the merger of two or three distinct subsystems. Scheick and Kuhn (1994) conclude from the smoothness of the diffuse light in the cD envelope, extending out to 230 h-1 kpc, that the cD envelope is old (greater than 1010 year, which would be hard to explain in the presence of interacting subclumps.

At a redshift of about 0.1, an Abell radius is 10 arcmin, thus an entire cluster fits comfortably within the halfwidth (30 arcmin) of the VLA primary beam. The observations consist of two pointings, one centered at the center of the cluster and covering the entire velocity range of 4400 km S-l and one centered 30 arcmin to the north east of the cluster, covering 2600 km s-l. The spatial resolution is 20 arcsec (20 h- 1 kpc) and 47 km s-l. Till so far half the data have been obtained and the 5a detection limit is

Figure 4. Contour image of the neutral hydrogen emission from Abell 2670. The first contour corresponds to aboll t 1020 cm -2. Crosses indicate positions of optically cataloged galaxies. Letters indicate membership of sub clumps as defined by Bird (1994), the solid triangles indicate the dynamical centers of the 3 subclumps.

5 X 108 M8 of II I, The total II I image of the cluster, of the center pointing only, is shown in Fig. 4, crosses indicate positions of optically cataloged galaxies. The first thing that one notes is that no galaxies within 250 h-1

kpc from the center of the cluster have been detected in II 1. For comparison in Virgo the galaxy closest to M87 that has been detected in H I is at a projected distance of 100 kpc. As in Virgo, the galaxy closest to the central cD, which has been detected in H I shows a remarkably asymmetric HI distribution, displaced from the optical away from the center of the cluster (Fig. 5). As if the galaxy were entering the hot X-ray gas now and is in the process of being ram pressure swept. Of special interest is the group of spirals detected to the north east of the cluster and shown in Fig. 6. These galaxies are extremely gas rich, with H I masses up to 1010 M 8 . The velocity dispersion of the group is small of order 150 km s-1, possibly creating the


5 6 7

~10 37 15

30

S" to 45 en ,... -c

i c 38 00

~ 15

23 51 40 39 38 37 36 35 Right Ascension (1950)

Figure 5. An overlay of the H I emission (contours) on an optical image (greyscale) from the digitized POSS of the nearest galaxy to the cluster core detected in H I. The projected distance to the central cD galaxy is 250 kpc. Note the asymmetric H I distribution. Contours begin at 3 x 1019 cm-2 •

ideal environment for the galaxies to merge before they enter the cluster. An intriguing result is that the mean velocity of this group is 23242 km s-l, similar to that ofthe cD and 450 km s-l higher than the cluster mean.

The H I results provide additional support for the suggestion by Bird (1994) that the cluster consists of 3 distinct subsystems. The clumps are spatially almost overlapping, but are distinct in velocity. She suggests that clump C, the most massive of the clumps, is located at what will be the center of the cluster, while A and B are still faUing in. Interestingly the H I detections in the 3 clumps are spatially distinct. In Fig. 4 we have indicated clump membership for each of the spirals with an A,B or C. The dynamical centers of the clumps, as derived by Bird (1994) are indicated by the filled triangle. The least massive clump with the lowest velocity dispersion, clump A, is by far the most H I rich. The non uniform distribution of galaxies around the cluster is real. Although the sensitivity of the HI observations decreases going outward, it is circularly symmetric around the cluster center. By looking at this image one could imagine that clumps A and B are in fact parts of filaments, that are corning together in cluster. In that scenario the cD has forrr.ed through some mergers in filament A and is now making its way to the cluster center.


·10 32 30

3300

30

- 3400 0 m Q) .... 30 -c .2 10 3500 .s "i.'.i IV 30 0

3600

30

3700

23 52 20 16 10 0$ 00

Right Ascension (1950)

Figure 6. The H I emission (contours) of the gas rich group to the NE of Abell 2670 at a mean velocity of 23242 km S-1 overlaid on an optical image (greyscale) from the digitized POSS. These galaxies have H I masses of 1010 M 8 and extent from 50 to 100 kpc. Contours begin at 6 x 1019 cm-2 .

Although this is still a very tentative result, it shows that detecting HI in emission at redshifts of 0.1 has become entirely feasible. Soon the WSRT and GMRT will be able to go considerably beyond that, out to z = 0.4 (or even further?). It also demonstrates that even a very rich seemingly relaxed cluster has lots of H 1. Locating the H I rich galaxies spatially and in velocity will help in distentangling the 3 dimensional distribution of the galaxies, since the mere presence of H I indicates that a galaxy is not in the core and most likely has not gone through it.


Acknowledgements. I thank Dwarakanath and Raja Guhathakurta for their collaboration on Abell 2670 and for allowing me to present the preliminary results here. This research was partly funded by NSF.

References

Bautz, L. P. and Morgan, W. W. 1970, ApiL 162, L49 Bird, C. 1994, Ap I 422, 480 Bottinelli, L. and Gouguenheim, L. 1974, AA 36, 461 Butcher, H. R. and Oemler, A. Jr. 1978, ApI 219, 18 Butcher, H. R. and Oemler, A. Jr. 1984, ApI 285,426 Cayat.te, V., van Gorkom, J. H., Balkowski, C. and Kotanyi, C. 1990, AI, 100, 604 Cayatte, V., Kotanyi, C., Balkowski, C. and van Gorkom, J. H. 1994, AI, 107, 1003 Couch, W. J. and Sharpless, R. M. 1987, MNRAS 229,423 Couch, W. J., Ellis, R. S., Sharpless, R. M. and Smail, l. 1994, ApI430, 121 Davies, R. D. and Lewis, B. M. 1973, MNRAS, 165, 231 Dickey, J. M. and Gavazzi, G. 1991, ApI 373,347 Dickey, J. M. 1994, BAAS Dressler, A. 1980, ApI 236351 Dressler, A. 1984, ARAA 22 185 Dressler, A. 1986, ApI 301 35 Dressler, A. and Gunn J. E. 1983, ApI 220,7 Dressler, A., Oemler, A., Butcher, H. R. and Gunn, J. E. 1994, ApI, 430, 107 Giovanelli, R. and Haynes, M. P. 1983, Al 88, 881 Giovanelli, R. and Haynes, M. P. 1985, ApI 292,404 Jones, C. and Forman, W. 1984, ApI276, 38 Evrard, A. E., Silk J. and Szalay, A. S. 1990, ApI365, 13 Hubble E. and Humason M. L. 1931, ApI 74,43 McMahon, P. M. 1993, PhD thesis, Columbia University McMahon, P. M., Richter, O.G., van Gorkom, J. H. and Ferguson, H. C. 1992, Al 103,

399 Postman M. and Geller M. J. 1984, ApI 281,95 Salpet.er, E. E. and Dickey, J. M. 1985, ApI 292, 426 Scheick, X. and Kuhn, J. R. 1994, ApI 423, 566 Sharpless, R. M., Ellis, R. S. and Gray, P. M. 1988, MNRAS 231,479 Sullivan, W. T. 1989, in The World of Galaxies, ed H. G. Corwin and L. Bot.tinelli (New

York: Springer), p404 Tully, R. B. and Shaya, E. J. 1984, ApI 281,67 van den Bergh, S. 1976, ApI 206, 883 van Gorkom, J. H., Balkowski, C. and Kotanyi, C. G. 1984, in Clusters and Groups of

GalaXies, edited by F. Mardirossian, G. Giuricin and M. Mezetti (Reidel, Dordrecht), p. 261

van Gorkom, J. H., Dwarakanath, K. S. and Guhathakurt.a, P. 1996, in prep. Verheijen, M., 1996, t.his conference Warmels, R. H. 1988a, AAS 72,19 Warmels, R. H. 1988b, AAS 72,57 Zabludoff, A.I., Franx, M. and Geller, M.J. 1993, Ap I 410,47

AN HI SURVEY OF THE BOOTES VOID

A. SZOMORU Kapteyn Astronomical Institute P. O. Box 800, 9100 A V Groningen The Netherlands

Abstract. We discuss the results of a VLA 1 (Napier et al. 1983) HI survey of the Bootes void and compare the distribution and HI properties of the void galaxies to those of galaxies found in a survey of regions of mean cosmic density. The Bootes survey covers 1100 Mpc3 , or rv 1% of the volume of the void and consists of 24 cubes of typically 2 Mpc X 2 Mpc X 1280 km s-1 , centered on optically known galaxies. Sixteen targets were detected in HI; 18 previously un catalogued objects were discovered directly in HI. The control sample consists of 12 cubes centered on IRAS selected galaxies with FIR luminosities similar to those of the Bootes targets and located in regions of 1 to 2 times the cosmic mean density. In addition to the 12 targets 29 companions were detected in HI. We find that the number of galaxies within 1 Mpc of the targets is the same to within a factor of two for void and control samples, and thus that the small scale clustering of galaxies is the same in regions that differ by a factor of rv 6 in density on larger scales. The galaxies found in the void are mostly late-type, gas rich systems. A careful scrutiny of their HI and optical properties shows them to be very similar to field galaxies of the same morphological type. This, combined with our finding that the small scale clustering of the galaxies in the void is the same as in the field, suggests that it is the near environment that mostly affects the evolution of galaxies.

1. Introduction

The possible existence of a very large void in the distribution of optically bright galaxies in the direction of Bootes was first discussed by Kirshner et

IThe NRAO is operated by Associated Universities, Inc., under a cooperative agreement with the National Science Foundation.

159

M. N. Bremer etal. (eds.). Cold Gas at High Redshift. 159-164. © 1996 Kluwer Academic Publishers.

160 A. SZOMORU

al. (1981). Redshift measurements of galaxies in three small fields 35 degrees apart showed an identical gap in the galaxy distribution around 15000 km s-1. A subsequent redshift survey of galaxies in 283 fields distributed between the original three confirmed the existence of this void (Kirshner et al. 1987). Assuming a uniform galaxy distribution, 31 galaxies should have been detected within the redshift range of the void. As none were found, they concluded that the density of the void can be no higher than onequarter of the cosmic mean. The largest sphere devoid of galaxies consistent with their data has a radius of 31.5 Mpc and is centered at O! = 14h50m , ~ = + 46°, at a mean redshift of 15500 km s-1 , which implies a volume of 1.3 X

105 Mpc3 . This is what is commonly called the Bootes void.

A large number of investigations followed this discovery. Two previously cataloged emission line galaxies were found to lie within the boundaries of the void. In objective prism surveys eight more emission line galaxies were detected (Sanduleak & Pesch 1982, 1987, Moody 1986, Tifft et al. 1986, Moody et al. 1987, Weistrop 1987 and Weistrop & Downes 1988). Strauss & Huchra (1988) and Dey et al. (1990) discovered 10 more void galaxies, selecting their candidates from the IRAS Point Source Catalog (PSC). Dey et al. find the density of IRAS galaxies in the void to lie between 1/6 and 1/3 of the mean, consistent with the original estimate of the void density by Kirshner et al. (1987). Bothun & Aldering (1988) and Aldering et al. (1996) have reported on a redshift survey ofthe void based on sources selected from the IRAS Faint Source Survey (FSS) data base. We have added to these a search for HI rich objects in the vicinity of known void members (Szomoru et al. 1993, Szomoru, van Gorkom & Gregg 1996, Szomoru et al. 1996). At the time of our first HI observations (December 1989), only 12 emission line galaxies were known within the boundaries of the void. Surveys based on the PSC and FSS and our own HI survey have since raised this number to a total of 58.

While the influence of environment on HI properties is well established for regions of high density (van Gorkom 1993), little is known about the effects of a low density environment on the formation and evolution of galaxies. Hoffman et al. (1992) suggest that initial conditions may have a strong influence on the structure of galaxies. In their models dwarf galaxies (10" fluctuations) are hardly affected by their environment. However, rare 30" peaks within voids will develop into massive low surface brightness galaxies like Malin I (Bothun et at. 1987). Brainerd & Villumsen (1992) use N -body simulations to determine the mass function of dark galaxy halos as a function of environment. They find that the formation of massive galaxy halos is strongly biased towards high-density regions and predict voids to be filled with low mass galaxy halos. Finally, Lacey et al. (1993) model the evolution of the galaxy luminosity function, assuming the rate of

AN HI SURVEY OF THE BOOTES VOID 161

RA

Figure 1. Observed fields in the Bootes void. The dashed line outlines the void, the numbers indicate the number of detected galaxies per field.

star formation to be controlled by the frequency of tidal interactions with neighboring galaxies. Their model implies that luminous galaxies can not have formed in underdense regions because of a lack of tidal interactions needed to trigger star formation. Instead, there would be dark galaxies with all their baryons in the form of cool gas. Indeed, Bothun et al. (1993) find that low surface brightness galaxies (LSB) have a deficit of nearby companion galaxies (at projected separations less than 0.5 Mpc) compared to high surface brightness galaxies (HSB). At larger radii the respective distributions begin to merge and on scales 2': 5 Mpc, LSB galaxies trace out the same structures as HSB galaxies. They argue, as do Lacey et al. (1993), that the lack of tidal interactions has served to suppress star formation.

Although the Bootes void galaxies are found in a large-scale low density region, their immediate neighbourhood is not necessarily underdense. The evolution of voids and the existence of substructure within them has been addressed by Dubinski et al. (1993) and van de Weygaert & van Kampen (1993) by means of N-body simulations. Dubinski et al. (1993) find that smaller scale voids disappear within larger voids but that frozen-in remnants of small void walls lead to small-scale substructure within large voids. Van de Weygaert & van Kampen (1993) find a void hierarchy inside proto-voids, which may survive as substructure in large voids until the present epoch.

162 A. SZOMORU

2. The Survey

Our HI survey of the Bootes void consisted of pointed observations of known void galaxies, with the VLA in D-configuration (Figure 1). Since the field of view of the VLA is about 2 X 2 Mpc at the distance of the Bootes void, each pointing provided us with data on the HI content of a large volume, typically 30 Mpc3 around each target galaxy. Optical observations of a number of Bootes void galaxies were carried out at Kitt Peak and La Palma, using the 0.9m telescope and the 1m Jacobus Kapteyn Telescope respectively. The motivation behind this project was to determine whether the HI properties of void galaxies are significantly different from those of cluster and field galaxies, and to search the void for uncataloged late-type systems and neutral gas clouds in the vicinity of the known void galaxies.

Sixteen targets were detected in HI; 18 previously uncataloged objects were discovered directly in HI. The galaxies found in the void are mostly late-type, gas rich systems. Many of the galaxies are large; however, optical sizes D 25 range an the way from '"V 9 kpc to > 30 kpc. The largest spirals in our sample are clearly massive objects, with rotational velocities of 200-250 km s-l . In nearly an cases the HI distributions are irregular. Optically, a large number of the void galaxies have either asymmetrical isophotes or are irregular in appearance, and a number show clear signs of interaction in the form of HI tidal tails and optical debris. Many objects are clustered in small groups of two to five galaxies at small separations. We find that the HI masses are similar for both targets and companions, but that the samples are clearly segregated in luminosity and central surface brightness, with the targets making up the bright end. The target galaxies are overluminous by several magnitudes with respect to the blue Tully-Fisher relation.

3. The Control Sample

A large amount of data is available in the literature to compare the intrinsic properties of our Bootes sample to those of galaxies in denser environments. From this comparison we find that the properties of the Bootes galaxies are similar to those of late-type field galaxies. Very little is known however about the clustering properties of IRAS selected samples on scales provided by the field of view of the VLA. To address this issue, we have done VLA HI observations of a sample of 12 IRAS galaxies with the same IR luminosity distribution as in our Bootes void sample, but in regions whose density smoothed on large scales is close to the mean. These galaxies were selected from the full-sky redshift survey of IRAS galaxies complete to 1.2 Jy at 60 l.Lm (Strauss et al. 1990, 1992b, Fisher et al. 1995). As the average flux is appreciably higher than that of the Bootes sample, this control sample is much closer, with a mean around 8000 km s-l . The local density relative

0.. 4 >< Q)

'-" Z o

AN HI SURVEY OF THE BOOTES VOID 163

... Bootes

[J Control

... ratio ... ... ...

... ... ...

[J [J [J [J [J [J

[J [J ... [J

~ ... ... ... ...

10

Figure 2. Ratio of the number of expected and observed galaxies with HI masses ~ the ordinal value, for the control sample (open squares) and Bootes sample (filled t.riangles). The stars show the ratio of expected and observed galaxies ill the Bootes sample normalized to the control sample.

to the mean around each galaxy was defined using a top-hat smoothing with a radius of 1100 km s-1 (Strauss et ai. 1992a).

All targets were detected; a total of 29 companion galaxies were found, many of which were previously uncataloged. Excluding the companions that could not have been observed in both samples, due to differences in distance and sensitivity, we find that the Bootes and control samples contain practically the same number of companions: 11 VR. 10. Considering that the void is underdense by a factor of 3-6, and the control sample lies in regions of rv 1.5 the mean density, we would expect 4.5-9.5 times more companions in the control sample; instead we find a factor of 2.

In a more quantitative fashion we can calculate the expected number of companions based on the two-point correlation function (Davis & Peebles 1983) and the HI mass function (Briggs & Rao 1993), and compare this expected number to the observed number. The result is shown in Figure 2; the ratio of expected to observed companions differs by only a factor of two for the two samples.

The resulting picture is that these galaxies have formed as normal field galaxies in local density enhancements within the void. These enhancements may be part of larger substructures in the void, comparable to that seen in various N -body simulations. In this picture, the void galaxies outline smaller, "true" voids within the Bootes void.

164 A. SZOMORU

References

Aldering, G., Bothun, G.D., Kirshner, R.P. and Marzke, R. 1996, in preparation. Bothun, G.D., Impey, C.D., Malin, D.F. and Mould, J.R. 1987, AJ, 94, 23. Bothun, G.D. and Aldering, G. 1988, BAAS, 20, 1087. Bothun, G.D., Schombert, J.M., Impey, C.D., Sprayberry, D. and McGaugh, S.S. 1993,

AJ, 106, 530. Brainerd, T.G. and Villumsen, J.V. 1992, ApJ, 394, 409. Briggs, F.H. and Rao, S. 1993, ApJ, 419,515. Davis, M., and Peebles, P.J.E. 1983, ApJ, 267, 465. Dey, A., Strauss, M.A. and Huchra, J.P. 1990, AJ, 99, 463. Dubinski, J., da Costa, L.N., Goldwirth, D.S. and Piran, T. 1993, ApJ, 410, 458. Fisher, K.B., Huchra, J.P., Strauss, M.A., Davis, M., Yahil, A. and Schlegel, D. 1995,

ApJS, in press. Hoffman, Y., Silk, J. and Wyse, R.F.G. 1992, ApJ, 388, L13. Kirshner, R.P., Oemler, A., Schechter, P.L. and Shectman, S.A. 1981, ApJ, 248, L57. Kirshner, R.P., Oemler, A., Schechter, P.L. and Shectman, S.A. 1987, ApJ, 314, 493. Lacey, C., Guiderdoni, B., Rocca-Volmerange, B. and Silk, J. 1993, ApJ, 402, 15. Moody, J.W. 1986, Ph.D. thesis, University of Michigan, U.S.A. Moody, J.W., Kirshner, R.P., MacAlpine, G.M. and Gregory, S.A. 1987, ApJ, 314, L33. Napier, P.J., Thompson, A.R., Ekers, R.D. 1983, Proc. Inst. Electron. Electr. Eng. 71,

1295. Sanduleak, N. and Pesch, P. 1982, ApJ, 258, L11. Sanduleak, N. and Pesch, P. 1987, ApJS, 63, 809. Strauss, M.A. and Huchra, J.P. 1988, AJ, 95, 1602. Strauss, M.A., Davis, M., Yahil, A. and Huchra, J.P. 1990, ApJ, 361, 49. St.rauss, M.A., Davis, M., Yahil, A. and Huchra, J.P. 1992a, ApJ, 385, 421. Strauss, M.A., Huchra, J.P., Davis, M., Yahil, A., Fisher, K.B. and Tonry, J.P. 1992b,

ApJS, 83, 29. Szomoru, A., van Gorkom, J.H. and Gregg, M.D. 1993, AJ, 105,464. Szomoru, A., van Gorkom, J.H. and Gregg, M.D. 1996, to appear in the February 1996

issue of AJ. Szomoru, A., van Gorkom, J.H., Strauss, M.A. and Gregg, M.D. 1996, to appear in the

February 1996 issue of AJ. van de Weygaert, R. and van Kampen, E. 1993, MNRAS, 263, 481. van Gorkom, J.H. 1993, in The Evolution of Galaxies and Their Environment, proceedings

of the Third Tetons Summer Astrophysics Conference, ed. J. M. Shull and H. A. Thronson, Jr. (Dordrecht: Kluwer Academic Publishers).

Weistrop, D. 1987, BAAS, 19, 1074. Weistrop, D. and Downes, R.A. 1988, ApJ, 331, 172.

AN HI STUDY OF URSA MAJOR SPIRALS

Dark Matter in spirals and the TF-relations

M.A.W. VERHEIJEN

Kapteyn Institute Postbus 800, 9700 A V Groningen, The Netherlands

1. Research Goals

This is a progress report on our investigation of the scatter in the TFrelations and study of the Dark Matter component of spiral galaxies as a function of luminosity, morphology, scale length etc. by means of a detailed kinematic and photometric study of individual galaxies in the Ursa Major cluster. Because all the galaxies are at the same distance there is no doubt about their relative masses and luminosities. In this paper we will briefly discuss the H I properties of spiral galaxies in the cluster and the TFrelations using the kinematic inclinations and H I rotation curves.

2. The Ursa Major cluster.

A galaxy is considered to be a member of the Ursa Major cluster if its position on the sky is less than 7.5 degrees from 0: = 11 h56~9 b = 49°22' and 700 < vsys < 1210 km s-1 (Tully et al. in prep.). The distribution on the sky of all 79 cluster members identified so far is shown in Fig. 1. The somewhat elongated cluster is located in the supergalactic plane at approximately the same distance as Virgo. It lacks a central concentration, is rich in spirals and has a low velocity dispersion of ~ 150 km S-1 which results in a crossing time of the order of the Hubble time. These properties suggest that the Ursa Major cluster is currently forming. The estimated depth of the cluster would contribute ~ 0~17 to the scatter in the TFrelations.

3. Observations.

B, R, I and K' surface photometry obtained with the UH 24" and 88" telescopes on Mauna Kea is available for 78 galaxies. Analysis of the luminosity

165

M. N. Bremer et al. (eds.). Cold Gas at High Redshi/t. 165-170. © 1996 Kluwer Academic Publishers.

166 M.A.W. VERHEIJEN

I

56° 0 o E-SO - --It -t;,. Sa-Sb - ........ /""

o Sc-Sd ,/ 0g- 0 ,

itSm-lm/ " 54° 0 I

0

"-/ 0

"-/ ~ ~

52° 0 I

/ t;,. 0 \

it t;,.~ \ / 0

o~ "* \ 50° 0

I I ~ 8 ,..... "* \ 0 0

c:i "* it 10 t;,.

I m +0 ..-I o BI it

--: 48° 0 t;,. "* ~ u it 0 t;,. ff CD

0 I 0 \ 0 t;,. 0 0

46° 0 I ff /

\ 0

"* t;,.

"* "* / \

/ 44° 0

I "- t;,.

t;,. 0 / "- 0 0

" ~ /

m ,/

42° 0 I , 0

/"" ........ 0 -~- -I

30m 15m 12h Om 45m 11 h30m

R.A. (1950.0)

Figure 1. The distribution of galaxies in the Ursa Major cluster. The dashed circle indicates the 7.5 degree distance from the center (cross).

and colour profiles is currently in progress. Therefore, in this paper we will use the H-o.s magnitudes from Tormen and Burstein (1995). HI 21cm-line synthesis observations done with the WSRT provide integrated H I maps, global H I line widths and velocity fields from which kinematic inclinations and rotation curves can be derived. A total of 62 galaxies is observed of which 35 are reduced and analyzed. Figure 2 shows the integrated H I maps of those 35 galaxies.

4. A comparison with field galaxies and the Virgo cluster.

The environment of the Ursa Major cluster might influence the H I properties of spirals as is the case in the Virgo cluster (Cayatte et al. 1994). Since

An HI study of Ursa Major spirals. 167

/

/

/

/ /

/

\

\ N4389

\

- -- -- N3972

U6969 N3992 ijb.

N4102 U6940. U6923

U693 ~ N3953

.....

" U6446 "-

o \. ~: N3718 \

\ N4157 II!.. N4088

... N3917 U6667

t7 U6399 \ ,,~ OU6917

__ N4085 , N4100

o U6930

N3B96tl N3893

N3906 N3769 o N3949 N3769A I"

" N4217 ,N4010 ~N3877 G N3726

"- .....

15m

, N4183

-

N4051 $

,N4013

,0 U6962

U6973

12h Om

--45m

RA (1950.0)

/ /

./ ./

11 h30m

/ /

/ /

Figure 2. Integrated HI maps of 35 galaxies. Contour levels are at 0.5, 2 and 4 Xl021 atoms cm-2 . Individual galaxies are four times enlarged and some are displaced.

the TF-relation is mostly applied to field galaxies we should investigate whether the H I properties of Ursa Major spirals differ from those in the field. Figure 3 shows a comparison of the H I properties between Ursa Major and Virgo spirals, normalized to what is observed in field galaxies. The upper panel shows the ratio of the radius of the H I disk where NHI = 1020

atoms cm -2 to the optical diameter D~; taken from the LEDA database as a function of projected distance to the cluster center. In the case of Virgo the H I disks are relatively smaller near the cluster center. This effect is not observed in the Ursa Major cluster. The lower panel shows the H I deficiency, a measure of the global H I surface density, defined by Chamaraux et al. (1986) as < log(aHI) >T -log(aHI) in which aHI = MHr/7rR~5· <>T refers to average values for field galaxies of morphological type T. It is clear


• UMa • N3769 ~ 1.5 o Virgo U6973 .... • P-

o • Q 1 • • o •

........... • • • s:: • ••• • 00 0.. 0 • 0 • 0"- ·0· , ..

Q 0 • ........... 0.5 0 ~ ~ • 00

CD 0 0

0 0 • UMa

>. 1 0 CJ 0 0 o Virgo 0 0 ~ 0 0 • o Q) 0 0 Q) .- 0 • ~o ~. .0 CJ 0 ••• • • -.. .- • •• • -Q) •• • • • •• 0 •

'U • • ..... • ::r:: -1

0 2 4 6 Distance to cluster center (degrees)

Figure 3. Upper panel: Ratio of HI to optical diameter of galaxies as a function of projected distance from the cluster center. N3769 and U6973 are interacting and have distorted disks. Lower panel: Global HI deficiency of galaxies as a function of projected distance from the cluster center. Values in both panels are normalized to field galaxies.

that Virgo spirals near the center of the cluster not only have smaller H I disks but also show an increased H I deficiency. This effect does not occur in the Ursa Major cluster. Therefore, we conclude that the H I properties of spiral galaxies in the Ursa Major cluster are typical for field galaxies.

5. TF-relations

For 18 galaxies with inclinations larger than 45 degrees H-O.5 magnitudes and new detailed kinematic information are available and we will now investigate whether use of the H I velocity fields can reduce the scatter in the TF relation.

The left panel in Fig. 4 shows the "classic" TF-relation using global properties taken from the LEDA database like the width of the global H I profile corrected for instrumental resolution (Bottinelli et at. 1990), turbulent motion (Tully and Fouque, 1985) and inclination as derived from the optical axis ratio (Fouque et al. 1990). The middle panel shows the relation when using the width of the global profile as measured by the WSRT and the kinematic inclination derived from the H I velocity field. The panel on the right shows the relation when using the maximum rotational

8

~ 10 ",I

::r:

12

slope=-B.B 0'~=0.31

An HI study of Ursa Major spirals.

slope=-B.9 O'r.,'l" =0.33 a~!t =0.31

slope=-9.6 ar~=0.35

169

2 2.2 2.4 2.6 2.8 2 2.2 2.4 2.6 2.8 2 2.2 2.4 2.6 2.8 Log (2Vmaz)

Figure 4. TF-relations using different sources of kinematic information. Left: Single dish profile widths and optical inclinations. Middle: WSRT profile widths and kinematic inclinations. Right: Maximum amplitude of the rotation curve.

velocity from the rotation curve. An inverse least squares fit was made and the slope, rms scatter and the

more robust biweight scatter (Beers et at. 1990) are plotted for each case. It is clear that the biweight scatter does not significantly decrease although the slope steepens somewhat. If we subtract the expected scatter (01l!17) due to the depth of the cluster and the estimated scatter (01l!20) due to measurement errors in quadrature from the total observed scatter (01l!29) we find an intrinsic scatter of 01l!12 or a distance uncertainty of 6%.

6. Summary.

The cluster environment does not seem to influence the H I properties of the spirals as is the case in Virgo. This makes the Ursa Major cluster of galaxies an ideal sample to study the TF-relations.

Using H-O.5 magnitudes and the maximum rotational velocities from 18 rotation curves, the biweight scatter in the TF-relation after an inverse least squares fit is 01l!29. A quadrature subtraction ofthe estimated scatters due to the depth of the cluster and observational uncertainties results in a tentative intrinsic scatter of 01l!12 or a 6% uncertainty in distance.

References

Beers, T.C., Flynn, K. and Gebhardt, K. (1990), A.J., 100, p. 32 Bottinelli, L., Gouguenheim, 1., Fouque, , P. and Paturel, G. (1990), A.f!lA.Suppl., 82,

p.391 Cayatte, V., Kotanyi, C., Balkowski, C. and van Gorkom, J.H. (1994), A.J., 107, p. 1003 Chamaraux, P., Balkowski, C. and Fontanelli, P. (1986), A.f!lA., 165, p. 15 Fouque, , P., Bottinelli, L.,Gouguenheim, L. and Paturel, G. (1990), Ap.J., 349, p. 1


Tormen, G. and Burstein, D. (1995), Ap.J.Suppl., 96, p. 123 Tully, R.B. and Fouque, , P. (1985), Ap.J.Suppl., 58, p. 67 Tully, R.B., Pierce, M.J., Huang, J., Verheijen, M.A.W. and Wainscoat, R. (in prep.)

HI AT HIGH REDSHIFT

A.G. DE BRUYN

Netherlands Foundation for Research in Astronomy, Dwingeloo and Kapteyn Astronomical Institute, Groningen

1. Introduction

The importance of searches for high redshift H I should be obvious to anyone attending this conference. This topic appears several times on the program, and there may well be some duplication there (see contributions by Chengalur, Carilli, Swarup and Braun). Most radio searches for cosmological HI, whether in emission or absorption, have thusfar been conducted around a frequency of 327 MHz. At that frequency we are tuned to the 21cm line at a red shift of about 3.4. There is nothing magic about this frequency except that the deuterium line is at a frequency of 327.4 MHz and many observatories in the past built receivers to try and detect it. Because of this the band around this line has received protection and is relatively free of interference.

For the closed universe cosmology that I will be using in this talk the conversion between angular and linear scales is as follows: I' ~ 207/ h kpc at z = 3.4, where h is the current Hubble constant in units of 100 km s-l Mpc-1. In the following I will be mostly using h = 0.5, hence I' ~ 415 kpc. At this frequency a band of 1 MHz corresponds to 917 km s-1 velocity shift.

To radio II I observers it would be nice if the universe were closed with baryons but all the current evidence points to a nb of about 0.05 to within a factor of 2. It is therefore realistic to assume that nbh2 ~ 0.01. Even with those parameters there is plenty of gas around. A cube of 10 Mpc proper dimension, which subtends an angle of 25' on the sky, then contains about 4 X lO14 Me;) of baryons. The biggest problem for observers is that the bulk of this gas appears to be ionized so let me begin with making a few comments on this issue.

171

M. N. Bremer et al. (eds.), CoLd Gas at High Redshift, 171-181. © 1996 KLuwer Academic Publishers.

172 A.G. DE BRUYN

2. Ionization of the Intergalactic Medium

H I is most easily seen in absorption in the ultraviolet through Lya and Lyman-limit absorption. We know that the universe is highly ionized, out to red shift of about 4.5, from the fact that there is not a continuous "trough" of absorption at wavelengths below the Lya emission line in high redshift QSOs (the so-called Gunn-Peterson "test"; Gunn and and Peterson, 1965). Hence the bulk of the intergalactic medium (IGM) must have been "reionized" at some redshift which we now know must lie at z > 5. There is no consensus on what caused the reionization of the IGM. QSOs, Population 3 stars and shockwaves generated during the formation of large scale structure are plausible mechanisms.

The "Stromgren" sphere of a single luminous QSO at a red shift of z = 3.4 can grow to a radius of several tens of Mpc, if it shines for at least 108 years and emits isotropically. This corresponds to a diameter on the sky of a few degrees, depending on redshift. The density of luminous QSOs above z = 4 appears to decrease so rapidly that QSOs alone probably do not provide enough ionizing photons (Shapiro et ai., 1994). If QSOs are responsible for the re-ionization of the IGM, its ionization could thus be very nonuniform. Non-isotropic ionizing cones in QSOs will exacerbate this, assuming the cones are stable in direction over long periods of time. This conclusion however, may need to be modified if we underestimate the number of high z QSOs due to dust extinction.

The Gunn-Peterson constraint really tests for the presence of diffuse H I moving with the Hubble flow. Large concentrations of neutral gas, e.g. those hypothesized to exist in protoclusters, would reveal themselves in QSO spectra via a line we would call a damped Lya absorber. If they exist they may be deduced via the lack of metal absorption lines at the redshift corresponding to the damped absorber. We therefore suggested previously (Wieringa et ai., 1992) that there may well be proto clusters "masquerading" a.s damped Lya absorbers. I will come back to this possibility at the end of my talk when discussing search strategies for protoclusters.

A further point worth making is that the Gunn-Peterson test has been "done" for only a few dozen lines-of-sight for redshifts greater than 3.5. There could therefore be large regions of the universe that are NOT fully ionized and I believe we have no direct observational evidence to prove this is not the case. Taking this conclusion a step further we may, in fact, state that if the projected size of a neutral protocluster is about 5' (2 Mpc) the surface density of high-z protoclusters could still be about one per square degree without conflicting with the results of QSO ultraviolet spectra. That is, there may be tens of thousands of protoclusters at redshifts above 3. If such regions exist they may well be so cool that their temperature dropped

HI AT HIGH REDSHIFT 173

below the temperature of the microwave background (MWB) which at z = 3.4 is about 12 K (d. Scott and Rees, 1990).

3. HI absorption

After aU this "wishful thinking" of the previous section let us see what we can do and have done in the radio domain. Let us begin with looking in somewhat more detail at H I absorption studies. We can distinguish absorption against the MWB (d. Hogan and Rees, 1979) or against discrete high redshift radio sources.

3.1. ABSORPTION OF VERY COLD PRIMORDIAL DIFFUSE GAS

If 1 is the neutral fraction of gas with nb=0.04 it will produce a smoothedout brightness temperature n of 0.0061 K at a redshift of 3.4 (Scott and Rees, 1990). The WSRT 60" detection limit, in brightness temperature n, for a beam of about 8' and a velocity width of 1300 km s-1 in the Wieringa et al. (1992) data is about 0.30 K. Therefore, in order to detect very cold primordial gas (with 1 = 1) in absorption against the MWB, the gravitational collapse must have already caused a surface density enhancement of a factor 50 (or a factor 7 in each of the two spatial dimensions). Whether this is realistic at those redshifts remains to be seen. The bottom line, however, is that observers should always be on the lookout for absorption as well as emission signals!

~L2. ABSORPTION AGAINST DISCRETE SOURCES

Detecting II I absorption against radio continuum sources is easier. This is because the brightest high-z radio continuum sources, with flux densities around 1 Jy and angular cross-sections usually less than one arcsecond squared, have continuum brightness temperatures of about 107 K or higher. This is so much more than the spin temperature T~pin of the If I that it is many orders of magnitude easier to see gas in absorption than it is to see gas in emission. The a.mount of gas corresponding to an absorption detection of course depends on the cross-section of the radio source and is in general much too small to see in emission. Another, not independent, way oflooking at this is that in order to see gas in emission it must fill the beam which is typically 1,000-10,000 times gr.eater than the angular cross-section of high redshift radio continuum sources (QSRs or radio galaxies).

Despite the fact that it is much easier to detect 21 em line absorption than it is to see emission it has been hard to find If I absorption. In fact, there are still only two cases of 21 em absorptioll at redshifts z > 3. MallY more have been detected at lower redshifts (see Carilli, 1994, for a review).

174 A.G. DE BRUYN

0 0902+34 WSRT raw maps, nat taper 83 [FRS, 160 sublracls Freq/B 323.000/1.25 MHz 50 channels (uniform taper)

....... channel separalion 18 km/sec

:::i -2 velocity resolution 22 km/sec ;;: ~

'" -!::->< -4 ;:s

&: <il ;:s

A·· '0 'iii -6 Cl)

0::

-8

0 10 20 30 40 50 60 Channel #

Figure 1. HI absorption spectrum towards the 1.2 J y radio galaxy 80902+34. The ftux density scale is in unit.s of 5 mJy. A simple straight baseline fit is indicated. Frequency increases to the right.

In one case we are dealing with intervening absorption, the other is due to associated absorption. Both happen to lie at z = 3.~i9, due to the fact that most searches have been conducted around a frequency of 327 MHz.

The intervening case is that towards the QSR 0201+11~~, with an emission redshift z = 3.61, and was discovered because there is a damped Lya absorption line system at a redshift of 3.39 (White et ai., 1993). Despite the very large II I column density (N H ~ 3 X 1021 cm 2 ) the 21 em line is feeble (the continuum strength of the source is only 350 mJy) and very narrow, about 10 km s-1 (de Bruyn et ai., 1996). The implied T,pin = 1100 K assuming that the radio and optical lines of sight are intercepted by the same gas, a reasonable assumption in view of the very compact nature of the radio source. The high Tspin is in line with a trend towards higher spin temperatures at high redshifts (de Bruyn et ai., 1996).

The z = 3.39 radio galaxy B0902+34 (Lilly, 1988) was found to have an associated 21 cm absorption line at the redshift of that of the Lya line (Uson et al., 1991). WSRT observations of this system with higher velocity


resolution are shown in Fig. 1. The absorption spectrum shows a resolved feature with an optical depth of 0.8% and 100 km s-l halfwidth and a weak much shallower absorption wing extending several 100 km s-l to the blue (i.e. lower redshift or higher frequency). The narrow feature agrees well with the Arecibo spectrum of Briggs et al. (1993). The broad shallow wing in the WSRT spectrum would have been hard to detect in the Areeibo data in view of baseline curvature problems and needs confirmation.

The radio source has a large scale jet (Carilli et al., 1994) ending in a luminous hot-spot and the most probable location for the gas responsible for the deepest 21 cm line absorption features is the northern hot-spot (if the absorption occurred against the flat spectrum core the optical depth would have to be of order unity, which is unlikely though not impossible). The shallow blue wing on the absorption profile could well be due to gas covering other parts of the radio source, including the nucleus. The hotspot has an angular size of about 0.2" and contains about 75% of the continuum emission from the radio source (here we must make an uncertain extrapolation to the frequency of 323 MHz where we see the absorption line). If the hot-spot is responsible for the narrow part of the absorption line the optical depth is about 1.2%. If the associated H I in B0902+34 too has a Tspin = 1000 K, the average H I column density is about 2 X 1021 cm 2 .

'tVith a cross-section of 0.2"x 0.2" (corresponding to about 1.5 kpc2 ) the corresponding amount of gas is about 3 X 107 lYlra. However, the surface area covered by Lyo: emission is about 500 times larger than that covered by the hot spot (Eisenhardt and Dickinson, 1992). If the region responsible for 21 em line absorption is representative of the region producing Lyo: emission the total H I mass associated with B0902+34 could therefore well be 500 times larger or about 1010 Mra. If we would have the sensitivity and spatial resolution we could map the 21 cm line absorption across the whole radio source. This is a task well suited to the SKAI (d. Braun, these proceedings).

4. Results of previous searches for protocluster emission

The density of the smooth IGM is 10-7 (1 + z)3(nbh2 /0.(1) cm-3 or 10-5 em -3 at z = 3.4, assuming a minor fraction of baryons to be locked up in stars and galaxies (which is probably correct at high redshift). A cube of 10 Mpc dimensions then contains a total of 4 X 1014 Mra of baryons. This is also the mass known to be present in baryons in rich clusters like Perseus and Coma as deduced from their X-ray emission (Bi::ihringer (1994)). The dynamical masses of these clusters appears to be about 3-5 times larger still, which is probably contained in dark matter. Some or all of this dark matter may well be baryonic. If this gas would all be neutral for a brief phase during its formation we should be able to detect it with some effort.

176

20

-10

R.A. = 09h 03.Bm

Dec. = +33d 52'

A.G. DE BRUYN

5' beam WSRT:

Uson et aJ.o

_20~~-L~~L-~~-L~~~L-~~~-L~~~L-~~-L-L~~~

o 10 20 30 40 50 60 Channel #

Figure 2. HI emission spectrum at the location of the Uson et al. (1991) "proto cluster" , after smoothing the data to a 5' beam. The band covers 1.25 MHz in 64 channels. Frequency increases to the right.

More than 1000 hours of telescope time on the WSRT and VLA have thusfar been spent in searches for H I emission. There have not been any confirmed detections. The claimed detection by Uson et at. (1991) of a proto-cluster in the vicinity of the radio galaxy B0902+34 has not been confirmed in Arecibo (Briggs et at., 1993) and WSRT observations (de Bruyn and Katgert, unpublished). The latter spectrum is shown in Fig. 2.

We are therefore still left with upper limits which have been well summarized in Fig. 6 of Wieringa et at. (1992). Note that the total volume surveyed in that work is enough to contain the progenitors of about 8 rich clusters. Limits for 25' structures (=10 Mpc) from Wieringa et at. (1992) are about 4 X 1014 Me;) depending on the adopted velocity width of the (undetected) signals, This is just about the mass of baryons that should be present in such a volume.

It is also worth noting that the total amount of baryons (most of them probably ionized) within the pillbox-shaped volume of a synthesis map (with WSRT jVLA size 25m dishes) is about 4 X 1015 Me;) for a bandwidth


of 5 Mpc or 4500 km s-l velocity width at z = 3.4. It will be a real challenge to analyze the data and pick out the neutral component if we have no a priori information on the spatial, spectral and ionization signature.

5. A new WSRT program to search for broadband HI signals

When surveying the sky blindly for protocluster emission speed in covering large volumes of space and sensitivity go hand in hand. Not knowing what to expect (i.e. we do not know the H I mass function of protoclusters) we can not decide on the optimal observing strategy. It thus remains unclear whether it would be better to survey a large area not so deep or whether a large amount of time should be spent in observing a small area very deep. Both approaches are possible with data we have in hand right now and I will briefly describe what we may learn from them.

Furthermore, we do not really know how wide the expected signals could be. The previous searches have concentrated on deep targeted searches with a relatively high velocity resolution. In view of the fact that we need of order L014 Me of neutral gas before we can detect it it is perhaps not unrealistic to assume that the lines will be rather wide (the "virial" velocity of this much matter in a volume of a few Mpc in diameter quickly runs into several 1000 km S-l). Simulations of the expected profile from a protocluster around the moment of "turning around" have been done by Kumar et al. (1995) and indicate lines that could be as wide as a few MHz where 1 MHz ~ 1000 km s-l.

Together with Robert Braun and Jayaram Chengalur the author has begun one more WSRT survey to look for emission with very broad signals as expected in proto clusters around turnover (see e.g. Kumar et al., 1995). In 1994 a new low frequency system was made available at the WSRT which provides data for 8 frequency bands simultaneously. It uses the WSRT continuum backend with 8 X 5 MHz to be spread over a range of about 80 MHz. This will yield information about H I in 8 shells in the universe with redhsifts in the range z = 2.7 to 3.5. The survey covers 3 fields in the north galactic pole and 48 hours of telescope time was spent on each field. Each frequency band is sensitive to H I in a pillbox shaped region with a velocity depth of about 4500 km S-l corresponding to a depth of about 10 Mpc. A region of 1.6° diameter (effective) corresponds to 40 Mpc diameter. The series of pillboxes sampled in this survey therefore add up to a total of 3x8x lOx 1250 = 300,000 Mpc3 at z ~ 3.2. At z = 0 this corresponds to a comoving volume of 20 X 106 Mpc3 • Such a volume contains the "seeds" of about 100 richness >2 Abell clusters. If about 50% of the data turn out to be useful we should have sampled a representative region of the universe.

We have not yet finished the reduction. The preliminary results from

178 A.G. DE BRUYN

o i' g .- VI.:·~· • • • , 00 0 Q: 0 ;:', • o~ , 80 0

Ie " 0; . . > ' • Ii • . 0 i

~, 0 08 " b. ~ <,~ .. ' (: :,~: t 8 . ~ _ \ ~ 0 ,0 · 8 '0 "." o. 0 '. c

, " '0" , Cb ,- ',.' • ,4 :J,' : " (\ .' , · , " . tv. ,,0' 0 , "0 o. no', ''cI4 '' '. • too 000 0 0 I

<a... 00, • 0 °0 11 0 • ," ,0' . ~ 9· a 0' eo' 0,' • t .,.. • • 0 'd'.. •• .!l ,

o 0 11.'1 . ~o • ~ 'o··.AA! 0'" 0 0, ':.0 . .u .,' • r "iIf 0 •• II

·"i. 0"00.. ~ ~o ."D":,~:.1t. '.t~o .,0'.'0' '~O ·ti '.~" ..... 0" .... , 0a.Ia>' i 0,,' .. g,o-~j, .. flCl .';';

(I.., .. • ~ :. e',A.. 110, 0 V". .• .• ' •. 0 ,0 ~ , . C!I ~ , .... O.d~" • I', 0 \' ,

oQ .... • ." '0' • ,f..~ f 0" "0' ~ a .. ' J)" .- ",0" ,

~ :GIj';Oi '1'0 .~. '~Itt:~~''1fr.,oo~\' 'j', " ~" ~"'o,,~ '8-e~ o 0 11.: , • • '" • l," ',: '.0 r :.'6." 0 0 •

00 0.: :~'.'.' ,,',t O. 0 c,.t> e, .' lJ 0.. '.

"" 8' 0' 0 0 0, li-t 8 .a I' .', -' 0 0 , o. . • "()' :W' 0,°, 000,:.'0 a·,d,,· ,'0"0 .U,O 0: ,°0• 0 0 •

D. :,0 10 0 0:'0" '00 o,.J '000, ,l)Cl

fI 0 o. ~ 0 ~ ,0 • , 0,0°0' • .' . G t ,8 0 0\' O. • ..... 0.. .' eo • , o· 0 .',00 v '~9" 0 •• : •

• 0, • . AD' 1/1' J, 0 0 • 01 A O. Q

Figure 3. Continuum image of one of the three fields in one of the 8 bands ar 325 MHz. The area measures about 2° X 2°, which is only one-third of the whole synthesized field. The noise level is about 0.3 mJy and limited by source confusion. Contours are at -1.25, 1.25, 2.5, 5, 10, 20, 40, 80 and 160 mJy per beam area (54" x 113").

the first field show that we could detect signals with a peak intensity of 2 mJy (averaged over a band of 4000 kms-1 ). This corresponds to about 4 X 1014 M0 of H I which is just about the baryon mass of a rich cluster and equal to the sensitivity reached by Wieringa et at. (1992) for signals with smaller velocity spread.

The wide fields of view means that the images are littered with thousands of continuum sources. As an example of what the sky looks like in continuum, down to 1 mJy flux levels, Fig. 3 shows the continuum image in one of the eight 5 MHz bands observed at l' resolution. The wide fields and the wide bands complicate the data reduction. The chromatic effects and error side lobe patterns due to strong off-axis sources seriously complicate the data reduction.


A second WSRT survey that could be used for searches for very large mass concentrations is the WENSS survey. This survey will cover 1/4 of the sky, i.e. 10,000 square degrees, at l' resolution. The survey, which is now nearing completion, observed the sky in a band of 5 MHz with about 0.75 MHz spectral resolution. The survey was done mainly for continuum, variability and polarimetric studies of galactic and extragalactic radio sources. At each location on the sky a net integration of about 1 hour was achieved. The total survey has taken up about 8 months of telescope time. The survey frequency resolution of 0.75 MHz corresponding to an H I Doppler spread of 700 km s-1. If instead of 1 image, 7 images of the sky would be made we could reach a 5a sensitivity per channel of about 25 mJy. If we smooth this to 4' (corresponding to 1.6 Mpc) this will go up to about 50 mJy. The corresponding H I mass detection limit should be about 2 X 1015 Mev. Now the question I am faced with is: "Should I invest about one year of my life to search for such massive structures or are there more promising things to do?" For the moment I have no plans to embark on such an endeavour.

6. New WSRT tunable receivers and search strategies

At the end of 1996 the WSRT will be equipped with a new series of frontends as part of the upgrade of its receivers and backend. These receivers, working in the UHF-band, will cover the frequency ranges from 260-450 MHz (UHF-low) and 700-1200 MHz (UHF-high) and were designed specifically for red shifted H I searches. The corresponding range in red shift space runs from z = 4.5 to 2.2 and from 1 to 0.2. A powerful new backend, with half a million spectral channels, should become available in 1997/8. The large RF bandwidth and high spectral resolution provided by this backend, are essential to improve the limits on proto cluster emission. Primary beam modelling out to many degrees, and possibly wide field mosaiced selfcalibration, will be needed to properly remove the "chromatic" and "error" sidelobe responses of the background radio sources.

What sort of programs will we embark on with these new systems? Sensitive searches for emission require a large investment of telescope time (many 100 hours of integration). With the UHF-low system we will undoubtedly conduct a deep survey at redshifts from 3.5 to 4.5 but the real question is, in which directions should we search? It is therefore important to select the target directions in a careful manner. We could search near known concentrations of high-z galaxies or quasars, but then perhaps we are to "late" in the sense that most of the gas has already collapsed. It may therefore be better to search "blindly" in those directions where no galaxies and quasars are found. At least we would then be safe against large scale inonization of the IGM. A more profitable approach may be to look

180 A.G. DE BRUYN

in the directions where (extended) damped Lya absorbers are found. In the contribution by Chengalur a blind search for damped Lya systems is described. It is done towards extended high redshift radio galaxies at redshift going up to 4.3. When we find H I absorption in such sources, which hopefully will be just a matter of time, we could spend a large amount of time in the direction of those objects where we detect a broad damped Lya line and for which the radio data indicate that the absorber is at least 50-100 kpc in diameter. Note that, if we would find such an absorber, there would be no possibility to check the metal content of the absorber because the ultraviolet continuum of high-z radio galaxies is extremely faint.

With the UHF-high system, on the other hand, we can investigate in detail the redshift evolution of damped systems at lower redshifts. Such studies may be useful to investigate the issue of 'selection effects through dust in damped absorbers. Radio interference in both UHF-bands will pose a real challenge.

7. Conclusions

There is no observational evidence that there could not be large numbers of "proto clusters" with detectable masses of H I, although none have been found. Current limits are about 1014 M 0 .

Absorption searches indicate that damped Lya absorbers at high redshift have a rather high spin temperature (r~pin = 1000 K) "Unbiased by dust" searches for absorbers in z > 4 radio sources may prove to be the best way of selecting the directions in which to spend very large amounts of observing time with the new tunable WSRT UHF-low system and the GMRT. Limits could be pushed down to about 1013 M0 for a I' beam if large amounts of telescope time are invested.

Acknowledgements. The WSRT is operated by the Netherlands Foundation for Research in Astronomy (NFRA) with financial support by the Netherlands Organization for Scientific Research (NWO). I am grateful to Peter Katgert, Robert Braun en Jayaram Chengalur for discussions and permission to present some of the results of our joint collaborations.

References

de Bruyn, A.G., O'Dea, C.P. & Baum, S.A. 1996, Astron.e9 Astrophys. 305,450. Briggs, F.H., Sorar, E. & Taramopolous, A., 1993, Ap.J. , 415, L99. Bohringer, H. 1994, in Cosmological aspects of X-ray Clusters of Galaxies, Kluwer Aca

demic Publishers, 123 Carilli, C.L. 1995, J.Astrophys.Astr. 16, 163. Carilli, C.L., Owen, F.N. & Harris, D.E., 1994, Astron.J. 107, 480.

HI AT HIGH REDSHIFT

Eisenhardt, P. & Dickinson, M. 1992, Ap.J. 399, L47. Gunn, J.E. & Peterson, B.A., 1965, Ap.J. 142,1633. Hogan, C.J. & Rees, M.J. 1979, Mon.Not.Royal.astr.Soc. 188,791. Kumar, A., Padmanabhan, T. & Subramanian, K., Mon.Not.R.astr.Soc 272,544. Lilly, S.J. 1988, Ap.J. 333,161. Scott, D. & Rees, M.J. 1990, Mon.Not.R.astr.Soc. 247,510. Shapiro, P.R., Giroux, M.L. & Babul, A., 1994, Ap.J. 427, 25. Uson, J.M., B agri, D.S. & Cornwell, T.J. 1991, Phys.Rev.Letters 67,3328. White, R.L., Kinney, A.L. & Becker, R.H. 1993, Ap.J. 407,456. Wieringa, M.H., de Bruyn, A.G. & Katgert, P. 1992, AstI'OTI.€3Astrophys. 256,331.

181

BUTCHER-OEMLER EFFECT AND RADIO CONTINUUM

K. S. DWARAKANATH Raman Research Institute, Bangalore 560080, India

AND

F. N. OWEN National Radio Astronomy Observatory, Socorro, NM 87801, USA

Abstract. The Abell clusters 2125 and 2645 have been imaged in the radio at 20 cm to the same luminosity limit using the Very Large Array. Both the clusters are very similar in richness and redshift, but Abell 2125 has a higher blue galaxy fraction (0.19) compared to Abell 2645 (0.03). The cluster Abell 2125 appears to be qualitatively different from Abell 2645 in having an order of magnitude more radio sources associated with the cluster. The radio sources belonging to Abell 2125 are distributed over a region (about 15 Mpc) much larger than the central regions of the cluster. The origin of radio emission could be due both to AGN-type of activity and/or due to massive star formation in these galaxies. On the average, the optical counterparts of the radio sources in Abell 2125 farther from the cluster center show signs of star formation and blue color. Most of the optical counterparts of the radio sources within the central regions ("-' 3Mpc) of Abell 2125 are red in color. No radio emission was detected to the 50" limit of 3.2 X 1022 W Hz-l from most of the blue galaxies in the central regions of the cluster which are responsible for the Butcher-Oemler effect in Abell 2125.

1. Galaxies in distant Clusters

The population difference between field and cluster galaxies have been recognized for a long time. The nearby Coma cluster of galaxies contains largely ("-'95%) red E and SO galaxies while most galaxies (......,80%) in the field are spirals. This is believed to be due to loss of gas in cluster galaxies resulting in very little recent star formation. This has left an old population of stars in cluster galaxies resulting in their red color. However, photome-

183

M. N. Bremer etal. (eds.), Cold Gas at High Redshift, 183-194. © 1996 Kluwer Academic Publishers.

184 K. S. DWARAKANATH AND F. N. OWEN

try of distant clusters revealed a different situation as to the color of their constituent galaxies (see Dressler 1984 for a review).

A broad-band photometric study of 33 nearby and distant (z = 0.5) clusters of galaxies which are rich, compact, and regular revealed an increasing trend of the fraction of blue galaxies with redshift. This is known as the Butcher-Oemler effect (BO effect). After taking into account various selection effects such as K-correction, the color-magnitude relation for E and SO galaxies, and field contamination Butcher and Oemler found an increasing trend in the fraction of blue galaxies in the cluster as the redshift increased to 0.5 (Butcher and Oemler 1978, Butcher and Oemler 1984a). This would imply that many of the early-type galaxies seen in the present day clusters were actively forming stars until quite recently.

This was a surprising result indicating a dramatic evolution of galaxies in clusters over a small fraction of the age of the Universe. Naturally, this result invoked skepticism and caution. Several causes like the range of U-V colors of early-type galaxies, differences in K-correction for red and blue galaxies, photometric errors associated with the faint galaxies were suggested as contributing to this effect (Kron 1982, Ellis 1983, DeGioiaEastwood and Grasdalen, 1980). Notwithstanding these complications further photometry and spectroscopy have confirmed the basic effect (Couch and Newell 1984, Butcher and Oemler 1984b, Dressler and Gunn 1982, 1983). However, no consistent picture has yet emerged as to the nature of the blue galaxies in high redshift clusters. In the cluster 3C295 the blue population appears to be made up of Seyferts and starbursts while in CI0024+1654 and C11447+2619 they appear to be spirals. In AC 103, AC 114, and AC 118 they appear to be more active Es and SOs. More recently Hubble Space Telescope results on AC 114, A370, CI0016+16, and CI 0939+4773 appear to indicate that most of the blue population in these clusters are disk-dominated systems involved in mergers and interactions (Couch and Sharples 1987, Couch et ai. 1994, Dressler et ai. 1994, Wirth et ai. 1994).

2. Motivation for the present Radio observations

While the nature of the blue galaxies in distant clusters is varied it appears that they have large amount of gas and are undergoing enhanced star formation or are in the post star burst phase. If this is indeed the case then it is indicative of significant evolution in cluster galaxies since one does not see such activity at the present epoch. However, there are some difficulties with this picture - the spectra and surface brightness of some of these blue galaxies are quite different from the present day normal galaxies and some high red shift clusters (for e.g., CI0016+16) do not contain blue

BUTCHER-OEMLER EFFECT AND RADIO CONTINUUM 185

galaxies (Koo 1981, Mathieu and Spinrad 1981, Dressler and Gunn 1982). In addition, it has been argued that the blue luminosity is not an accurate indicator of current star formation rate due to extinction of massive stars in giant molecular clouds and a wide range of spectral types that contribute to the blue luminosity of a galaxy (Sage and Solomon 1989). Hence it will be interesting to test the star formation picture in blue galaxies from an alternative point of view.

It is known that many near-by spiral galaxies are radio sources. They emit significant amounts of non-thermal radio emission. There is also a well-known correlation between this and their infrared luminosity (Condon 1992). One possible explanation of this correlation is as follows: massive stars which are born in giant molecular clouds are surrounded by gas and dust. These stars ionize and heat their surroundings and most of their energy comes out in the infrared. When these stars become supernovae they accelerate the cosmic rays to produce non-thermal radio emission. A study of local spiral galaxies has shown that the 1.4 GHz spectral radio luminosity varies from 1020 to 1023 W Hz-1 while the far infrared luminosity varies from 108 to 1011 L0 . We can expect similar radio emission from the blue galaxies; it might even be more due to the enhanced star formation that might be going on in them. It will be instructive to compare them with the local sample. The poorer resolution (0.5' at 12/Lm) and sensitivity of the IRAS survey makes it unsuitable to look for the expected infrared emission from the blue galaxies. On the other hand, the sensitivity, the resolution, and the field of view achievable with radio synthesis telescopes seem adequate to detect the expected radio emission from the blue galaxies in high redshift clusters. A detection limit of 100/LJy beam -1 corresponds to 3.2 X 1022 W Hz-1 at a red shift of 0.25 (Ho = 50 km s-1 Mpc-1 ; n = 0).

3. Source selection and Radio observations

The main criteria used to select the appropriate clusters are as follows -redshift range, richness, number of constituent galaxies, the blue fraction, the solid angle over which the blue galaxies are distributed, the position of the cluster in the sky, and availability of photometric and/or spectroscopic data. On the basis of these the following 4 clusters were chosen: A2125, A2645, A370, and Cl 0024+1654. Both A2125 and A2645 are at a redshift of about 0.25 and are similar in richness (class 4). Their blue galaxy fractions are 0.19 and 0.03 and should make a very useful comparative study. The clusters A370 and CI0024+1654 are at a redshift rv 0.38 a.nd ha.ve richness class between 2 and 3. Their blue galaxy fractions are 0.21 and 0.16 respectively. More details of these clusters can be found elsewhere


(Butcher, Oemler, and Wells 1983, Butcher and Oemler 1984a, Struble and Rood 1991).

The radio observations were carried out using the Very Large Array in the 20 cm band. The 4 clusters were observed for a total ",70 hr divided in such a way as to get similar detection limits in luminosity. All the observations were made in the multichannel continuum mode (50 MHz bandwidth, 8 channels). The cluster A2125 was observed in the 3 km configuration of the VLA (the C array) during July 1993 while the other 3 clusters were observed in the 11 km configuration of the VLA (the B array) during August - September 1994. The full widths at half maxima of the synthesised beams in the C and B configurations were 13" and 6" respectively. These correspond to 80 kpc and 40 kpc at z '" 0.25.

The observations of A2125 and A2645 were analysed using the Astronomical Image Processing System developed by the National Radio Astronomy Observatory. In both the cases the entire primary beam of the VLA dishes (",60' at 1.4 GHz) was imaged. The r.m.s. value of the noise in the images of A2125 and A2645 was 20 - 25 j1Jy beam-I, close to the expected value. To a detection limit of 100 j1Jy beam-1 :n8 and 174 sources were detected in the fields of A2125 and A2645.

4. Optical Identifications

4.1. CENTRAL REGIONS OF THE A2125 CLUSTER

Photometry ofthe two clusters A2125 and A2645 are available from Butcher, Oemler, and Wells (1983, hereafter BOW). The central 55 square arcmin of the clusters have been studied on J and F plates taken at the prime focus of the KPNO and CTIO 4 m telescopes. The details of these observations are found in BOW. Photometry is expected to extend to a red magnitude of 22 mag. A red image of the A2125 cluster center can be seen in Fig. 22 in BOW. A radio image of A2125 covering this region is shown in Fig. 1. There are 19 radio sources within the area covered by the red plate of BOW (see Table 1). There are 17 blue galaxies according to the definition (i.e., (J - F) < 1.6) of Butcher and Oemler (see Fig. 2d in Butcher and Oemler 1984a). Of the 19 radio sources within the BOW frame 13 have optical counterparts. Of these, 3 have (J - F) values less than 1.6 while the other 9 have (J - F) values more than 1.6. One source has a. (J - F) value close to 1.6 (no. 149 in Table 1). According to the definition of BOW this is the dividing line for defining blue and red galaxies. So, although there are 17 optical sources which are blue we detect radio emission from only 3 of these. The following table (Table 1) summarises these findings.


s l£ e. z o i= <C Z :::; (,) w o

6630

26

c& 24 •

22

20

ll.

'{J

00 4030 00 RIGHT ASCENSION (B1950)

Figure 1. A radio image at 1.4 GHz of the central region in A2125. The contours are at -100,-60,60,100,140,180 pJy/beam. An arc min at the distance to A2125 is about 0.4 Mpc. The' +'s indicate the boundaries of the region observed by BOW. Although there are 17 radio sources within the boundaries defined by the' +' s the central pear-shaped source is resolved into 3 sources in higher resolution images. Thus the total number of sources has been counted as 19.

4.2. LARGER REGIONS OF THE A2125 CLUSTER

The full primary beam of the VLA dishes cover a region which is "'60' in diameter. A radio image of the field containing A2125 in grey scale representation is shown in Fig. 2.

The photometry of BOW covers only the central 7.4'x 7.4'. Optical identifications outside this region require large-field photometry and preferably spectroscopy to identify cluster members. This has been possible recently. A detailed discussion of these observations is beyond the scope of this article. Only brief information will be given here. Owen, Keel & Morrison (1995, OKM hereafter) had an observing session this August on the KPNO 0.9 m with a 2K X 2K CCD. This CCD produces an image which is 23'x 23' with 0.68" pixels. The R images go to a limiting magnitude of 24.5 while the B images go down to 22.5. This has enabled us to obtain good colors


TABLE 1. Optical identifications in the central region of A2125

Radio no. Radio Lum. BOW no. F (J - F) sepn. (1022 W HZ-I) (mag) (mag) (")

8 723.9 33 18.63 1.29 3.4,0.2 18 520.5 20 18.28 1.85 4.7,4.5 38 33.9

71 13.3

116 7.7 170 20.43 1.96 4.7,0.1 121 8.5 22 18.31 1.77 3.1,0.1 134 6.0 137 4.9 44 18.81 1.36 1. 7,-5 149 10.4 330 21.63 1.64 -5.5,1.9 168 5.3 132 20.06 2.22 6.4,-2.5 172 4.4 190 5.9 17 18.18 1.51 0.5,2.3 240 9.8 248 3.2 101 19.68 1.83 0.8,-1.4 292 2.8 6 17.41 2.11 2.6,2.0 300 4.2 368a 7.4 10 17.79 2.07 0.2,-0.5 368b 39.8 5 17.37 2.13 0.2,-0.6 368c 194.9 3 17.16 2.14 0.6,-0.6

and magnitudes. The spectroscopic observations covered the region 3600 -7200 A. For a redshift of 0.25 this covers the wavelength region 2900 - 5800 A. This covers the region of Ca H, K, the D(4000) break, the G-band, H;3, Mg b, [0 II] 3727 & [0 III] (4959, 5007). There are 38 optical counterparts in the region covered by the CCD of which 20 have spectroscopic data. The positions of the 38 optical counterparts are plotted in Fig. 3.

Following the definition of BOW we find that J = B - 0.23(B - V) and F = R + 0.28(V - R). Hence, (B - R) = (J - F) + 0.23(B - V) + 0.28(V -R). From Coleman et al. (1980) we find that (B - V) = 1.0 and 1.6 for blue/red spirals and blue/red ellipticals respectively at a redshift of 0.25. Similarly (V - R) = 0.8 and 1.2 for blue/red spirals and blue/red ellipticals respectively at a redshift of 0.25. Assuming that the optical identifications we have in A2125 are spirals we find that (B - R) = (J - F) + 0.45. Since BOW consider (J - F) = 1.6 as the dividing line between blue and red galaxies, it translates to a (B - R) value of 2.05. Based on the optical observations of OKM we find that there are 31 identifications with color information. Of these, 4 have colors very close to 2.05 (within 0.05 mag).


6650

45

40

0' It) 35 (I)

~ Z

30 a i= c( z ::J 25 0 w c

20

15

10

05

.'·)t::l

• .

' ..

"

.,

'.

.' . }

150

-

-. '.

: ..

•

-. , .o

..

' ..

• •

•

.

"

•

. .

•

'.

1544 43 42 41 40 39 38 37 RIGHT ASCENSION (81950)

Grey scale flux range= 80.0 200.0 MicroJYIBEAM

200

Figure 2. A grey scale representation of the radio image at 1.4 GHz of the field containing A2125. This field covers the full primary beam of the VLA dishes. At the distance to A2125 an angular size of 5' corresponds to ..... 2 Mpc.

They are border-line cases. Of the remaining 27, 8 are red (B - R > 2.02) while 19 are blue (B - R < 2.05). However, if we restrict ourselves to those optical identifications for which we have both color and red shift information there are 5 red galaxies and 10 blue galaxies. Thus, the majority of radio sources identified with the optical candidates in A2125 appear to be blue in color. This is in contrast with the conclusion one arrives at by restricting to the BOW field where majority of the optical counterparts are red (see Table 1).


6650

45

40 + +

S + + lI) 35 0)

§. +

z 30 0 +

~ + Z * +~ ::::i 25 + + -tt + 0 w

+ +: +. * c + + + ++:t 20

+

15 +

10

05

43 42 41 40 39 38 37 RIGHT ASCENSION (61950)

Figure 3. Positions of the optical identifications. The CCD covers an area 23' x 23' which is smaller than the full size of the primary beam of the VLA dishes.

4.3. THE A2645 CLUSTER

We used the COSMOSjUKST southern sky object catalogue to obtain optical identifications of the radio sources detected in this cluster. A total of 23 optical counterparts were found. Purely by chance we expect only 2 optical candidates to be associated with the radio sources. These 23 candidates are classified as galaxies in this catalogue. All these optical counterparts are within 5" of the radio sources and more than half of them are within 2". Although the positional accuracy of the catalogue is believed to be 1 - 2" searching for optical counterparts up to 5" has been found to be more useful to allow for extended sources, real positional offsets between radio and optical positions, etc. Of the 23 identified candidates spectra were obtained of 19 by OKM. Only 4 of these belong to this cluster. Of these 4 sources


that belong to this cluster 2 are blue and 2 are red.

5. Discussion

The primary motivation for the present exercise was to look for enhanced radio emission from the blue galaxies in the central region of A2125 that are responsible for the BO effect. To the detection limit (5a) of 3.2 X 1022 W Hz-l most of these blue galaxies were not detected. Based on their average magnitude and color (see BOW) their blue magnitude is expected to be about -22. From Fig. 4 one can see that there is considerable spread in the expected radio luminosity for a given blue luminosity. It is clear from this figure that the radio luminosities of the blue galaxies must be very similar to the local spirals and no enhanced radio emission is associated with the blue galaxies. If enhanced star formation is responsible for their blue colors it remains a puzzle as to why no enhanced radio emission is detected from these blue galaxies.

The cluster A2125 appears to be qualitatively different form A2645. The number density of radio sources belonging to A2125 is an order of magnitude more than that in A2645. This is quite surprising since the number density of galaxies as seen in optical is very similar in both the clusters. The radio sources that belong to A2125 are distributed over a region much larger (",15 Mpc) than just the central parts of the cluster observed by BOW (see Fig. 3 and Fig. 1). About 2/3rd of these radio sources are blue. Thus, although we do not detect radio emission from the blue galaxies in the center of the cluster that are responsible for the BO effect majority of the radio sources detected in A2125 are blue. One might be witnessing merger induced star formation and radio emission from galaxies farther away from the cluster center. It seems possible that the distribution of radio sources belonging to A2125 (Fig. 3) is indicative of enhanced gas density distribution in the cluster due to the merger. In such a case one can expect the X-ray surface brightness distribution to be similar to this. The BO effect observed in A2125 could well be a result of such a merger activity.

The present division of the optical identifications into blue/red galaxies does not necessarily classify them into star-forming and non star-forming galaxies. There are several reasons for this. At the outset the definition of blue/red used in this paper comes from the BOW definition who define in terms of the (J - F) color. This translates into a (B - R) value depending on whether the galaxies are spirals or ellipticals. This requires a knowledge of the detailed morphology of the cluster galaxies which is hard to discern. Merging of galaxy images causes further complication in estimating magnitudes and colors. Another unknown factor in estimating colors is extinction

192

..

V N

o .0 E N ~

.;:, 0\ o

00

K. S. DWARAKANATH AND F. N. OWEN

D

D. D D D.

: '. . ~ . ",.'

blue luminosity in mag (loA)

Figure 4. The absolut.e blue magnit.ude (M) is obtained from the apparent blue magnit.ude (B) using the following relation: M = B - 0.25 - 1.1 - 41.1 where the three t.erms are corrections due to interstellar absorption, K-correction, and dist.ance modulus respectively. It is assumed that Ho=50 km s-1 Mpc-1 and n = O. The squares are for A2125 while the points are for the local sample of spirals and irregulars. The data for the local sample are from Condon (1989).

due to dust. A combination of photometry and spectroscopy is more helpful in identifying star forming galaxies rather than color alone.

Spectroscopy of the optical identifications in A2125 has revealed that about 1/2 of the spectra resemble those of the ellipticals while the other 1/2 show some signs of star formation in the form of, e.g., suppressed D( 4000) break. In some of the very blue galaxies there appear to be signs of enhanced star formation. Thus the origin of radio emission from the sources in A2125 could be due to AGN-type of activity and/or due to massive star formation. See Fig. 5 for a comparison of different types of radio sources in terms of their number distribution as a function of the radio luminosity. Those that show signs of star formation are generally found to be blue as opposed to


0 0 ~

~p.andirr. Ul

" local starbursts e " 0 Ul

.... 0

ci Zo local monsters

If)

local E/SO

A2125(opl id) ~ 0

A2125(spectra)

18 20 22 24 26

log(lum at 1.4 GHz)(W/Hz)

Figure 5. The data for the local sample is from Condon (1989) and Sadler et al. (1989). The zeros for different plots have been shifted for the sake of clarity.

the the AGN-types which are found to be red in the optical observations of OKM. In addition, the galaxies which show signs of star formation are on the average farther away from the cluster center than the AGN-type galaxies. This might explain the interesting fact that most radio sources within the BOW frame (which covers only the central regions ofthe cluster) are identified with red galaxies while if we consider the radio sources all over the cluster about 2/3rd of them are blue.

Radio observa.tions like the present one ca.n be effectively used in identifying star forming galaxies in high red shift clusters in particular. The power of radio observations comes from their ability to detect such a.ctivity in clusters over linear scales as large as 10 - 20 Mpc. It is also quite evident that the existing optica.l observa.tions (like in BOW) of most of these clusters a.re inadequate to trace large scale activity such as the one witnessed in A2125.


Acknowledgements. One of us (KSD) wishes to thank the Scientific Organisers of this workshop for inviting to give this review and for the financial support of travel and local stay. KSD also wishes to acknowledge a Jansky post-doctoral fellowship at the NRAO during which time the radio observations described in this article were made. The Very Large Array is a part of the National Radio Astronomy Observatory which is operated by the Associated Universities, Inc., under a cooperative agreement with the NSF. This work has used data from the COSMOS/UKST Southern Sky Catalogue provided by the Anglo Australian Observatory.

References

Butcher, H. R. & Oemler, A. Jr 1978, ApJ, 219, 18. Butcher, H. R., Oemler, A. Jr & Wells, D.C. 1983, ApJSS, 52, 183. Butcher, H. R. & Oemler, A. Jr 1984a, ApJ, 285,426. Butcher, H. R. & Oemler, A. Jr 1984b, Nature, 310, 31. Coleman, G. D., Wu C.-C., Weedman, D. W. 1980, ApJSS, 43, 393. Condon, J. J. 1989, ApJ, 338, 13. Condon, J. J. 1992, Ann Rev Astron & Astrophys, 30, 575. Couch, W. J., Ellis, R. S., Sharples, R. M. & Smail, I. 1994, ApJ, 430,121. Couch, W. J. & Newell, E. B. 1984, ApJSS, 56, 143. Couch, W. J. & Sharples R. M. 1987, MNRAS, 229, 423. DeGioia-Eastwood, K. & Grasdalen, G. L. 1980, ApJ, 239, L1. Dressler, A. 1984, Ann Rev Astron & Astrophys, 22, 185. Dressler, A. & Gunn, J. E. 1982, ApJ, 263, 533. Dressler, A. & Gunn, J. E. 1983, ApJ, 270, 7. Dressler, A., Oemler A. Jr, Sparks, W. B. & Lucas, R. A. 1994, ApJ, 435, L23. Ellis, R. S. 1983, In The Origin and Evolution of Galaxies, ed. B. J. T. Jones, J. E. Jones,

p. 255, Dordrecht: Reidel. Koo, D. C. 1981, ApJ, 251, L75. Kron, R. G. 1982, Vistas Astron., 26, 37. Mathieu, R. D. & Spiurad, H. 1981, ApJ, 251, 485. Owen, F. N., Keel, W. & Morrison, G. 1995, private communication Sadler, E. M., Jenkins, C. R. & Kotanyi, C. G. 1989, MNRAS, 240, 591. Sage, 1. J. & Solomon P. M. 1989, ApJ, 342, L15. Struble, M. F. & Rood H. J. 1991, ApJSS, 77, 363. Wirth, G. D., Koo, D. C. & Kron, R. G. 1994, ApJ, 435, LI05.

WARM MOLECULAR GAS IN AGNS AND COOLING FLOWS

WALTER JAFFE AND MALCOLM BREMER

Leiden Observatory P.B. 951,,), 2,')00 RA Leiden, Netherlands

AND

RODERICK JOHNSTONE

Institute of Astronomy Madingly Road, Cambridge, England

Abstract. We have looked for infrared line emission from warm (1000 K) molecular hydrogen in cooling flows. We have two detections, with luminosities of about 1041 ergs s-1, but these are small sources which are associated with the active nuclei of the central cluster galaxies, rather than with the general cooling flow.

There are at least two reasons to expect lots of neutral gas in cooling flow clusters. First, the gas that has cooled from very high temperatures and recombined should be somewhere, at least until it is incorporated into stars, and second, soft X-ray spectra of cooling flow clusters show strong absorption in excess of that expected from our Galaxy. The latter observation seems to require that a substantial area, say 104 square kiloparsecs, about the central cluster galaxy be covered with a neutral gas with a column density of about 1021 atoms em -2. The total mass of such gas would be about 1012 M0 (Daines et al., 1993). Despite extensive searches, nobody has found any evidence for this gas outside the X-ray window. Searches for absorption or emission in the 21 em HI line or the CO rotation lines have in general found nothing (e.g. Jaffe and McNamara, 1994, O'Dea et al., 1994). The upper limits of these surveys exclude the possibility of gas as required by the X-ray measurements, if its physical regime of temperature and pressure resemble the local interstellar medium. This has led Daines et ai. (1993) to suggest that the gas is in the form of extremely small « 1 pc) extremely cold « 10 K) clouds.

195

M.N. Bremer etal. (eds.), Cold Gas at High Redshift, 195-197. © 1996 Kluwer Academic Publishers.

196 WALTER JAFFE ET AL.

Since such clouds, if they exist, would be surrounded by the ultrahot cluster gas they should have outer shells that are respectively ionized, atomic, warm molecular, and cold molecular as we go from the outside in. Having failed to find the atomic or cold molecular fractions we decided to look for the warm molecular fraction in the light of the 2.1 J.L vibrationrotation line of molecular Hydrogen. This emission arises only in dense gas where the temperature is '" 300 - 1000 K. The only regions in normal galaxies that meet these conditions are in strong shocks.

We have run two experiments, one on the United Kingdom Infrared Telescope (UKIRT) using narrow band (1:1>../>.. '" 1%) filters, and one on the European Southern Observatory's (ESO) 2.2m telescope using a FabryPerot filter ('" 0.1 %). The results ofthe two experiments seem inconsistent.

The UKIRT experiment included observations of Abell 426 (Perseus A, NGC 1275), Abell 478, and PKS 0745-19, all famous cooling flow clusters. The first two showed detectable H2 emission, but only in the immediate vicinity of the nucleus, within about 1 kpc. In both cases the luminosity in the line was a few times 1041 ergs s-l. No emission was detected from the last source, with a 3a upper limit also of order a few times 1041 ergs s-1.

At ESO we had no positive detections for the clusters PKS 0745, Hydra A, Abell 2029, M 87, and Abell 3115. The typical line detection limit was'" 5 X 10-16 ergs s-1 cm-2 which for the distant cooling flows corresponds to 1040 ergs s-l. With the exception of M 87, these clusters are not drastically different in mass cooling rate, central radio power, or distance than the two detected clusters. We are investigating the possibility that the wavelength calibration of the Fabry-Perot instrument was in error.

The two detections come from areas immediately surrounding the radi%ptical nuclei of the central cluster galaxies and almost certainly are related to the active nuclei, not to the cooling flow in general. NGC 1275 is a famous, powerful, flat-spectrum radio galaxy, while A 478 harbors a rather modest radio source. The warm molecular gas could arise in the "obscuring torus" purported to surround the nuclei of all radio galaxies, or could be connected with other dense clouds in the inner kiloparsec, generally the extent of the narrow line region, of these galaxies. HST pictures of NGC 1275, for example, show that the inner kiloparsec is full of dusty clouds.

A real interpretation of this would require some real physics of the cloud surfaces. In place of that we have estimated that the surface brightness of a layer of warm H2 is '" 1O-2o NH ergs s-l cm-2 sr-1 where NH is the warm molecular surface density. If N H > 1024 the gas will be optically thick and the surface brightness will be '" 104 ergs s-1 cm -2 sr-1. Thus a luminosity of 1041 ergs s-1 requires an emitting area of at least 1037 cm2, or 1 square parsec and probably rather more if the warm molecular phase

WARM MOLECULAR GAS IN AGNS AND COOLING FLOWS 197

is not optically thick. This places this phase outside the broad line region, possibly in the obscuring torus region (1- 10 pc) or possibly in the narrow line region (100 -+ 1000 pc), depending on how thick the warm molecular gas is.

The detection of very extended sources, such as the general cooling flow, is more difficult than the detection of point sources, because various effects limit the sensitivity to low surface brightness radiation. The UKIRT observations could have detected extended emission with a surface brightness above 3 X 10-15 ergs s-I cm-2 arcsec-2 , or "-J 10-4 ergs s-I cm-2 sr-I . Ferland et al. (1994) present a model of ultracold gas in cooling flows and predict that the surface brightness in the H2 IR lines would be "-J 10-7 ergs s-1 cm-2 • This would be undetectable, but other authors predict warmer clouds and probably higher line emission (O'dea et al., 1994) so the issue of whether such dense clouds exist is still open.

References

Daines, S. J., Edge, A. C., Steward, G. C. 1993, MNRAS 262,901 Ferland, G., Fabian, A. C., Johnstone, R. M. 1994 MNRAS 266, 399 Jaffe, W., McNamara B. R. 1994 AA 281,673 O'Dea, C. P., Baum, S. A., Maloney, P. R., Tacconi, L. J., Sparks, W. B. 1994, ApJ 422,

467

THE SEARCH FOR COLD GAS IN THE INTRACLUSTER MEDIUM

CHRISTOPHER P. O'DEA AND STEFl A. BAUM Space Telescope Science Institute :'1700 San Martin Dr., Baltimore, MD 21218 USA

Abstract. We present the results of searches for cold gas in the lCM of nearby (z < 0.1) clusters of galaxies. These searches include (1) Arecibo observations sensitive to atomic hydrogen with broad line widths, (2) VLA observations with high velocity resolution sensitive to cold optically thick atomic hydrogen, and (3) SEST observations to search for warm CO. We do not detect significant amounts of cold gas in the lCM of these clusters. Limits on column densities are in the range N(H2) < few X 1020 cm-2, N(H) < 1 X 1018 cm-2 and limits on the mass of gas in the central regions of the cluster are M(H2 + H) < 109 Me!). We also argue that heating ofthe gas by the X-rays is sufficient to keep the gas at temperatures above 10 K, thus ruling out the possibility that the gas would be too cold to emit significantly. Thus either cloud destruction is efficient (e.g., star formation, shredding, evaporation) or sources of cold gas (ram pressure stripping of galaxies, infall into the cluster, cooling flows) are not currently very important.

1. Why should there be cold gas in the ICM ?

There several potential candidates for a population of cold clouds which might exist in the intracluster medium (ICM) of clusters of galaxies (e.g., Sarazin 1988): (1) Cold gas which has been removed from the individual galaxies, possibly by ram pressure or by galaxy collisions; (2) Primordia.l clouds or protogalaxies which are currently falling into the cluster; (3) clouds which condense from thermal instabilities in a cooling flow in the inner 100 - 200 kpc of the cluster center (e.g., Cowie and Binney 1977; Fabian and Nulsen 1977; Mathews and Bregman 1978; Fabian 1994). Mass accretion rates in cooling flows are estimated to be in the range m f'V 10 -100 Me!) per year. If these accretion rates last for the lifetime of the cluster (f'V 1010

199


200 CHRISTOPHER P. O'DEA AND STEFl A. BAUM

yr) the accumulated mass in gas would be 1011 _1012 MG' In some cooling flow clusters there is evidence for low energy X-ray absorption in excess of that which would be produced by the known Galactic column density of HI (White et ai. 1991; Mushotzky 1992; Allen et ai. 1993; Allen & Fabian 1994). White et ai. (1991) have suggested that this is evidence for the existence of a population of cold clouds with column density N H rv 1021 - 1022

cm -2 and a covering factor of order unity.

On the other hand potential sinks and destructive processes for cold gas include (1) star formation (e.g., Allen 1995), (2) heating and evaporation via thermal conduction, mixing layers, etc (e.g., Sparks 1992; Bohringer & Fabian 1989), and shredding (e.g., Loewenstein & Fabian 1990).

If substantial amounts of cold gas do exist in the ICM, this would have implications for our understanding of (1) Cooling flows, (2) the physics of the ICM (is it a multiphase medium like our ISM?) (3) the intracluster magnetic field (can it suppress conduction?), (4) galaxy formation and evolution, and (5) Ly 0' forest systems. In order to take a census of the cold gas contribution to the ICM, we have undertaken a program of observations to search for cold gas in several forms. We summarize the results of those searches here and discuss them within the context of other searches.

2. An Arecibo Search for Atomic Gas with Broad Line Widths

A population of HI clouds in the ICM might have a velocity dispersion which is comparable to that of the cluster galaxies. A velocity dispersion (Tv rv 500-1000 km s-l corresponds to a line width of FWHM rv 1200-2400 km s-l. A weak broad line might have been missed in previous observations with low total bandwidth (Burns et ai. 1981; Valentijn and Giovanelli 1981; McNamara et ai. 1990; Jaffe 1991).

We obtained Arecibo observations with a total bandwidth of 40 MHz giving an instantaneous velocity coverage of 8000 km s-l (Table 1, see O'Dea & Payne 1995 for details). We obtain the following upper limits: (1) mass of optically thin HI with a broad range of line widths M (H) ~ 109 -1010 Mev within our Arecibo beam; (2) column density of optically thin atomic hydrogen of N(H) ~ 1019 cm- 2 which is two orders of magnitude lower than the column densities of the X-ray absorbing gas; (3) covering factor ~ few xl0-4 ; (4) number of clouds along the line of sight Nlos ~ 1.

3. A VLA Search for Cold Atomic Gas

If the hydrogen is indeed atomic, then the fact that it is not detected in searches for HI emission restricts the gas to be very optically thick, so that the signal due to emission is greatly reduced (e.g., Jaffe 1992; Daines,

COLD GAS IN THE ICM 201

TABLE 1. Arecibo HI results

Cluster Mclust Se T"pp NH Cf Nlos xI09 M0 mJy xIO~9 cm-2 xIO-4

(1) (2) (3) (4) (5) (6) (7)

A262 2.0 121.0 0.014 2.7 7.8 0.9 A397 5.4 <3.0 1.4 4.3 0.5 2A0335+096 9.6 33.0 0.034 2.6 9.0 0.9 A539 4.9 11.0 0.091 2.3 4.4 0.7 MKWls 1.9 8.0 0.24 2.2 7.4 0.7 All26 59.3 5.5 0.18 2.6 4.4 0.9 AII85 8.7 3.3 0.36 2.5 4.4 0.8 MKWll 4.6 5.0 0.36 3.1 8.8 1.0 AI983 8.1 <3.0 1.4 3.7 0.5 A2063 5.4 31.0 0.019 1.6 3.3 0.5 A2162 6.2 100.0 0.0075 2.1 9.2 0.7 AWM5 9.3 75.0 0.020 2.6 9.1 0.9 A2572 7.3 27.0 0.034 1.5 4.4 0.5 A2666 3.1 <3.0 1.5 4.4 0.5

(1) Cluster. (2) Upper limit to the mass of optically thin HI of the cluster core assuming a 'broad' line width given by the cluster velocity dispersion (1v. For these large velocity widths, the noise is limited by the systematic errors. (3) The measured continuum flux density Se (mJy) within the Arecibo beam. (4) The maximum apparent optical depth corresponding to unity covering fraction and a 3(1 detection limit. (5) The 3(1 upper limit to the column density of optically thin atomic hydrogen in the Arecibo beam using the flux integral for the cluster. (6) The upper limit to the covering factor Cf of the neutral Hydrogen within the Arecibo beam assuming the HI is optically thick and has a brightness temperature of 20 K. C f < 3(1b/[(T. - T2 .7 - TRCR)] where (1b is the 1(1 channel-to-channel uncertainty due to noise and systematic errors, T., T2 7, and TR are the temperature of the HI (spin temperature), the microwave background, and the continuum radio source, respectively, and CR is the covering factor of the radio source. (7) The upper limit to the number of clouds along the line of sight Nlos ::; 3(1T~ Vens.mbl./[~ Vcl (1 - e-T)(T. - T2.7)] where (1T is the rms brightness temperature, ~ Vens.mbl. is the FHHM of the spread in velocity of the ensemble of clouds in the beam, ~ Vcl is the FHWM of the thermal line width for an individual cloud, and T is the cloud optical depth and we have assumed a spin temperature of 20 K for the calculation (see section 5).

Fabian, & Thomas 1994). The only way to achieve sufficient optical depth given the total column densities of'" 1021 cm- 2 is to make the clouds very cold T ~ 10 K. This should make the gas easily detectable in absorption. We have obtained VLA observations with 1.6 km s-1 spectral resolution to optimize our sensitivity to very narrow lines.

Our results are given in Table 2 (see O'Dea, Gallimore, & Baum 1995 for additional details). Our observations also place upper limits of about a parsec on the maximum cloud size which would be consistent with the non-


TABLE 2. VLA HI absorption results, and non-detection statistics for the surveyed cooling-flow clusters

8v /peak 0 u Tmax NH Tel

Cluster (km S-I) (mJy beam-I) (mJy beam-I) (3u) X1018 cm-2 pc

A 496 1.37 67.5 6.51 0.34 8.5 0.5 A 1795 1.46 177.9 6.06 0.11 2.9 0.3

A 2584 1.62 851.6 6.01 0.021 0.62 0.1 A 2597 1.51 486.7 6.77 0.035 0.97 0.2 A 1795 46.60 162.1 0.31 0.0058 5.0

The columns are, in order from left to right: (1) Cluster name; (2) the velocity resolution; (3) the peak surface brightness, averaged over each band pass; (4) the (1 u) rms noise in the channel maps; (5) the maximum optical depth corresponding to unity covering fraction and a 3u detection limit; (6) the maximum HI column density assuming unity covering fraction, a line width equal to the velocity resolution in column 2, and a spin temperature T. = 10 K; and (7) the upper limit on the cloud size obtained by assuming that the non detection is due to dilution Tel < (36.S/S)I/2(T, .• /1 pC)I/2 Nc!(V)-1/2 pc, where T rs is the radius of the background radio source, S is the continuum flux density, 6.S is the depth of the line, and Ncl( v) is the number of clouds within the beam.

detection. The estimated limits on column density (for clouds in this regime of parameter space) are 2-3 orders of magnitude less than the 1021 cm-2

required to explain the X-ray absorption seen in some cooling flow clusters. Similar results are obtained by Dwarakanath et at. (1994) for additional clusters.

4. A SEST Search for Molecular Gas

If the cold gas is predominantly molecular, it may be bright in CO. The roughly 1043 ergs S-1 which is absorbed in X-rays must be reradiated somewhere. We observed three cooling flow clusters with the Swedish-ESO Submillimeter Telescope (SEST). We searched for the J=l~O line of 12CO in PKS 0745-191, Hydra A, and NGC 4696 and the J=2~lline in NGC 4696. Details are given by O'Dea et al. (1994).

We find the following upper limits: (1) Mass of molecular hydrogen M(H 2) ~ 108 - 10 M0 ; (2) Column density of molecular hydrogen N(H2) ~ few X 1020 cm-2 ; (3) covering factor of molecular gas (assuming T=20, and uniform covering in velocity) cf ~ fewx10- 4 ; (4) number of clouds along the line of sight Nlos ~ 10. Thus our observations rule out substantial amounts of warm CO in these clusters. Similar constraints can be obtained for many additional clusters (Antonucci & Barvainis 1994; Braine & Dupraz 1994; Braine et al. 1995; Braine & Wiklind 199:{; Bregman & Hogg 1988;

COLD GAS IN THE ICM 203

TABLE 3. CO results

SOURCE UC vel. res Wco N(H2) Mmot Cf Ntos

mK km/s K km/s cm-2 M(') (1) (2) (3) (4) (5) (6) (7) (8)

0745-19 4.3 10.00 < 0.57 1.6 x 1020 8 X 109 6.5 X 10-4 8.6 0915-11 5.0 9.55 < 0.66 1.8 x 1020 2 X 109 7.5 X 10-4 10.0 1246-41 5.2 9.16 < 0.66 1.8 x 1020 7 X 107 7.8 X 10-4 10.4 1246-41 J=2-+1 14 11.0 < 1.80 5.0 x 1020 2 X 108 5.3 X 10-4 7.0

(1) The Source name. (2) The channel-to-channel rms of the main beam temperature T mb.

(3) The smoothed velocity channel width in km/s. (4) The integrated CO line intensity (fTmbdV) in K km/s. Upper limits are 3u, assume a velocity width of 200 km/s (for a rectangular line profile). (5) The upper limit to the column density of CO within the SEST beam assuming a covering factor of unity. (6) Upper limit to mass of molecular gas assuming a covering factor of unity. (7) The covering factor of the gas within the SEST beam assuming a kinetic temperature of 20 K. (8) The number of clouds along the line of sight assuming a kinetic temperature of 20 K (Sect. 5).

Grabelsky & Ulmer 1990; Jaffe 1987; McNamara & Jaffe 1994).

5. What is the temperature of the gas?

Since we did not detect CO, this raises the question of whether the ga.s might be too cold to emit significantly in the CO line. We have performed an equilibrium temperature calculation balancing heating by the intracluster X-rays with cooling from the atomic and molecular lines (O'Dea et al. 1994). We find minimum equilibrium temperatures in the range 20 - 30 K. (See also Braine et al. 1995 and Voit & Donahue 1995). Thus, the lack of CO emission cannot be explained by very cold gas temperatures.

Acknowledgements. We are grateful to our collaborators Jack Gallimore, Phil Maloney, Harry Payne, Bill Sparks, and Linda Tacconi for their many contributions to the work reported here. We also thank Carolin Crawford, Megan Donahue, Andy Fabian, Tim Heckman, Craig Sarazin, and Mark Voit for many helpful discussions.

References

Allen, S. W., 1995, MNRAS, 276, 947 Allen, S. W. & Fabian, A. C. 1994, MNRAS, 269, 409 Allen, S. W., Fabian, A. C., Johnstone, R. M., White, D. A., Daines, S. J., Edge, A. C.

& Stewart, G. C., 1993, MNRAS, 262, 901 Antonucci, R. & Barvainis, R. 1994, AJ, 107, 448


Bohringer, H., & Fabian, A. C., 1989, MNRAS, 237, 1147 Braine, J. & Wiklind, T. 1993, A&A, 267, L67 Braine, J., Wyrowski, F., Radford, S. J. E., Henkel, C., & Lesch, H., 1995, A&A, 293,

315 Braine, J., & Dupraz, C., 1994, A&A, 283, 407 Bregman, J. N. & Hogg, D. E. 1988, AJ, 96, 455 Burns, J. 0., White, R. A., & Haynes, M. P., 1981, AJ, 86, 1120 Cowie, 1. L. & Binney, J. 1977 ApJ., 215, 723 Daines, S. J., Fabian, A. C., & Thomas, P. A., 1994, MNRAS, 268, 1060 Dwarakanath, K. S., van Gorkom, J. H., & Owen, F. N. 1994, ApJ, 432,469 Fabian, A. C., 1994, Ann. Rev. Astron. Astrophys., 32, 277 Fabian, A. C., & Nulsen, P. E. J., 1977, MNRAS, 180,479 Grabelsky, D. A. & Ulmer, M. P. 1990, AP J, 355, 401 Jaffe, W. 1987, A&A, 171, 378 Jaffe, W. 1991, A&A, 250, 67 Jaffe, W. 1992, In proceedings of the NATO ASI "Clusters and Superclusters of Galaxies,"

ed. A. C. Fabian, (Kluwer, Dordrecht), 109 Loewenstein, M. & Fabian, A. C. 1990, MNRAS, 242, 120 Mathews, W. G. & Bregman, J. N. 1978, ApJ, 224, 308 McNamara, B. R., Bregman, J., N. & O'Connell, R. W. 1990, ApJ, 360, 20 McNamara, B. R., & Jaffe, W. 1994, A&A, 281, 673 Mushotzky, R. F. 1992, In proceedings of the NATO ASI "Clusters and Superclusters of

Galaxies," ed. A. C. Fabian, (Kluwer, Dordrecht), 91 O'Dea, C.P., Baum, S. A., Maloney, P. R., Tacconi, 1. J, Sparks, W. B., 1994, ApJ, 422,

467 O'Dea, C. P., Gallimore, J. F., & Baum, S. A., 1995, AJ, 109, 26 O'Dea, C. P. & Payne, H. E., 1995, in preparation Sarazin, C. 1. 1988, X-ray Emission from Clusters of Galaxies, (Cambridge, Cambridge

U ni versity Press) Sparks, W. B. 1992, ApJ, 399, 66 Valentijn, E. A., & Giovanelli, R. 1982, A&A, 114, 208 Voit, G. M., & Donahue, M., 1995, ApJ, 452, 164 White, D. A., Fabian, A., C., Johnstone, R. M., Mushotzky, R. F., & Aruaud, K. A.

1991, MNRAS, 252, 72

X-RAY OBSERVATIONS OF COLD GAS IN CLUSTERS

R. M. JOHNSTONE Institute of Astronomy Madingley Road, Cambridge CB.'J OHA, UK

1. Introduction

There is now a diverse body of evidence that shows the presence of about 1012 Me of cold gas within the central regions of nearby clusters of galaxies. This gas is seen through X-ray observations of clusters by its absorption of the low-energy part ofthe spectrum. In this contribution I will briefly review the observations that show the strongest evidence for X-ray absorption and indicate what the main uncertainties are in the parameters derived from the observations. I shall also indicate what the likely physical state of the absorbing gas is and finally point out that searches for such large amounts of cold gas in other wavebands have so far not been successful.

2. Evidence for X-ray Absorption

Evidence for X-ray absorption comes from a range of instruments which have significantly different characteristics.

2.1. EINSTEIN OBSERVATORY SSS

This instrument was a single pixel spectrometer with a field of view of 6 arcmin (well matched to the size of the core of nearby clusters) and a moderate energy resolution of 160 eV over the 0.5-4.5 keY band. White et al. (1991) discovered cold gas in clusters from an analysis of archival data from this instrument. They found that single temperature models with absorption inferred from Galactic neutral hydrogen emission often gave poor fits to the data. Thirteen out of twentyone clusters in this sample required excess absorption of .-v 1021 cm-2 and eleven of these thirteen also required a second cooler emission component which was well fit by a constant pressure cooling flow. Figure 1 (left panel) shows the deficit of

205


206 R. M. JOHNSTONE

low-energy counts in Abell 2199 compared with an isothermal model with Galactic absorption.

2.2. ROSAT PSPC

In contrast with the SSS this detector gives spectra which have spatial resolution (of'" 30 arcsec), but with only 4-5 independent energy bands between 0.1 and 2.5 keY.

Spatially resolved spectral fits for Abell 478 (Allen et al. 1993) and Abell 2199 (Siddiqui 1995) show that the excess absorption is concentrated towards the centre of these clusters. In Abell 478 the excess absorption is present over at least the central 300 kpc region and LlNH '" 1 X 1021 cm-2

for a single temperature model and LlNH '" 2.5 X 1021 cm-2 for a model including a cooling-flow component.

Looking to slightly higher redshift clusters Abell 1068 (z = 0.1386) and Abell 1664 (z = 0.1276), Allen et al. (1995) has shown that the PSPC data (although not resolving the central regions of these clusters) required excess absorption of", 3 - 4 X 1021 cm-2 over that from our Galaxy.

2.3. BBXRT

The BBXRT flew on the Astro-1 mission in 1990, carrying detectors based on similar technology to the SSS. Each detector did however have 5 independent segments, giving some spatial resolution on a scale of '" 5 arcmin.

With data from this mission, Mushotzky (1992) confirms the excess absorption seen by the SSS in the Perseus Cluster and that the central region of Abell 2256 (a non-coaling-flow cluster) does not require excess absorption. Additionally data from this instrument first showed that Abell 262 requires excess absorption.

2.4. ASCA

Recently, using the CCD detectors of the ASCA SIS, which combine both a good spectral resolution with moderate spatial resolution ('" 1 arcmin), Fabian et al. (1994) have shown that the excess absorption in the Perseus cluster is in the range 1.1- 4.1 X 1021 cm- 2 , for the Centaurus cluster it is in the range 1.5 - 4.6 X 1021 cm-2 and for the cluster Abell 1795 it is in the range 2.5 - 3.5 X 1021 em -2 • The ranges of excess absorption correspond to different models fitted to the data sets. Figure 1 (lower panel) shows residuals from an isothermal model with Galaxtic absorption fitted to the SIS data from the Perseus cluster. The marked positive residuals near 1 keY due to the cooling-flow component are strongly absorbed towards lower energies.

X-RAY OBSERVATIONS OF COLD GAS IN CLUSTERS 207

A211111 S8S Kekal 4.7kIIV. Galactic ab8.

~~~--~------------~~ i :i t + i ~ -----ftt "4± f~tl+t+l++tt±t+*Nt+~~+tt+I""'¥l~ -

)(

~L-~~+~~~ __________ ~ ____ ~ ____ ~ __ ~ 125

cbllDllel .... 1117 (keY)

Peneus Cluster

0.5 1 2 5 Channel Energy (keY)

Figure 1. Upper panel (from White et al. 1991): SSS spectrum of Abell 2199 (crosses in top section), with a single temperature plasma and Galactic absorption model (solid line) fitted above", 1.5keV. The residuals (lower section) are negative below 0.9 keY due to intrinsic absorption and are positive near 1 keY due to line emission from the cooling flow. Lower panel (from Fabian et al. 1994): residuals after fitting the ASCA SIS spectrum ofthe Perseus cluster above", 3keV with an isothermal model with Galactic absorption. The strong residuals between 1-2 are keY due to the cooling flow emission are cut-off by intrinsic absorption at lower energies.

208 R. M. JOHNSTONE

2.5. UNCERTAINTIES IN THE EXCESS ABSORPTION

The statistical requirement for excess absorption is often very high, although the actual value of the excess column density derived depends strongly on the model fitted to the data. Higher values of excess absorption typically correspond to models that include (and statistically require) additional cool or cooling-flow emission components. This point has been explored too by White et al. (1994).

For the SSS data there was a build-up of ice on the detector window during observations. The amount of ice is parameterised (Arnaud et al. 1989) and modelled in the subsequent spectral fitting. Errors in this modelling could lead to uncertainties of'" few X 1020 cm-2 in the measured intrinsic cluster absorption. This effect is not relevant to data from other detectors.

A further source of uncertainty in the value of the excess column density is the abundance of oxygen. The absorption in the band that we observe is chiefly due to oxygen, but the column densities quoted are scaled to the equivalent hydrogen column density, calculated assuming a solar abundance of oxygen. If the abundance of oxygen were above the solar value then this would lead to a lower equivalent hydrogen column density, whereas a lower than solar abundance of oxygen would lead to a higher equivalent hydrogen column density. Typically, the abundances measured (mostly from the iron KG line in cluster hot gas are 1/4 to 1/3 of the solar value (Edge & Stewart 1991). However, there is a suggestion from the Einstein Observatory FPCS data (Canizares et al. 1987) that the abundance of oxygen in some clusters may be 2-3 times the solar value.

Details of the lines and line strengths included in the plasma emission codes which generate the emission spectra fitted to the X-ray data make small differences to the inferred excess absorption at the few x1020 cm-2

level.

3. Physical State of the Gas

3.1. MASS OF THE ABSORBING GAS

A simple estimate of the total mass of gas corresponding to the excess absorption can be made from

11 R 2 f1NH M '" 3 X 10 (100kpc) (1021 cm-2 )MG

So, for R = 200kpc and f1NH = 1021 cm-2 , M '" 1012 MG.


Note that this assumes that the covering fraction of the cold gas over the hot X-ray emitting gas is 100 per cent. It was found by White et al. (1991) that fits including partial covering models for the absorption constrain the covering fraction to be larger than 70 per cent. A similar figure is found for Abell 496 from ASCA data (Fabian, priv. comm.). In practice, however, we expect the situation to be much more complicated with the cold gas mixed in with the hot X-ray emitting gas in the central regions and little or no cold gas associated with X-ray emitting gas further out. What we observe is, of course, the sum of the emitted spectrum for all these different regions.

Cooling flows typically have mass deposition rates of hundreds of solar masses per year, and these have probably been maintained for the life ofthe cluster rv 1010 yr, resulting in approximately 1012 M8 of cooled material. Previously it had been assumed that the cooled gas formed low-mass, and hence difficult-to-observe, stars very efficiently. Since central cluster galaxies are not nearly as blue or bright as they would be if undergoing star formation with a normal IMF at the X-ray inferred mass deposition rate (e.g. Johnstone, Fabian & Nulsen 1987). The similarity of the mass of the observed cold gas to the total mass cooled by a cooling flow over its lifetime does suggest that low-mas star formation may not be very efficient, and in fact the observed cold gas is just that hot gas which has cooled during the lifetime of the cluster. Also relevant to this issue is the fact that the noncooling-flow clusters, Coma and Abell 2256 do not have excess absorption and hence not large amounts of cold gas.

3.2. TEMPERATURE OF THE ABSORBING GAS

In order to absorb the X-rays, oxygen atoms with at least 1 electron must be present. This constrains the absorbing gas to have T < ;{ X 106 K. In the temperature region 104 K - 3 X 106 K the emitted luminosity from 1012 M8 of gas is approximately:

M T L rv 3 X 1047( )(_)-3/2 erg s-1 . 1012 M8 105

Most of this luminosity will emerge from the gas in optical/uv emission lines. On the one hand, gas in this temperature range is clearly ruled out by observations, while on the other hand if it were in this temperature range, it would require an energy input equal to this enormous luminosity to maintain the gas within this temperature range.

These constraints require the gas to have a temperature less than rv

104 K. Detailed calculations of cold clouds irradiated by the spectrum of a cooling flow have been made (Ferland, Fabian & Johnstone 1994, Fabian, Johnstone & Daines 1994, Daines, Fabian, & Thomas 1994, O'Dea et al.

210 R. M. JOHNSTONE

1994 and Voit & Donahue 1995), and while there is some uncertainty about which models are appropriate, they all require the cold clouds to be cooler than"" 30 K, and some (Ferland, Fabian & Johnstone 1994) find temperatures for cloud cores as cold as the local microwave background.

4. Confrontation with Low-Energy Observations

It might be expected that 1012 MG of gas with a temperature in the range 3-30K would be easily detectable in nearby clusters by emission or absorption in the 21cm line of neutral hydrogen, or through CO emission or absorption in the J = 1-0 and J =2-1 rotational lines at wavelengths of 1.3mm and 2.6mm.

In recent years many workers have sought such tracers of the cold gas in clusters, in general without success. Some recent works in the literature are: Jaffe (1990, 1991), McNamara et at. (1990), O'Dea et at. (1994), Antonucci & Barvanis (1994), Braine & Dupraz (1994) while other pertinent work is presented in this conference. Upper limits to HI masses or total masses inferred from the CO observations are generally in the region of 109 MG for beams that match the central few hundred kiloparsec region of clusters where the X-ray absorption is seen.

So, why is the bulk of the X-ray absorbing gas not detected"! One possibility is that the conversion factor from the observed strength of (or limits on) the CO lines to the total gas mass present is substantially different for the conditions in cooling flows compared than it is in the Galaxy.

Alternatively large dust grains may form over the lifetime of the cold clouds in clusters (assuming they remain undisturbed) and the CO molecules then freeze out from the gas-phase on to the surfaces of these dust grains (Fabian, Johnstone & Daines 1994). This would still allow them to produce X-ray absorption but avoid detection in the mm-wave bands. Such large dust grains are probably stable to sputtering from the hot electron gas and do not produce very much optical absorption.

A second alternative is that the cold gas remains as gas, but forms more complex molecules thereby depleting the CO. Future work in this field requires the molecular chemistry in cooling-flow clouds to be explored in much more detail.

References

Allen, S.W., Fabian, A.C., Johnstone, R.M., White, D.A., Daines, S.J., Edge, A.C. and Stewart, G.C., 1993. MNRAS, 262 901.

Allen, S.W., Fabian, A.C., Edge, A.C., Bohringer, H. and White, D.A., 1995. MNRAS, 275, 741.

Antonucci, R. and Barvainis, R., 1994. Al, 107, 448. Arnaud, K.A., Szyrnkowiak, A., and White, N., 1989. HEAD Newsletter, Vol 1, No 2.


Braine, J. and Dupraz, 1994. Af9A, 287, 407. Canizares, C., Markert, T.H. and Donahue, M.E., 1987. In "Cooling Flows in Clusters

and Galaxies", Kluwer Academic Publishers, ed. Fabian, A.C., p63. Daines, S.J., Fabian, A.C. and Thomas, P.A., 1994. MNRAS, 268, 1060. Edge, A.C. and Stewart, G.C., 1991. MNRAS, 252, 414. Fabian, A.C., Johnstone, R.M. and Daines, S., 1994. MNRAS, 271, 737. Fabian, A.C., Arnaud, K.A., Bautz, M.W. and Tawara, Y., 1994. ApJ, 436, L63. Ferland, G., Fabian, A.C. and Johnstone, R.M., 1994. MNRAS, 266, 399. Jaffe, W., 1990. Af9A, 240, 254. Jaffe, W., 1991. Af9A, 250, 67. Johnstone, R.M., Fabian, A.C. and Nulsen, P.E.J.N., 1987. MNRAS, 224, 75. McNamara B.R., Bregman J.N., and O'Connell R.W. 1990. ApJ, 360, 20. Mushotzky, R.F. 1992. In "Clusters and Superciusters of Galaxies", Kluwer Academic

Publishers, ed. Fabian A.C., p91. O'Dea, C., Baum, S., Tacconi, L.J., Maloney, P.R., Sparkes, W.B., 1994. ApJ, 422,467. Siddiqui, H., 1995. PhD Thesis, University of Leicester. Voigt, G.M., and Donahue, M., 1995. ApJ, 452, 164 White, D.A., Fabian, A.C., Johnstone, R.M., Mushotzky, R.F. and Arnaud, K.A., 1991.

MNRAS, 252,72. Whit.e III, R.E., Day, C.S.R., Hatsukade, I., and Hughes, J.P., 1994. ApJ, 433, 583.

ABSORPTION MEASUREMENTS

ABSORPTION MEASUREMENTS OF MOLECULAR GAS

F. COMBES DEMIRM, Observatoire de Paris 61 Ave. de l'Observatoire F-75014 Paris, FRANCE

AND

T. WIKLIND

Onsala Space Observatory 5-43 992 Onsala, Sweden

Abstract. Molecular absorption in front of radio-loud quasars is a unique tool to probe cold gas at high redshifts, with the benefit of high angular and velocity resolutions. Up to 30 molecular lines have been detected in a single object, and 4 absorbing systems have been detected until now. Molecular abundances, and excitation temperatures can be studied at length in remote galaxies. The column densities sampled by this technique are between N(H2) = 1020 to 1024 cm-2 ; since the cross section for absorption is very small around a galaxy, the absorbing systems are either internal, or coming from a gravitational lens. The molecular measurements can then contribute to cosmographic studies.

1. Introduction

Emission and absorption measurements do not sample the same gas, they are complementary to determine the temperature and density ofthe medium. Emission is strongly biased in favor or warm and excited gas, and there can exist large quantities of molecules unobserved because they are not excited above the cosmic background temperature Tbg' On the opposite, the absorption is biased in favour of cold gas, and does not trace dense regions, especially since they have a low surface filling factor. For optically thin emission, the integrated signal (fTadv) is proportional to NtotT-le-Eu/kT(ehv/kT -l)(J(T) - J(Tbg)), where N tot is the total column density, Eu the upper

215


216 F. COMBES AND T. WIKLIND

level energy, and J(T) = hvjkj(ehll / kT -1). It vanishes at low temperature, when T ~ Tbg. For the absorption, the integrated signal, or the integrated optical depth J rdv is proportional to NtotT-le-EI/kT(l- e-hll / kT ). In particular, when the absorption is from the fundamental level El = 0, the signal is ex: NtotT-l(1- e-hll/ kT ), considerably enhanced at low temperature T.

This temperature T is not the kinetic temperature Tk, but the excitation temperature Tex determining the fractional populations in the various energy levels of the molecule (adopting the weak LTE hypothesis, Tex is the same for all levels ). Molecules are generally excited by collisions, and their Tex depends in fact on the H2 density. For example the critical density for which collisions can balance spontaneous emission, and Tex = Tk is 4 X 104

cm-3 for the CO molecule (low dipole moment), and 1.6 X 107 cm-3 for the HCN molecule, and other high density tracers with high dipole moment, such as HCO+. For density much lower than that, only the low energy levels of the rotational ladder are populated, corresponding to radiative equilibrium with the cosmic background temperature (Tex ~ Tbg)'

If both dense gas and diffuse gas are present on the line of sight, with the same total column density, the strongest absorption will be due to the diffuse gas, and the apparent excitation temperature could be very low.

2. Galactic versus extragalactic absorptions

Molecular galactic absorption has been detected towards Cas A or the Galactic center more than a decade ago (Encrenaz et ai. 1980, Linke et ai. 1981), but only towards very strong and extended galactic continuum sources. More work on these lines, using extragalactic radio sources, had to await the advent of millimetric interferometers. Absorption in most cases is indeed buried among strong emission from galactic clouds, when observing with a single dish. Even when absorption only is detected, it is impossible to deduce reliable optical depths and column densities because of emission confusion (e.g. Liszt & Lucas 1995). With interferometers, there has been recently a renewed interest in galactic molecular absorption measurements (e.g. Marscher et al. 1991, Lequeux et al. 1993, Liszt & Lucas 1994, Hogerheijde et al. 1995). Among the surprises, it has been found that abundances vary strongly from one line of sight to the other, by factors 10-100, and that HCO+ and HCN are very abundant in diffuse clouds. This was impossible to see through emission measurements, since these high dipole moment molecules are not excited there. In some cases, HCO+ absorption is found without any CO absorption. These abundances are incompatible with current equilibrium chemical models, and it has been suggested that high tubulence can modify the chemistry (Hogerheijde et al. 1995, Falgarone et al. 1995). Basically, energy coming from large chaotic motions accelerates

ABSORPTION MEASUREMENTS OF MOLECULAR GAS 217

reactions, especially endo-thermic ones, or even can help to overcome potential barriers. The time scale to reach chemical equilibrium being larger than the turbulence time scale, the turbulent gas can be maintained far off chemical equilibrium. This would explain the high HCO+ abundance found in diffuse clouds (enhanced by 100), and also the broad line widths (HCO+ absorptions are significantly wider than the CO ones).

Paradoxically, it is much more easy to observe extragalactic absorption lines. Unresolved emission from a galaxy falls down as the square of distance, while the absorbing signal does not depend on distance, it is as easy to detect at any redshift (of course smaller than that of the radio source behind). As soon as the redshift of 0.005 is reached, emission is no longer a problem, and an interferometer is not necessary. The first extragalactic molecular absorption lines in the millimeter range were found in Centaurus A (Israel et al. 1990, Eckart et aZ. 1990), towards the edge-on nuclear disk: the line widths are quite small (5 - 10 km s-l), abundances appear normal for dark clouds. A few other molecules had already been detected in absorption in the em range (OIl, H2 CO, C3 H2 , e.g. Gardner & Whiteoak 1979, Bell & Seaquist 1988).

The first high-z molecular absorption in the mm range was found towards the BL Lac object PKS1413+135 (Wiklind & Combes 1994a). Since then, 4 systems in total have been discovered, that are listed in Table 1 (Wiklind & Combes 1995, 1996a,b). The redshifts of absorption vary between 0.25 to 0.9. At least two of the systems are confirmed gravitational lenses, with multiple images and even an Einstein ring (B0218+:{57).

TABLE 1. Absorption systems detected ill the millimeter range

Galaxy Continuum source Absorber z

Milky Way SgrA, CasA, QSOs Nuclear disk, spiral arms O.

NGC5128 Cen A Nuclear disk, spiral arms? 0.002

PKS1413+135 BL Lac, core-jet radio source Spiral arms? 0.247

Q1504+377 Core-jet radio source Nuclear disk, spiral arms 0.673

B0218+357 Core-jet, A-B images+E-ring Lens 0.685

PKS1830-211 A-B images +jet? Leus 0.886

Milky Way - Linke et al. (1981), Marscher et al. (1991), Lucas & Liszt (1994); NGC5128 - Eckart et al. (1990); PKS1413+135 - Wiklind & Combes (1994); Q1504+377 - Wiklind & Combes (1996b); 80218+357 - Wiklind & Combes (1995); PKS1830-211 - Wiklind & Combes (1996a)

Absorption studies are a unique way to probe cold gas at high redshift, they are sensitive to a single molecular cloud (only 1 Me;) is enough to


produce a detectable signal), while emission can only detect 1011 Mra or so at high redshifts. Molecular absorption measurements benefit from a high spatial resolution, since the high-z radio sources have an angular size of the order ofthe milli arcsecond (mas) or less in the millimetric domain. At z = 0.5-1, the corresponding sizes on the absorbing galaxy are a few parsec, and diffuse gas from a single molecular cloud can cover entirely the source. This results in an optically thick absorption (T ~ 1), while in the centimeter domain, the radio sources are much more extended (synchrotron emission from the jets essentially), and the apparent T is very low, which makes the H I absorption measurements more difficult. The power of the absorption signal is due to the strength of the radio sources: while we observe a typical antenna temperature of 100mK with a beam of 20", an instrument able to resolve the 0.5 mas source would detect a signal of 1.6 X 108 K!

The greatest difficulty is to find the absorbing system, since they are much rarer than in other wavelength domains. The H2 column densities sampled vary from 1020 cm-2 (PKS1413+135, T ~ 2, ~v ~ 2 km s-1), to 1024 cm-2 (B0218+357, T ~ 2000, Llv ~ 20 km s-1), while in the optical, the Lya systems sample H I column densities from 1013 cm-2, MgII systems, from 1017 em -2, and 21 cm absorption, from 1019 em -2. Our observational upper limits are for CO 1015 cm-2, and for HCO+ 1012 cm-2. The 21cm a.bsorption cross section corresponds to the extended disk of galaxies, while the molecular gas has a much smaller extent: we therefore expect a lower cross section (but more statistics are needed, to draw any conclusion). For the line-of-sight towards a quasar to intercept the restricted molecular disk, the impact parameter must be quite small, and this implies in general a gravitational lens phenomenon. The critical stellar surface density within a few kpc from the center is in general above the critical one for multiple images, that explains the frequency of lenses in our absorbing systems: either there is a lens or the absorption is internal, i.e. comes from the galaxy associated to the radio source.

3. Selection of candidates

Up to now, we have observed 50 systems, selected to be good candidates for molecular absorptions, according to several criteria:

first, a necessary condition is to find a strong enough continuum source in the millimeter domain. They are much less numerous than in the cm range, since the spectrum of synchrotron radio sources is steep. The minimum for our selected sample wa.s 0.15 Jy at 3 mm. then, we selected candidates among systems already detected in absorption either at 21 cm, or damped Lya systems for z > 2 (but with negative results, e.g. Wiklind & Combes 1994b), or optical Mg II sys-


terns (0.4 < Z < 2) and Call systems ( z < l.3). or, even in the absence of previously detected absorptions, we selected systems with at least a known redshift, gravitational lens candidates from the VLBI radio morphology (the redshift could be that of the lens), BL Lac, strongly variable sources, or very red objects with a galaxy on the line of sight, obscured quasars (Webster et ai. 1995), and radio sources from the 1 Jy survey, with flat spectrum, and coinciding with an optically detected galaxy (Stickel & Kiihr 1993). finally, we also tried objects following the same criteria as above, but without any known redshift. This is the less biased search, since heavily obscured systems (with a molecular absorption) should not be detected in the visible. In this case, the millimeter continum level has been chosen larger than 0.5 Jy. The PKSI8~~0-211 system, chosen for being a gravitational lens, was detected by s11ch a search (Wiklind, this meeting).

We report elsewhere all negative results, with their upper limits (Wiklind & Combes 1995, 1996b). On the 50 systems searched for, the detection rate is about 10%. The probability is even lower than for H I absorption, since most 21 cm absorbers have not been detected in molecules: for example the well known BL Lac AO 02~~5+164 (ze = 0.94, z" = 0.524), although quite strong in the 3mm continuum (0.7.Jy) was not detected with a good upper limit. Of the 10 high-z 21 cm absorbers listed in the Carilli (1994) review, 6 had high enough 3 mm continuum, but only 2 were detected (PKSI413+135 and B0218+357).

4. Discussion of individual cases

4.1. PKS141~H135, ZA = 0.247

PKS1413+135 has been classified as a BL Lac object in the optical, since there is no strong emission lines. It is a peculiar object in the infrared, where the point source is highly variable: it has varied by 20% at 2.2 J.lm in 1 day (Bregman et ai. 1981). However, it could be a misclassified BL Lac, since there is even no point source in the optical, as if the central AGN was highly obscured. Instead, there is an edge-on spiral galaxy, with normal weak emission lines and stellar absorption features, located at less then 0.1" of the AGN (McHardy et ai. 1991, Stocke et ai. 1992). The recent HST image from McHardy et ai. (1994) reveals the presence of a conspicuous dust lane in the middle of the elongated image.

In the radio, it is a flat spectrum, variable and compact source, which appears to be a mini-triple (core+.iet+counterjet) in the VLBI, with tens of milli-arcsecond (~30pc) scale (Perlman et al. 1994,1995). The presence of the counterjet rules out the standard jet geometry accounting for BL Lacs,


Frequency (~Hz)

92460 92458

"'-'1 ""'-1 -1'-'-'-

0.3

0.2

0.1

O~~LL~-L~-L~~LL~-L~ -10 -5 o

Velocity (km/s)

Frequency (~Hz)

5 10

106750 106700 106650 106600 106550

0.07 ._-

0.06

o

1504+377

HCO+(2-1)

200

Velocity (km/s)

400

Frequency (~Hz)

130360 130340 130320 130300

0.02

o

-100 -50 o 50 100

Velocity (km/s)

Frequency (~Hz)

96200 96150 96100 0.09

0.08

0.07

0.06

0.05

-200 -100 a 100 200

Velocity (km/s)

Figure 1. Synoptic view of the 4 detected molecular absorption systems. Line widths are quite different, from ~ 1 km S-l to 100 km S-l. Signals are in T~ (K).

where the relativistically beamed continuum mask the strong emission lines. Also, BL Lac objects are believed to have elliptical hosts, which does not fit with the presence of the spiral galaxy there.

Stocke et al. (1992) have derived from the heavily absorbed X-ray spectrum an extinction of at least Av = 30 mag, or N(H) > 1Q22cm-2 , and suggest that the AGN is a background source for the spiral galaxy, since there is no evidence for the absorbing gas to be heated by the AGN (bright emission lines, or mid-IR. thermal continuum). However, the probability of such coincident projected positions is very low, and also would imply gravitationallensing with multiple images, which is not seen in the radio VLBI morphology. To avoid images, the source should be very close to the spiral galaxy, at about of few percent of its distance to the observer, which also is highly unlikely.

Carilli et al. (1992) detected H I 21 cm redshifted absorption, with an apparent optical depth of T = 0.3, which implies N(HI) = 1.31021 cm -2,


-4 -2 o

PKS1413+135

CO(I-O)

2 4

Velocity (km s-I)

-4 -2 o

PKS1413+135

HCO+(2-I)

2 4

Velocity (km 8- 1)

Figure 2. Comparison of the CO(I-D) and HCO+{2-1) optical depth spectra towards PKS1413+135, at 40ms-1 velocity resolution. Here we plot T = -In{max{Tl,a}/Tcont}.

if Ts = 100 K is assumed, with a filling factor of 1. We detected towards PKS1413+135 several molecular lines, including CO(1-0), HCO+, HCN and HNC, all three in two different transitions, J= 2.-1 and J=3.-2. However, the isotopic lines were not detected e3 CO(l-0), H13CO+(2-1)) with very good upper limits, suggesting optically thin gas. Also, rarer molecules such that N 2 H+, eN, etc. were not detected. One peculiarity of this absorbed system is the very narrow line widths: in CO(1-0) the spectrum splits in two components, and one of them is narrower than 1 kms-1 (a velocity resolution of 40 m s-1 was necessary to study them, cf. Fig. 2).

This is just slightly higher than the H2 thermal width at 10 K: 0.25 km s-1. This means that there is only a small turbulence. The HCO+ profile is wider, a phenomenon already found in absorbing galactic diffuse clouds. This is not a saturation effect, since the H13CO+ is not detected, but probably an abundance effect. Abundances and excitation are quite typical of a diffuse cloud like ( Oph (Hogerheijde et al. 1995). We derived a covering factor of at least 80% of the continuum source, and a T( CO) = 1. 7. The CO and H I central velocity coincide (there was a technical error in the CO velocity previously reported by Wiklind & Combes 1994b), but the H I line width is much broader, which could be due to the much more extended continuum source, at cm wavelengths.

4.2. B0218+357, ZA = 0.685

B0218+357 is also a BL Lac object, which has been identified as a gravita,tional lens on basis of its radio structure (O'Dea et al. 1992, Patnaik


et al. 1993, Blanford & Narayan 1992). Its radio morphology consists of two compact flat spectrum objects and a steep spectrum ring. The separation of the two point sources (A to the SW and B towards the NE; see Patnaik et al. 1993) is 335 milli-arcseconds (mas), which is similar to the diameter of the ring. The B-component is situated inside the ring, only 40 mas from the center, whereas the A-component is placed well outside the ring. The A and B components, of similar spectrum index and polarisation, are interpreted as the two images of the compact core, while the ring (dubbed "the smallest Einstein ring") represents the image of a jet structure in the background source, possibly with one or two hot spots. Optical spectroscopy shows narrow emission lines of [0 IIj-\3727 and [0 mj-\5007, absorption lines of Ca II K and Hand Mg II, at a red shift of z = 0.685, as well as very tentative emission lines of Mg II and Fe II at z = 0.94 (Browne et al. 1993).

Adopting the redshifts 0.685 and 0.94 for the lens and the background source, the required mass for a spherical lens inside the Einstein ring is 1.8 X

1010 M ra , which corresponds to a spiral galaxy bulge. The recent VLBA map at 0.5rilas resolution of Patnaik et al. (1995) reveals a further decomposition A1-A2, B1-B2, which requires a non-spherical lens potential.

The intensity ratio AlB is ",3 at 5, 8.4, 15 and 22 GHz, but a change in the flux of the B-component of ",10% in a few mouths has been seen (O'Dea et al. 1992, Patnaik et al. 1993). In the optical, recent NOT observations have shown that only B is detected, A being certainly heavily obscured (Grundahl & Hjorth 1995).

The intervening galaxy is most likely a late type disk galaxy, since 21 cm H I absorption has been detected by Carilli et al. (1993), at a redshift of z = 0.68466. The implied column density is N(HI) = 4 X 101S Tsl f cm-2, where Ts is the spin temperature of the gas, and f is the H I covering factor. The peak optical depth is only 0.048, if f is assumed to be l.

Absorption of molecular rotational lines, CO(1-2),(2-3), HCO+(1-2) and HCN(1-2), was detected at the same redshift as the H I absorption (Wiklind & Combes, 1995), indicating large column densities of molecular hydrogen. The molecular lines are saturated, except the HCO+ line, since H13CO+ is not detected. This again suggests that the HCO+ is coming from the diffuse envelope, where its abundance is highly enhanced. The CO transitions are highly saturated, since both the 13CO and C1S0 are detected at the same level, implying also saturation of both lines. We observed the C l70 to estimate the optical depth of C1SO, which is about 3. The optical depth of the 12CO is therefore about 1500, which leads to an H2 column density of abou t 1024 cm -2.

Due to this high column density, B0218+357 is a good candidate for a search of molecular oxygen in the interstellar medium. This essential ele-


ment, expected by chemical models to be of similar abundance to CO, has never been detected in the Galaxy, because of the atmospheric absorption. Searches of 0 180 in the Galaxy, or O2 in emission in redshifted galaxies, like NGC 6240, have only brought upper limits (e.g. Combes et al. 1991, Liszt 1992). The best upper limit until now comes from redshifted galaxies, but the O2 could suffer a higher dilution factor than CO, and it is necessary to observe an individual molecular cloud, avoiding dilution. The non-detection of the O2 424 GHz line in absorption in B0218+357 implies an abundance ratio 02/CO < 1.2 X 10-2, implying that the C/O ratio becomes larger than 1 in the gas phase (certainly oxygen is depleted on grains, Combes & Wiklind 1995). Other molecules, such as CN, CS and H2CO have been detected, in various excitation levels ( 2<-1, ~J<-2, 4<-;J), but H13CO+, H13CN, OCS and CCH have not been detected. Abundances of the dense medium appear comparable to those of a typical dark cloud in the Milky Way, and the excitation temperature is slightly larger than the cosmic background temperature of 2.76(1 + z) = 4.6 K.

As mentioned above the B-component is situated very close to the center of the intervening galaxy (rv 200 pc) while the stronger A-component is 1.8 kpc further out in the disk. If the A/B intensity ratio is :J, then our covering factor f > 0.8 implies that both A and B are covered by molecular clouds. While the 21 cm H I absorption line consists of only a single broad component with a linewidth of rv 30 km s-1, in the CO absorption there appears to be two blended components which can be decomposed in two gaussians separated by rv 10 km s-1. This is too close a separation, for a galactic center, where we expect to find the rotational gradient. Unless the lens is seen almost exactly face-on (which is not likely, in view of the required ellipticity of the lens), or A and B are exactly on the minor axis, the solution is that the A/B ratio is more around 4 in the mm domain, and that only the Acorn ponent is covered (as suggested by the heavy obscuration observed by Grundahl & Hjorth 1995). It is interesting to note that we found a molecular cloud at 1.8 kpc from the center of the galaxy, but not at the very center (component B).

4.3. Q1504+377 ZA = 0.673

This is a flat spectrum radio source, composed of a core and a jet, and detected with 0.9 Jy at 3 mm (Steppe et al. 1993). In optical, there is no conspicuous point source, but instead a faint galaxy is seen on deep R-band CCD images (Stickel & Kiihr 1994). The spectra reveal narrow emission lines ([0 III], H,fJ, [Ne III], [0 II]) superposed on a weak continuum. The object is detected in the far-infrared with IRAS, with a flux of 0.19.Ty at 60,""m . Again, it appears as though a galaxy is seen on the same line of


sight as the AGN and is obscuring it in the optical. The R-band CCD images show clearly that the galaxy is an edge-on spiral.

We detect towards this system, at the redshift of the galaxy, two absorbing components, separated by 330 km s-1 (Fig. 1). Since for a galaxy viewed edge-on, in circular rotation, we should see only one component at systemic velocity, this suggests that one of the component is in highly non-circular motion, such as in a bar, or elongated nuclear ring. This phenomenon is frequent in the center of spiral barred galaxies, including our own (d. Combes 1994). Since the very wide component is more likely to come from the nuclear ring, we interprete the low redshift component as an outer spiral arm, at the systemic velocity, while the nuclear ring displays non-circular motion at v = +330 km s-1, in apparent infall towards the nucleus.

As for PKS1413+135, the galaxy is coinciding with the radio source, and the radio morphology does not reveal any gravitational lens, so we tend to conclude that the absorption is internal to the AGN host.

We detected in this system CO(2-1) and (3-2), HCO+(2-1), (3-2) and (4-3), HCN(2-1), HNC(2-1) with different ratios between the two components for the various lines. From the HCO+(3-2), we can infer the density of the components, and from the HNC(2-1) the actual temperature of the medium (Tk = 16 K), since the HCN fHNC ratio acts as a chemical thermometer (Wiklind & Combes 1996b). Molecular lines, such as 13CO(2-1) are undetected, while HNC(3-2) is tentatively detected. The absorption level is not equal to the continuum level, which could be due either to low covering factor, or low intrinsic optical depth. From the non-detection of the isotopic line 13CO(2-1), we tend to favor the second hypothesis, especially since the absorption levels vary for the various detected lines. The optical thickness is then T ~ 0.1 - 0.3 for CO and HCO+, and the column densities implied N(H 2 ) ~ 1020 - 1021 cm-2 . The excitation temperatures are around 10-15 K, and vary between components (even between the several components of the nuclear ring absorption).

4.4. MOLECULAR TORI

Molecular absorption lines could be a way to test the presence of the dusty tori required around AGN in the unification model. Barvainis & Antonucci (1994) have tried to observe CO( 1-0) in absorption towards Cygnus A without success. We improved their upper limits by about a factor 6 with the IRAM-30m telescope, and the upper limits on the optical depth is now a few percents. We also did not detect HCO+(2-1), and CO(2-1). The same kind of negative results is found in Hydra A. Maloney et al. (1994) have discussed the possible interpretations: the gas in the torus could be


completely ionised (but H I absorption has been detected towards Cygnus A, Conway & Blanco 1995, and Hydra A, Dwarakanath et ai. 1995), or the molecular gas could be radiatively excited at a high temperature. Even in the latter hypothesis, we should have detected the absorption, if the covering factor is high enough to obscure optically the AGN.

We observed also a certain number of red and obscured radio-loud QSOs, at the red shift of the AGN, in the hope of detecting the molecular tori. We selected strong mm continuum sources from the catalog of obscured quasars from Webster et ai. (1995). The colours of some of these ob jects are surprisingly red (B - J( ;:::: 8 mag), and Webster et al. (1995) claim that may be 80% of optical QSOs are missed through obscuration. We obtained only upper limits, with typically N(Hz) < 3 X 1019cm -Z.

5. Conclusions

Molecular absorption studies open new possibilities to probe cold gas at high redshift. Hz column densities from 1020 cm- 2 to 1(}24 cm-2 can be sampled, at the high end of the neutral hydrogen distribution. The technique benefits from high spatial resolution (since the typical size of the millimeter continuum sources is 1 mas or less), and high velocity resolution, from the heterodyne detection method. Since the absorbing signals are as easy to obtain at any redshift, it is a good opportunity to observe individual molecular clouds at frequencies impeded by atmospheric absorption at z = 0 (0 2 , H2 0, HCO+(2-1), HNC(2-1), etc.).

The cross section for molecular absorption is small around a galaxy, and the detected systems are either internal absorption, or from a gravitational lens. There are many interests in studying these objects: first, it is possible to study in detail abundances and excitation temperatures; since absorption is biased towards un-excited diffuse gas, in radiative equilibrium with the cosmic background temperature, we can determine Tbg

as a function of redshift. Also, for the gravitational lens systems, molecular absorption lines contribute to refine the geometry of the lens, explain the intensity ratios (through obscuration), and determine the time delays (through monitoring the absorption and continuum levels, since only one of the image is absorbed). It is possible therefore to contribute to cosmography measurements. However, clumpy molecular clouds at those redshifts can produce micro-lensing, preferentially on one image, which complicates the time delay estimation.

References

Barvainis R., Antonucci R. 1994, AJ 107, 1291 Bell M.B. Seaquist E.R. 1988, ApJ 329, L17


Blandford R.D., Narayan R. 1992, ARAA 30, 311 Bregman J.N., Lebofsky M.J., Aller M.F., et al. 1981, Nat.ure 293,714 Browne, I.W.A., Patnaik, A.R., Walsh, D., Wilkinson, P.N. 1993, MNRAS 263, L32 Carilli, C.L. 1994, in 6th Asian-Pacific Regional Meeting, to appear in JAA Carilli, C.L., Perlman E.S., Stocke J.T. 1992, ApJ 400, L13 Carilli, C.L., Rupen, M.P., Vanny, B. 1993, ApJ 412, L59 Combes F. 1994 in "Nuclei of normal galaxies: Lessons from the Galactic Center", ed.

R. Genzel & A.I. Harris, p. 65 Combes, F., Casoli F., Encrenaz P. Gerin M., Laurent C. 1991, A&A 248,607 Combes, F., Wiklind T. 1995, A&A 303, L61 Conway J.E., Blanco P.R. 1995, ApJ 449, L131 Dwarakanath K.S., Owen F.N., van Gorkom J.H. 1995, ApJ 442, L1 Eckart A., Cameron M., Genzel R. et al. : 1990, ApJ 365, 522 Encrenaz P.J., Stark A.A., Combes F., Linke R.A., Lucas R., Wilson R.W. 1980, A&A

88, L1 Falgarone E., Pineau des Forets G., Roueff E.: 1995, A&A 300, 870 Gardner F.F., Whiteoak J.B.: 1979, MNRAS 189, 51P Grundahl F., Hjorth J.: 1995, MNRAS 275, L67 Heisler J., Ostriker J. 1988, ApJ 332, 543 Hogerheijde M.R., de Geus E.J., Spaans M., van Langevelde H.J., van Dishoeck E.F.,

1995, ApJ 441, L93 Israel F.P., van Dishoeck E.F., Baas F. et at.: 1990, A&A 227, 342 Lequeux J., Allen R.J., Guilloteau S.: 1993, A&A 280,23 Linke R.A., Stark A.A., Frerking M.A.: 1981, ApJ 243, 147 Liszt H.S. 1992, ApJ 386, 139 Liszt H.S., Lucas R.: 1994, ApJ 431, L1.31 Liszt H.S., Lucas R.: 1995, A&A 294,811 Lucas R., Liszt H.S.: 1994, A&A 282, L5 McHardy I.M., Abraham R.,Crawford C.S. et al: 1991, MNRAS 249, 742 McHardy I.M., Merrifield M.R., Abraham R.,Crawford C.S.: 1994, MNRAS 268, 681 Maloney P.R., Begelmall M.C., Rees M.J. 1994, ApJ 432,606 Marscher A.P., Bania T.M., Wang Z.: 1991, ApJ 371, L77 O'Dea, C.P., Baum, S.A., Stanghelli, C., Van Brengel, W., Deustua, S., Smith, E.P. 1992,

AJ 104, 1320 Patnaik, A.R., Browne, I.W.A., King, L.J., Muxlow, T.W.B., Walsh, D., Wilkinson, P.N.

1993, MNRAS 261, 435 Patnaik, A.R., Porcas R.W., Browne, I.W.A., 1995, MNRAS 274, L5 Perlman E.S., Stocke J.T., Shaffer D.B., Carilli C.L., Ma C.: 1994, ApJ 424, L69 Perlman E.S., et at. 1995, ApJ preprint Steppe, H., Paubert, G., Sievers, A., Reuter, H.P., Greve, A., Liechti, S., Le Floch, B.,

Brunswig, W., Menendez, C., Sanchez, S. 1993, A&AS 102, 611 Stickel M., Kiihr H. 1993, A&AS 100, 395 & 101, 521 Stickel M., Kiihr H. 1994, A&AS 105,67 Stocke J.T., Wurtz R., Wang Q., Elst.on R., Jannuzi B.T. 1992, ApJ 400, L17 Webster R.L., Francis P.l., Peterson B.A., Drinkwater M.l., Masci F.l., 1995, Nature

375, 469 Wiklind, T., Combes, F. 1994a, A&A 286, L9 Wiklind, T., Combes, F. 1994b, A&A 288, L41 Wiklind, T., Combes, F. 1995, A&A 299,382 Wiklind, T., Combes, F. 1996a, Nature 379, 139 Wiklind, T., Combes, F. 199Gb, A&A in prep

A NEW MOLECULAR ABSORPTION LINE SYSTEM

The Gravitational Lens PKS1830-211 at z=O.88582

T. WIKLIND Onsala Space Observatory S-4,"] 992 Onsala, Sweden

AND

F. COMBES DEMIRM, Observatoire de Paris 61 Ave. de l'Observatoire F-75014 Paris, FRANCE

1. Introduction

The detect ability of molecular absorption lines is largely independent of the excitation of the gas and can give accurate column densities, abundance ratios and excitation temperatures for a wide range of different gas conditions. Moreover, when seen towards distant radio-loud quasars, the small angular extent of the background source makes the observational sensitivity of absorption lines independent of distance (Combes, these proceedings).

Another advantage with molecular lines is that the energy of the first rotationallevel of several abundant molecules corresponds to temperatures in the range 4-5 K, while the second level corresponds to 12-15 K. Excitation of rotational lines is governed by collisions with H2 molecules and coupling to an ambient radiation field. In diffuse molecular clouds the radiation field is likely to be dominated by the cosmic microwave background (CMB) radiation, while collisional excitation can be low. Big Bang cosmology predicts that the temperature of the CMB radiation increases linearly with redshift; Tcmb(Z) = Tbb(1+z), where Tbb = 2.726±0.0010 is the present va.lue. Thus, at redshifts z 2: 0.5, Tcmb becomes comparable to the J=l rotational level of several abundant molecules, while it remains less than the J =2 level as long as z < 3.4. Molecular absorption lines therefore has the potential to give upper limits to the CMB temperature at high redshifts.

The main difficulty in observing molecular absorption lines at high redshift is to identify and derive the redshift of the absorbing system. Back-

227


228 T. WIKLIND AND F. COMBES

ground sources can easily be selected from radio-source catalogs, but the presence of molecular gas along the line of sight means that the background QSO is heavily obscured, making it difficult to obtain an optical spectra. However, observations of molecular absorption have shown that the lower rotational transitions of CO, HCO+, HCN and HNC are likely to be saturated, even in the case of diffuse molecular gas (Wiklind & Combes 1994, 1995). Hence, given a sufficiently strong background continuum source, it is possible to search for molecular absorption in the millimeter band by repeatedly tuning the receiver and integrating for a relatively short time at each tuning. For redshifts z S 3.3 one or more of the J=1<--O, J=2f-l or J =3f-2 lines of CO, HCO+, HCN and HNC will fall within the 3-mm band (80-115 GHz), except for a small interval between z = 0.44 - 0.55. The latter interval can, however, be covered by observing a restricted part of the 2-mm band (149-160GHz).

2. The gravitational lens PKS1830-211

Since molecular gas is usually found in the central regions of galaxies, a small impact parameter between an intervening galaxy and a background continuum source will increase the likelihood of molecular absorption. It also means that the systems are likely to be gravitationally lensed. One such system is the strong, fiat-spectrum radio source PKSI830-211, which has a radio morphology consisting oftwo compact components separated by I" and connected by an elliptical ring (Subrahmanyan et al. 1990; Jauncey et al. 1991). This is interpreted as two lensed images of a background quasar consisting of a core and a jet (Nair et al. 1993). The extended feature of the lensed images makes PKS 1830-211 ideally suited for modeling of the potential of the lensing galaxy and for deriving cosmographic parameters. One major drawback has been the unknown redshift of both the lens and the source. Whereas the latter is not necessary for cosmographic considerations, the red shift of the lens is crucial (Narayan 1991). Deep imaging in several optical bands has failed to detect an optical counterpart (Djorgovski et al. 1992) and a search for redshifted 21 cm HI absorption was also unsuccessful (McMahon et al. 1993).

2.1. A SEARCH FOR ABSORPTION TOWARDS PKS1830-211

The lens models constructed for PKSI830-211 implies that the line of sight towards the background source passes within a few kpc from the center of the lensing galaxy and the galaxy has a high inclination. PKSI830-211 is a strong continuum source at mm-wavelengths (2:: 2 Jy) and we selected it as a good candidate to search for molecular absorption lines in the 3-and 2-mm bands using the 15-m SEST telescope. The SEST telescope is

A NEW MOLECULAR ABSORPTION LINE SYSTEM 229

"l HCN(2-1)

<0 o

~ HCN(3-2)

<Xl o

"l CS(3-2) -

'" o

Velocity (km 5-1)

HCO+(2-1) HNC(2-1)

HCO+(3-2) HNC(3-2)

Velocity (km 5-1)

Figure 1. Molecular absorption lines at z = 0.88582 towards PKS1830~211. The continuum levels have been normalized. The velocity resolution is ~4 km S~l .

equipped with low-noise SIS receivers with an instantaneous bandwidth of 1 GHz and has a remote tuning facility, making rapid changes of the observed frequency possible. This allowed us to search over a large part of


the 3-mm band within a few hours. When the first molecular absorption line was detected we searched at selected frequencies in the 3- and 2-mm bands to establish the identity of the line and obtained the redshift of the absorbing system to be z = 0.88582 ± 0.00001. In total we observed 12 transitions: the J=2f-1 and J=3f-2 lines of HCN, HCO+, HNC, N2 H+ and H13CO+, and the J=3f-2 and J=4f-3 lines of CS (Fig. 1). A possible HI absorption at z ~ 0.2 has recently been reported by Lovell et ai. (ATNF News no. 27). No molecular absorption is seen at this redshift.

2.2. DOES THE ABSORPTION ORIGINATE IN THE LENSING GALAXY?

There are two properties which independently suggest that the absorption lines originate in the lensing galaxy rather than being intrinsic to the radio source. (1) The two images of PKS1830-211 are situated on opposite sides of the center of the lensing galaxy (Nair et ai. 1993). If both core images are covered by molecular gas we expect two absorption profiles separated by a few hundred km S-l due to the rotation of the galaxy. However, the absorption consists of a single component with a maximum width of ",3D km s-l. (2) The absorption profiles do not reach the zero level, as expected for an optically thick molecular cloud covering the entire background source. Since the H13CO+(2f-1) line is detected and since all the J=2f-1lines of HCN, HCO+ and HNC reach a similar depth, the absorption is nevertheless likely to be saturated, implying that only part of the background continuum is covered by molecular gas, but that this gas is optically thick. The fraction of the continuum source area that is obscured by optically thick molecular gas is 36%. This is the same as the flux contribution of the SW component (N air et al. 1993). From this we conclude that the molecular absorption is likely to take place in the lensing galaxy and that the SW core is obscured by molecular gas, while the NE component is not.

2.3. EXCITATION TEMPERATURES AND COLUMN DENSITIES

The saturated lines of HCN, HCO+ and HNC, give upper limits to the excitation temperature Tx and lower limits to the column densities. The unsaturated profiles directly give Tx and corresponding column densities, but since these lines are weaker, the uncertainties are larger. The excitation temperatures are listed in Table 1. For HNC, which appears to be only moderately saturated, Tx is only slightly higher than the expected temperature of the CMB of 5.16 K. The unsaturated CS, H13CO+ and N2 H+ lines give Tx less than 5.16 K, but given the errors associated with these values, they are still compatible with Tcmb = 5.16 K. Preliminary results obtained with the IRAM 30-m telescope indicate that Tx = 6 ± 1 K for N2 H+ and Tx < 6.3 K for H13CO+. From our molecular absorption line data we derive

A NEW MOLECULAR ABSORPTION LINE SYSTEM 231

a conservative upper limit to the temperature of the cosmic background radiation at z = 0.886 of Tcmb ~ 6.0 K.

Using Tx = 6.0 K, we derive the column densities presented in Table 1. We estimate the CO and H2 column densities in the absorbing gas at z = 0.886 to be at least 3 X 1018 cm -2 and 3 X 1022 cm -2, respectively.

3. Future prospects: Time-delay measurements

PKS1830-211 is known to be a highly variable source (Steppe et al. 1993). If this variation is intrinsic to the background source, the saturated molecular absorption lines can be used as a sensitive probe of the time-delay between the two images of the background core. However, microlensing by structures in the obscuring molecular gas may affect the observed continuum flux. The time scale for this is of the order of years, but can be much shorter if the continuum emanates from a region exhibiting superlumina.l motion. Nevertheless, the molecular absorption profiles remain the best way to easily and accurately monitor the time-delay in PKS1830-211. The expected delay between the SW and NE cores is unknown but is estimated to be in the range 3-7 weeks.

TABLE l. Molecular gas properties (errors are 30-)

fTvdV Tx logN [kms-I] [K] [cm-2]

J=2<-1 J=3<-2 J=4+-3

CS 27.0 ± 4.8 6.9 ± 3.3 4.0~~ ~ 14.682~g.g~; HCN ~ 82.3 ± 6.6 48.8 ± 5.4 :::: 7.6~~·~ > 14.499+0034 - -0.036 HCO+ ~ 91.7 ± 6.3 59.2 ± 6.0 < 8.4+14 _ -1.1 ~ 14.466~g g;~ HI3 CO+ 17.5±3.3 3.6 ± 2.4 4.0~~ ~ 13.746~g.g~~ HNC ~ 53.3 ± 5.7 21.5 ± 5.4 :::: 6.0~~ ~ > 14.405+0 044 - -0049 N2 H+ 18.7 ± 4.2 3.4 ± 3.0 4.1~;.~ 13.821~g·~~~

References

Djorgovski, S., Meylan, G., Klemola, A., Thompson, D.J., Weir, W.N., Swarup, G., Rao, A.P., Subrahmanyan, R. Smette, A. 1992, MNRAS 257, 240

Jauncey, D.L. et al. 1991, Nature 352, 132 McMahon, P.M., Moore, C., Hewitt, J.N., Rupen, M.P. CariIIi, C. 1993, BAAS 25, 1307 Nair, S., Narashima, D. Rao, A.P. 1993, ApJ 407, 46 Narayan, R. 1991, ApJ 378, L5 Steppe, H., Paubert, G., Sievers, A., Reuter, H.P., Greve, A., Liechti, S., Le Floch, B.,

Brunswig, W., Menendez, C. Sanchez, S. 1993, A&AS 102, 611 Subrahmanyan, R., Narasimha, D., Rao, A.P. Swarup, G. 1990, MNRAS 246,263


Wiklind, T. Combes, F. 1994, A&A 286, L9 Wiklind, T. Combes, F. 1995, A&A 299, 382

DEEP HST IMAGING OF A DAMPED LYMAN ALPHA ABSORBING GALAXY AT Z = 2.811

P. M0LLER

Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD21218, USA. On assignment from the Space Science Department of ESA

AND

S. J. WARREN

Astrophysics, Blackett Laboratory ICSTM, Prince Consort Rd, London, SW7 2BZ, UK

1. Introduction

The subject of this meeting "Cold gas at high redshift" in combination with the anniversary of the Westerbork Synthesis Radio Telescope, quite naturally has prompted that most of the work presented here has been concentrating on the prospects for finding this cold gas at radio wavelengths. However, we actually already know where the majority of the cold gas at high redshift is, namely in what in optical astronomy is know as the "Damped Lya absorbers", or DLAs.

Optical studies of DLAs in absorption only probe one single sight-line through an absorber. Whereas such a sight-line provides a quite accurate measure of the HI column density, it will give no hint as to what the size, or the total mass, of the absorbing cloud is. To learn when and how galaxies were born, those two numbers are of course of central interest, and it has therefore been exciting here to learn that the next generation of HI radio surveys will be able to probe total gas masses in regimes where they may be able to detect, or seriously constrain the masses of, the DLAs.

lBased on observations obtained with the NASA/ESA Hubble Space Telescope

233


234 P. M0LLER AND S. J. WARREN

2. Damped Lya absorbers

DLAs form the high column density end of the distribution of Lya absorbers, N(RI)::: 1020cm-2 • The shape of the column density distribution is such that virtually all the neutral gas in the Universe is contained in the damped systems, but, as discussed by Fall and Pei (1995), we may only be seeing some fraction of those systems if there is any significant amount of dust in them, because dust would tend to bias surveys against finding background quasars with strong DLAs.

Thus, the total mass in neutral gas in the early universe may have been underestimated. This effect is notably important at intermediate redshifts whereas the estimated mass in neutral gas at z = 3, for current models of chemical evolution, not will be seriously affected by the corrections for dust. This mass is roughly equal to the mass in stars in spiral galaxy disks today, and the standard picture has been that the damped systems at z = 3 are the fully formed gaseous disks of spiral galaxies, that turn into the stellar disks of today. The disks would have been a factor> 2 larger in the past, in order to explain the large number of absorbers found. Recent work on the detailed velocity structure of metal line complexes associated with DLAs (Wolfe, 1995), points out that those systems often exhibit a profile asymmetry consistent with the rotating disk model.

A competing scenario is that of interpreting DLAs as a numerous class of gas-rich dwarfs. Recent simulations as well as semi-analytical modelling (Navarro et al., 1995; Kauffmann, 1995) seem to lend support to this picture, predicting that proto galaxies are in fragments, and that these small and compact gas rich objects grow to become todays galaxies by merging. In such a scenario the profile asymmetry would have to be interpreted as high velocity outflow following the initial star formation event.

3. The DLA in PKS 0528-250 as a test case

To test the two competing interpretations of high red shift DLAs: i) fully formed spiral disks or ii) galaxy sub-units, we decided to try and image the DLA at z = 2.81 in the Zem = 2.77 quasar PKS 0528-250. This particular system was selected for this study because of the similarity in redshift between the DLA and the underlying quasar. On the assumption that the DLA hence is very close to the quasar, the UV flux from the quasar could possibly help us identify the absorber by a "Lya silver lining effect" (see Fig. 1 in M¢ller and Warren 1993a). If this would be the case we should easily be able to measure the size, and determine the morphology, of the absorber since the quasar flux only would make the optically thin circumference of the DLA shine in Lya, thus making the outline of the absorber visible. In the perfect case we should see the entire outline of the absorber,

HST IMAGING OF DAMPED LYMAN ALPHA GALAXY 235

but several effects (e.g. non-symmetric gas density profile of absorber or non-symmetric quasar UV emission) could cause only part of the outline to be visible.

Ground based imaging (Mj1jller and Warren 1993b, 1995) revealed three Lya emitting objects (named SI, S2, and S3), at projected distances of 1.2, 11 and 21 arcsec (corresponding to 4.5h-1 , 41h-1 , and 78h-1 kpc at z = 2.81 for qo = 0.5) from the quasar line of sight. Spectroscopic confirmation of the Lya emission, and resolved line profiles, have been obtained (Warren and Mj1jller 1995a). The images did not, however, show the predicted outline of the absorber. Due to the insufficient resolution of the images it was not clear ifthe nearest object (SI) was in fact part ofthe Lya silver lining of a small object, or if it was simply Lya emission due to star formation inside S1. To get a final answer to this question we requested HST observations of the field, and those observations were obtained in 1995.

4. The HST images

The field was observed in three bands (F467M, 18 orbits; F450W 6 orbits; F814W, 6 orbits) for at total of 30 orbits. All three sources are detected in all three bands. The sources are hence not Lya-only objects, they do have an underlying continuum in both Band 1.

Part of the HST field is reproduced in Fig 1. The three sources SI, S2, and S3 are marked. Note that the brightness of the quasar makes SI invisible unless a model for the psf is subtracted. To give an impression of how severe this problem is, we have in Fig 1 only subtracted the part of the psf around S1. Fig 1 was smoothed to bring out the faint galaxies better. The field shown is 39" by 39". PKS 0528-250 is the bright object in the upper right corner, there is a bright star just outside the field in the lower left corner, two relatively bright foreground galaxies can be seen in the other corners, but notably there is a large number of very faint galaxies.

In Fig 2 we show the enlarged regions around the three confirmed high redshift objects, and also around a non-saturated star for comparison. The pixels in Fig 2 are 0.1" on a side, and the data are unsmoothed. Note how small and compact S2 is, not much larger than the star. S3 is clearly resolved into two components (S3a and S3b) separated by 0.3" (1.1h- 1

kpc), and both of them as compact as S2. After correction for the stellar psf, the typical size of the three objects (S2, S3a, and S3b) is about 0.5h-1

kpc FWHM. Sl is clearly visible in the image both before and after the subtraction

of the psf model. However, due to the large uncertainty in the modelling of the psf, it is difficult to say anything else at this point about S1. It appears to be just as compact as S2, and there seems to be no evidence for


Figure 1. HST image of the field around PKS 0528-250. The image is 39" on a side, and the three known high redshift Lya emitters are marked S1, S2, and S3. Note the large number of other faint, compact galaxies.

any extended morphology. A better model for the psf-subtraction would be needed for a clearer answer, but it is not trivial to obtain.

5. Comparison with simulations

The striking similarity between the alignment of proto-galaxy substructures in recent simulations (Evrard et ai., 1994; Navarro et al., 1995), and the alignment observed in this and two other fields containing Lya emitters, has been discussed in detail elsewhere (Warren and M0ller 1995b). We shall here limit the discussion to simply point out another striking sim-

HST IMAGING OF DAMPED LYMAN ALPHA GALAXY 237

Figure 2. H8T images of the LyCl' emitters at z ::::: 2.81. Each pixel is 0.1" on a side. Note that 83 is resolved into two sources separated by 0.3", and that all the sources are very compact. A stellar psf taken from the same frame is shown for comparison.

ilarity. Namely the fact that proto-galaxy substructures in recent models are predicted to be very small and compact (see e.g. Kauffman 1995). The small sizes we find for these high redshift objects are fully consistent with those predictions.

6. Conclusions

In conclusion we would like to put forward just two statements.

The first is that even though we cannot from this one case generalise to all DLAs, the spectroscopic and high resolution imaging data we have obtained for this field so far all point towards that this DLA is a galaxy in the process of forming by merging of small compact sub-


units. The large number of similar faint compact objects found in deep counts with the Keck (Smail et al., 1995) suggests that this could be a fairly common type of object at high redshifts. In conclusion, we see no evidence that DLAs are fully formed disk galaxies. The mass of the HI in the absorber (S1) is estimated to be 1 x 1Q9h-2M8 (Warren and M~ller 1995a). Our second tentative conclusion shall therefore be that if our interpretation of these data is correct, and if this turns out to be typical for high redshift galaxies, then it is probably bad news for radio astronomy since HI clouds of such a low mass are going to be extremely hard to detect.

References

Evrard, A. E., Summers, F. J. and Davis, M. (1994) ApJ 422, 11 Fall, M. and Pei, Y. C. (1995) In QSO Absorption Lines, Meylan, G., (ed.), Springer,

p23 Kauffmann, G. (1995) this volume M9111er, P. and Warren, S. J. (1993a) In Observational Cosmology, Chincarini, G., Iovino,

A., Maccacaro, T., Maccagni, D. (eds.), A. S. P. Conference series, 51, 598 M9111er, P. and Warren, S. J. (1993b) A&.A 270, 43 M9111er, P. and Warren, S. J. (1995) In Galaxies in the Young Universe eds. Hippelein,

Meisenheimer &. Roser, p88 Navarro, J. F., Frenk, C. S. and White, S. D. M. (1995) MNRAS 275,56 Smail, I., Hogg, D. W., Van, L. and Cohen, J. G. (1995) ApJ-L, 449, L105 Warren, S. J. and M9111er, P. (1995a) A&.A in press Warren, S. J. and M9111er, P. (1995b) In New Light on Galaxy Evolution, IAU Symposium

171, Heidelberg June 1995, in press Wolfe, A. M. (1995) In QSO Absorption Lines, Meylan, G., (ed.), Springer, p13

ASSOCIATED X-RAY ABSORPTION IN HIGH REDSHIFT QUASARS

MARTIN ELVIS AND FABRIZIO FIORE Harvard-Smithsonian Center for Astrophysics 60 Garden St., Cambridge MA 02138 USA

1. Introduction

When ROSAT was launched in June 1990 it provided X-ray astronomy with the capability to study fainter sources that had hitherto been accessible. This opened up the range of AGN quasars that could be reached to virtually the whole span of their properties. So, instead of revisiting the same objects made famous by earlier missions, we adopted a strategy of exploring this new parameter space (Elvis 1994). The most obvious of the parameters to push was redshift, since almost no X-ray spectra of quasars at z > 0.5 were then available. Redshift three quasars were the most distant for which we could obtain strong ROSAT detections.

The ROSAT PSPC spectra of z = 3 quasars immediately surprised us by often showing strong low energy cut-offs (Elvis et al., 1994), far larger than those expected due to absorption by our galaxy. We now believe that these cut-offs are associated with the quasars. Assuming photoelectric absorption by cold gas 1 to be the cause this proximity allows us to derive some properties of the absorber.

2. Low Energy Cut-offs in High z Quasars

The signature of these low energy cut-offs is not subtle. Figure 1a shows a PSPC pulse height spectrum of a high z quasar with only absorption due to our galaxy. Figures 1 b,c show two other quasars in which the low energy counts are clearly missing.

If we fit this cut-off with absorption at the quasar we find we need column densities of order 1022 atoms em -2, however we cannot determine the absorber red shift from the PSPC data. If the absorber is closer to us

1 "Cold gas" is a relative term. For an X-ray astronomer it means, roughly < 1 06 K, so that metal atoms are not fully stripped.

239


240

101 ...... i ~ 0. 101

11'1 .. c 6 10' U

III 3

MARTIN ELVIS AND FABRIZIO FIORE

! O~--~~~~~~~ .. ~~~~~--~~~~~~ Cii .3

~1~~'~~~~~~.~~I~~'--"~~1~~~I~~~'~~~~~~.~ EnerlY (kev) EnerlY (keV) EnerlY (kev)

Figure 1. ROSAT PSPC pulse height spectra of high z quasars with (a) Galactic absorption only; (b-c) implied excess absorption (plus Galactic absorption).

then, because of the red shifting, it has to absorb lower energy photons and so needs less column density, ~1021 atoms cm-2 at z = O.

This assumes solar abundances and neutral material. Neither assumption can be tested by the PSPC data alone. Most of the absorption at 1 keY X-ray energies is due to metals, primarily 0, Fe, Ne, Mg, Si and S (Morrison & McCammon 1983). Reducing the abundances to primordial H+He alone, increases the implied hydrogen column by a factor of four (Elvis et at., 1994). Increasing the abundances decreases the hydrogen column decreases the required hydrogen column proportionately. (Here the "hydrogen column density" refers to the total number of hydrogen atoms, whether neutral or ionized.)

To make any further progress we must pin down the redshift of the absorber. This we can do statistically.

3. Associated Absorption & Evolution

The entire ROSAT pointed data archive is now publicly available. Searching through the more than 3000 pointings and over 60,000 X-ray sources has been made much easier by the compilation of two source catalogs, ROSATSRC from MPE, and WGACAT from GSFC (White, Giommi & Angelini 1995). Using WGACAT we have been able to extract X-ray "colors", i.e. a "hardness ratio", for almost 1000 previously cataloged quasars. Among these 81 are at z > 2, and 15 at z > 3. Figures 2a,b show the color derived at the lowest energies for these quasars as a function of redshift, z. This color is sensitive to absorption. The "color" is expressed as an effective

-I

X-RAY ABSORPTION

2 3 - .. SNR>5; ... <-~. lotNH<21. No poin .... 2.'

I

! r e-o ..

...... 1.0 ... -..

2 nctehllt. I

SNR>S .... <-~ . IDtNH<21. No poiau-1S&

241

Figure 2. Soft PSPC X-ray "color" for (a) radio-quiet quasars, (b) radio-loud quasars.

0.1- 0.8 keY slope, assuming Galactic absorption, so negative slopes imply falling spectra to low energies, suggestive of excess absorption.

Comparing the two plots it is immediately obvious that they are different. The radio-quiet quasars show far less scatter at all redshifts. While this will be of great interest in other contexts the key point for the present is the number of quasars with soft X-ray colors below the horizontal lines. These lines are the expected colors for quasars with the measured Ginga low z slopes (2-10 keY, rest frame). The lines are slightly different for radio-loud and radio-quiet because they have systematically different slopes (Williams et al., 1992.) Absorbed quasars would populate this region. Two of the quasars with measured low-energy cut-offs (see Figs. 1b,c) do in fact lie in this region (PKS2126-158, PKS0438-436). Also on the plot is an Einstein IPC point for the z = 4.3 quasar GB1508+5714 (Mathur & Elvis 1995).

3.1. ASSOCIATED CUT-OFFSjABSORPTION

It is apparent from Fig. 2 that only the radio-loud population has a population of quasars below the line. The difference is significant at the 99.5% level for quasars with z > 2. Immediately this tells us the crucial point:

The low energy cut-offs are associated with the quasars themselves. They are not simply lying randomly along the line of sight, otherwise they would affect radio-loud and radio-quiet equally. With WGACAT we are at the limit of detecting this effect. For z > 2.5 this difference is 89%, so a few more objects at z > 2.5 would strengthen this results greatly. Stew-

242

~ 0.8 ,Q

'" o 1l 0.6 <0

g 0.4 ~ u <0 .= 0.2

MARTIN ELVIS AND FABRIZIO FIORE

2 3 redshifl. z

4 5

~ 0.8 ,Q

'" o 1l 0.6 <0

c ..2 0.4 ... u <0 .:: 0.2

10 20 30 40 path length. X

50

Figure 3. Fraction of "absorbed" radio-loud quasars vs. (a) redshift, (b) path length.

art et ai. (1994) find no cut-offs in a sample of radio-quiet (X-ray selected) quasars, which strengthens our result. There is one potential way out of this conclusion: while radio-selected quasars can be discovered even when reddened, absorbed X-ray- or optically-selected quasars will be biased against discovery. So dusty intervening absorbers, might produce a selection effect. However, the X-ray cut-off quasars have normal strong UV continua (Kuhn et al., 1996) making this unlikely.

3.2. EVOLUTION

Although there are cut-off radio-loud quasars at all redshifts, there appears to be a greater fraction of them "below the line" at higher redshifts. We can quantify this by making a plot of the fraction of quasars "below the line" in Fig. 2 as a function of red shift (Fig. 3a). The Poisson derived error bars are large, but there is a strong hint that the cut-off fraction of quasars rises rapidly from z '" 2. An alternative binning, by absorption path length, X (Elvis et ai., 1994, Bahcall & Peebles 1969), is more convincing (Fig. 3b). This indicates that the cut-offs are more common in the past than now, i.e. they are an evolutionary effect. Redshift z'" 2 is special, as for quasar population evolution (Boyle et ai., 1987), which may not be a coincidence.

4. Physical Conditions in Associated Absorbers

Knowing that the cut-offs are at the redshift of the quasar allows us to begin interpreting the results. If we assume absorption is the cause then we can constrain the physical conditions of the absorbers. A bsorption by a neutral hydrogen column would produce strong absorption in the opticaljUV spectrum of these quasars.

X-RAY ABSORPTION 243

For Milky Way dust-to-gas ratios NH = 1022 atoms cm-2 produces Av = 4.2 (Jenkins & Savage 1974, Seaton 1979). The strong UV bumps in these quasars limits Av ~ 1 (Kuhn et at., 1996), so the absorber must have a low dust-to-gas ratio, as in the BALQSO PHL5200 (Mathur, Elvis & Singh, 1995), but unlike the red quasar 3C 212 (Elvis et at., 1994, Mathur 1994).

At 1022 atoms cm-2 damped Lyman Q absorption should be found, which is not seen; nor is Lyman limit absorption near the quasar redshift seen, so the neutral hydrogen column must be below 1017.5 atoms cm- 2 (Kuhn et at., 1996; Jauncey et al., 1978). The fractional ionization of H must thus be < 10-5 .5 . Higher ionization potential ions should be more sensitive. The CIV .\1550 lines are promising, and Jauncey et at., (1978) and Wilkes et at., (1986) both showed evidence for weak CIV absorption in PKS2126-158. This is not seen in high resolution data (Young et al. 1979) so the line must be broad, if it is real. Assuming it to be real gives an ionization parameter, U = Q!(41rr2nHc) rv 1.5 - 3.5, as in the ionized absorber in the Seyfert galaxy NGC5548 (Mathur, Elvis & Wilkes 1995).

Any absorber is likely to be ionized around these quasars: they are the most luminous objects in the universe and have their luminosity concentrated in the ionizing UV and soft X-ray region. Mathur & Elvis (1995) note that GB1508+5714 will ionize gas out to a radius of 3n5 kpc from the nucleus (where n5 is the gas density in units of 105 cm-3 . Thus the absorber need not be part of the active nucleus itself, but may be in the host "galaxy", or even on a cluster of galaxies, Mpc, scale, for present day cluster core densities of 10-2 cm-3 (Sarazin 1986). That these absorbers are seen only in radio-loud quasars, then requires that radio-loud and radio-quiet quasars have quite different environments at early epochs.

A bundances can be at least 10 times solar in quasar broad emission line regions at high z (Hamann & Ferland 1994). On the other hand the early galaxies that these quasars lie in should be relatively metal poor, down to the 1!10th solar values in damped Lyman a systems. X-ray grating spectroscopy will be able to measure individual K-edges and so derive the absorber abundances.

5. Conclusions

Low energy X-ray cut-offs are seen in radio-loud high red shift quasars. A large sample of quasars with X-ray colors shows: (1) cut-offs occur only in radio-loud quasars, implying that the cut-offs are associated with the quasars; (2) that the fraction of quasars with cut-offs increases rapidly above z = 2, implying that evolution with cosmic time is important; (3) the strong quasar UV continuum shows that the absorbers are virtually dust-free. These conclusions are independent of the physical cause of the

244 MARTIN ELVIS AND FABRIZIO FIORE

cut-offs. Assuming the cut-offs are due to photoelectric absorption their location

near the quasar tells us: (1) NH f'V 1022 cm-2 for solar abundance; (2) the absence of Lyman limit absorption shows the ionization parameter, U>2, so the absorber must lie 3n5 kpc from the nucleus, which allows sites for the absorber range from nuclear outflows to the surrounding cluster; (3) radio-loud and radio-quiet quasars lie in different environments at z = 3.

These high z quasar cut-offs seem likely to be a fruitful new field of investigation at many wavelengths, but especially for the large new X-ray satellites due for launch in the next 5 years.

We thank our fellow team members involved in this work - Jill Bechtold, Olga Kuhn, Tom Aldcroft, Smita Mathur, Jonathan McDowell, Aneta Siemiginowska, and Belinda Wilkes. This work was supported in part by NASA grant NAG5-3066 (ADP) and NASA contract NAS8-39073 (ASe).

References

Bahcall J. & Peebles J. (1969), Astrophysical Journal, 156, L7 Boyle B., Fong R., & Shanks T., (1987), MNRAS, 227, 717 Elvis M., Fiore F., Mathur S., & Bechtold J., (1994), Astrophysical Journal, 425, 103 Elvis M., (1994), X-rays and High Redshift Quasars, New Horizon of X-ray Astronomy,

[Tokyo: Universal Academic Press]' p.323 Elvis M., et al., (1994), Astrophysical Journal Supplement, 95, 1 Hamann F. & Ferland G., (1992), Astrophysical Journal, 391, L57 Jauncey D.L., Wright A.E., Peterson B.A., & Condon J.J., (1978), Astrophysical Journal,

223, 1 Jenkins E.B., & Savage B.D., (1974), Astrophysical Journal, 187, 243 Kuhn O. et al., (1996), in preparation. Mathur S. & Elvis M. (1995) GBI508+5714, Astronomical Journal, 110, 1551 Mathur S., Elvis M. & Wilkes B.J., (1995), Astrophysical Journal, 452, 230 Mathur S., Elvis M. & Singh K.P., (1995), Astrophysical Journal Letters, 455, L9 Morrison R. & McCammon D., (1983), Astrophysical Journal, 270, 119 Sarazin, C.S. 1986, Rev. Mod. Phys., 58, 1 Seaton M.J., (1979), MNRAS, 187, 785 Stewart G.C., Georgantopoulos I., Boyle B., Shanks T., & Griffiths R., (1994) The Spec

tral Evolution of High Redshift QSOs Observed with ROSAT, New Horizon of X-ray Astronomy, [Tokyo: Universal Academic Press], p. 331.

White N.E., Giommi P. & Angelini 1. (1995), http://heasarc.gsfc.nasa.gov/ ~white/wgacat/wgacat.html

Wilkes B.J. et al., (1983), Proc. Astron. Soc. Australia, 5, :t Williams O.R. et al., (1992), Astrophysical Journal, 389, 157 Young P. J., Sargent W.L.W., Boksenberg A., Carswell R..F., & Whelan J.A.J., (1979),

Astrophysical Journal, 229, 891

OPACITY OF SINGLY IONIZED HELIUM FROM VERY TENUOUS INTERGALACTIC ABSORBING GAS

W. ZHENG Center for Astrophysical Sciences Johns Hopkins University Baltimore, MD 21218, USA

The He II Gunn-Peterson test has been successfully carried out to detect the intergalactic medium (IGM). Jakobsen et at. (1994) reported a flux cutoff shortward of redshifted He II Lya emission in the HST prism spectrum of the quasar 0302-003 (z = 3.286), with the optical depth T = 3.2~ro. The spectrum of quasar HS1700+64 (z = 2.743) recently obtained with the Hopkins Ultraviolet Telescope (HUT) reveals a flux depression due to He II absorption (Davidsen, Kriss, & Zheng 1996), with an average optical depth of T = 1.0±0.07. The findings confirm the Gunn-Peterson absorption produced by primordial helium in the IGM.

The measurements of the Gunn-Peterson optical depth rely on accurate subtraction of the forest-line contribution as the current instruments are not able to resolve individual UV absorption lines. It is therefore important to distinguish the absorption by the diffuse IGM from that of the discrete intergalactic clouds that produce known Lya forest lines. The latter blankets the spectral region shortward of 304{ 1 + z) A as does the He I I Gunn-Peterson trough. Jakobsen et al. (1994) and Madau & Meiksin (1994) estimated the contribution due to Lya forest line blanketing and concluded that it will not account for the observed He II opacity unless there is a significant number of clouds with neutral hydrogen column density (column density hereafter) N < 1013 cm-2 • Giroux, Fardal & Shull (1995) derive generous values of the He II opacity that all the observed opacity can easily be attributed to forest lines.

Simulations are made using the empirical formulas governing the distribution of forest lines. The number of Lya forest absorption features, n, in quasar spectra may be described by:

245


(1)

246 W. ZHENG

(Press & Rybicki 1993). The normalization factor A affects the forest opacity appreciably, but its value varies widely in the literature. We adopt (3 = 1.5, '"f = 2.4 and A = 2.0 X 107 from Miralda-Escude & Ostriker (1990) because with these parameters we are able to produce the line number that matches recent published data (Hu et al. 1995). The average forest opacity is calculated in the 1050-1170 A range (restframe). This wavelength range, free of confusion from the proximity effect and Ly(3 lines, is used to calculate the average absorption decrement D A.

Simulated Lya absorption lines are generated according to their columndensity distribution. The Doppler parameter b of lines is assumed to follow a f-function distribution pCb) <X bbo/b*-1 exp( -b/b*) where bo = 38 km s-1 and b* = 14 km s-1 (Press & Rybicki 1993). The centroid wavelengths of each line are randomly generated according to the z-dependence of Equation (1). The Voigt profiles of these lines are then generated in the quasar restframe, with a pixel size of 0.01 A. The effect of line blending is naturally taken into consideration under such a treatment. Simulations start from lines of high column density (rv 1017 cm -2) and stop at a given low cutoff column density.

The corresponding He II spectrum is constructed by changing the column density of each simulated line from Ni to TfNi, and the Doppler parameter from bi to (k The helium to hydrogen velocity ratio ( is between 0.5 and 1.0, depending on the nature of line broadening. A recent estimate (Cowie et al. 1995) suggests ( '" 0.8. We will use this value in most of our calculations, as the results can be scaled to suit other ( values. The ratio of He+ to H population Tf affects the He II opacity appreciably. Most models have assumed Tf ~ 100 if quasars are the dominant source of the metagalactic radiation.

The optical depth in the vicinity pixels of each simulated line is calculated using the Voigt profile. The average decrement DAis calculated between 262.5 and 292.5 A for the He II spectrum, and 1050-1170 A for H I. The average forest optical depth is expressed as T = - In( 1 - D A). Figure 1 displays the average forest optical depth of H I and He II vs. the lower cutoff column density. In the middle panels it can be seen that, at N < 1013 cm -2, the optical depths approach values of 0.22 and 0.33, for z = 2.7 and 3.3, respectively. The simulations show that, for the parameters adopted, rv 80% of the observed HI Lya forest opacity is produced by the strong lines of N > 2 X 1013 cm -2.

As shown in Figure 1, the total H I opacity increases toward lower cutoff column densities, but it becomes saturated at N ::::: 1013 cm-2 . The He II

opacity, on the other hand, continues to increase significantly toward cutoff column densities of N ::::: 1011 cm- 2 • Note that strong lines produce nearly the same amount of absorption for He II and H I, but they contribute only

HELIUM OPACITY 247

,.1.6 , • 1.2

1.6

~ •..........................•.. , ..............................................•.................

0.4

0.9 .............................. . .........................•..

0.2 r-------.. _ ... _ .. _............... . .................... . 0.1

2

Cutoff Column Density logNHO (em-I)

Figure 1. Average optical depth of absorption between 1050 and 1170 A in restframe accumulated by forest lines of column density greater than the cutoff value. Their ratio is displayed in the lower panels. In each panel, the solid and dotted curves are for z = 2.74 and 3.29, respectively. The values at the current observation limit, log N = 12.3 are marked, representing the maximum amounts of opacity the observed forest lines can account for. The He+ to H population ratio '1/ is assumed to be 100, and velocity ratio ~ = 0.8. The dashed line in the lower left panel represents the hypothetical value derived from Equation (2). The curves on the right panels are produced with an assumption of f3 = 1.2 for N < 2 X 1012 cm-2 •

tV 30% of the total He II opacity for 'fJ = 100. Weak lines contribute f"V 40% of the total He II opacity, compared with only tV 5% for H I. This is because that, at low column densities, the H I absorption is proportional to the column density and decreases significantly, while the He II absorption is still in the flat part of the curve of growth. The opacity ratio p therefore becomes higher when more weak forest lines are included.

It should be noted that extension of a power law of f3 = 1.5 to column densities below 1013 cm-2 is speculative. Because of line blending many weak lines are not observable. Therefore an assumed incompleteness correction of up to a factor of 4 is applied to the observed line counts for N < 1013 cm-2 (Hu et al. 1995). There is, however, growing evidence that the number of these weak lines is smaller than a power-law index of f3 = 1.5 requires. Theoretical models (Cen et al. 1994) also suggest that the distribution of forest lines flattens at low column densities. In such cases, the total He II opacity due to forest lines would be smaller. As a comparison,

248 W. ZHENG

we also simulate the opacity for f3 = 1.2 for N < 1013 cm -2. The total forest He II opacity thus produced is considerably smaller, suggesting an even larger contribution from the diffuse matter. In general, the opacity ratio p ~ 3 when lines of N > 2 X 1012 cm-2 are taken into account.

At very low cutoff column density (N ~ 1011 cm-2 ), the ratio offorest opacity approaches the theoretical limit:

(2)

(Miralda-Escude 1993). For 'fJ = 100 and ~ = 1.0, this would yield p = 5. Such a high p value is needed to account for the observed He II opacity or its low limit. Figure 1 indicates that applying Equation (2) at N ~ 2 X 1012 cm-2 leads to a considerable overestimation of the He II forest opacity.

Previous calculations do not take into account the line blending, hence overestimating the He II forest opacity by IV 20 - 60%. When weak lines are blended with strong lines, they may not contribute additional opacity. This leads to a significant deviation at higher redshift. Our calculations show that most He II opacity is produced by tenuous medium of column density N < 1013 cm -2. Assuming a power-law index of f3 = 1.5, He+ to H population ratio 'fJ = 100 and velocity ratio ~ = 0.8, the forest He II opacity is f'V 3 X the H I opacity.

The high-quality Keck spectra should provide the best estimate of the forest opacity. Using all the resolved absorption lines and converting them into He II profiles, calculations show that the observed forest lines will not account for more than 60% of the observed He II opacity. Therefore, a significant contribution must arise in the diffuse IGM.

References

Cen, R., Miralda-Escude, J., Ostriker, J. P., & Rauch, M. 1994, Ap. J., 437, L9 Cowie, L. L., Songaila, A., Kim, T.-S., & Hu, E. M. 1995, A. J., 109,1522 Davidsen, A. F., Kriss, G. A., & Zheng, W. 1996, preprint Giroux, M. L., Fardal, A. A., & Shull, J. M. 1995, Ap. J., 451, 477 Hu, E. M., Kim, T.-S., Cowie, L. L., Songaila, A., & Rauch, M. 1995, A. J., 110, 1526 Jakobsen, P., Boksenberg, A., Deharveng, J. M., Greenfield, P., Jedrzejewski, R., &

Paresce, F. 1994, Nature, 370, 35 Madau, P., & Meiksin, A. 1994, Ap. J., 433, L53 Miralda-Escude, J. 1993, M.N.R.A.8., 262, 273 Miralda-Escude, J., & Ostriker, J. P. 1990, Ap. J., 350,1 Press, W. H., & Rybicki, G. B. 1993, Ap. J., 418, 585

HEAVY ELEMENTS IN THE LYMAN ALPHA FOREST: ABUNDANCES AND CLUSTERING AT Z=3

D.S. WOMBLE, W.L.W. SARGENT, R.S. LYONS California Institute of Technology Astronomy 105-24, Pasadena, CA 91125 USA

Abstract. For the purpose of studying the properties of heavy elements associated with the Lyman a forest, we observed the gravitational lens Q1422+2309. We used the HIRES instrument on the W.M. Keck telescope to obtain a high resolution, very high signal-to-noise spectrum of this z = 3.63 quasar; the spectrum covers wavelengths from below Lyman f3 up to the C IV emission line. Consistent with previous estimates, we find that a moderate fraction of the Lya forest clouds have been enriched with heavy elements at a level significantly below solar abundance. However, unlike the fairly uniform distribution of Lya forest lines, we show that the C IV absorption lines are clustered on large velocity scales.

1. Background

The spectra of high redshift quasars show remarkable complexity in H I absorption at wavelengths shortward of Lya emission. The nature and composition of these ubiquitous Lya "forest" lines remains as a prominent topic in cosmology. For more than a decade, the consensus has prevailed that high redshift Lya forest lines are due to an intergalactic population (cf. Sargent et al. 1980). Motivated to ask whether these Lya forest clouds have primordial or enriched heavy element abundances, we report on a very sensitive study of C IV absorption corresponding to the Lya forest in Q1422+2309.

Several measurements and upper limits on the metallicity of the Lya forest have been reported in the literature (Norris et at. 1983, Meyer & York 1987, Lu 1991, Tytler & Fan 1994). Using a variety of techniques, these results have provided some evidence for heavy element enrichment in the forest. However, it is not clear what fraction (if any) of the population

249


250 D.S. WOMBLE, W.L.W. SARGENT, R.S. LYONS

show primordial abundances and if the enriched clouds originate from a physically distinct class of object (such as extended galaxy halos). The exceptional high resolution facilities on the Keck 10 m telescope provide a timely opportunity to explore the properties of metals associated with the Lya forest. We have obtained a spectrum, of unprecedented quality, on Q1422+2309 for this purpose.

The quasar 1422+2309 is a bright, high redshift (V=16.5, Ze = 3.63) gravitational lens (Patnaik et al. 1992). This source is composed of four unresolved components located within an angular diameter of 1.3 arcseconds. Hammer et al. (1995) report a redshift for the lens of Z = 0.647 however we have not confirmed this measurement - owing to the absence of any associated intervening absorption lines in the quasar spectrum. The density of the Lya forest in this high-z quasar makes it difficult to identify any z < 1 metal line systems which might signify the redshift for the lensing mass.

2. Observations

We observed Q1422+2309 using the HIRES spectrograph on the Keck 10 m telescope during June 1994 and Spring 1995. The data were taken through a 0.86 by 7 to 14 arcsec slit resulting in a resolution of 6.6 km s-1 FWHM. Because the position angle of the slit rotates during an exposure (since Keck is Alt-Az), the spectrum contains significant amounts of light from the three brightest images of this gravitational lens.

We have achieved a typical signal to noise ratio of 150:1 per resolution element from a total exposure time of 25000 seconds. By using several different instrumental setups, the spectrum has complete wavelength coverage over AA3650-7150 (over 94000 pixels) with no inter-order gaps. At the redshift of the quasar, this spectrum extends from below 912A to just short of the broad C IV emission line peak.

Excluding a few isolated regions, we reach a limiting sensitivity under 1012 cm -2 in the column densities of both C IV (unresolved) and Lya (for b '" 20 km S-1 ) over most of the spectrum.

3. Results

In principle, the extended (lensed) nature of the background source may adversely affect intervening absorption line statistics because of cloud covering factors and inhomogeneous density distributions. The maximum angular separation of lensed images, at a lens redshift of 0.647, corresponds to an impact parameter of 5 h-1 kpc; we are using Ho = 100 h km s-1 Mpc-1, qo = 0.5 herein. Fortunately, this length is much smaller than recent estimates for the spatial extent of Lya clouds (Dinshaw et al. 1995) and the inferred diameters of galaxies which produce typical C IV absorption (Sar-

HEAVY ELEMENTS IN THE LYMAN ALPHA FOREST 251

gent et al. 1988). If, for example, a Lya forest cloud did not completely cover the background source, we would expect to see a highly non-Voigt line profile due to a zero-point offset in the residual flux. No such line profiles have been observed. The question of density fluctuations relative to the individual images has not yet been addressed and it may remain as an insoluble problem.

Concentrating on the portion of the spectrum between Ly f3 and Lya emission, 2.95 ~ Za ~ 3.61, we find that a significant fraction of the strong Lya forest lines have detectable C IV absorption. Using an optical depth criterion to select H I lines with T 2: 5, we count 66 Lya lines over this redshift path. For a Doppler parameter, b = 30kms-1, this limit corresponds to a column density, N(H I) 2: 2 X 1014 cm -2 • At the corresponding positions of C IV absorption, we detect 26 metal line systems; for these purposes, all C IV components spread over 6..v ~ 500 km s-l are counted as a single red shift system. We also detect an additional 4 C IV systems at redshifts corresponding to Lya lines which have apparent optical depths below our selection threshold. These latter systems, and many components of C IV absorption appear to have different velocity distributions than their associated Lya lines. Allowing that the H I and C IV ions may occupy physically distinct regions in the absorbing clouds, we find that 40-45% of the Lya forest lines with log N(H I) 2: 14.30 have been enriched with heavy elements. This fraction is roughly consistent with the values of 60% (Tytler et al. 1995) and 50% (Cowie et al. 1995) for log N(H I) 2: 14.5; note that we see 30/66 C IV systems compared with 13/22 and 15/31, respectively.

For the majority of these H I systems, the Lya lines are too saturated to obtain reliable column densities. In the six weakest systems, we obtained accurate measures of the C IV and H I column densities using Voigt profile fitting of the lines. In these systems, the median difference between the redshifts ofthe C IV and Lya components corresponds to a velocity offset of 35kms-1 • The mean relative abundance, (logN(CIV)/N(HI)) = -2.65± 0.22.

In order to convert to an absolute abundance, we must make an a priori assumption on the ionization state of the gas. Adopting C/C IV =10 and H/H 1= 10\ we obtain a typical abundance of [C/H]=-2.3 at (z) ~ 3.1; this value is highly dependent on the (unknown) ionizing conditions in the clouds. For comparison, Lu (1991) found [C/HJ'" -3.2 for relatively strong Lya lines whereas Tytler & Fan (1994) placed a limit of [C/H]~ -2.0 at lower H I column densities and Tytler et al. (1995) claim [C/HJ> -2.5 for log N(H I) 2: 14.5. Given uncertainties in the ionization corrections, our value is entirely consistent with previous abundance estimates.

In Fig. 1, the lower panel shows the portion of normalized quasar spectrum for Lya lines with 2.95 ~ Za ~ 3.62. On the same scale, the upper

252 D.S. WOMBLE, W.L.W. SARGENT, R.S. LYONS

C IV Syatems in Q1422+2309

II 3 3.1 3.2 3.3 3.4 3.5 3.8

REDSHIP"l'

3 :3.1 3.2 3.3 3.4 3.5 3.8

Lya Forest

4800 4900 5000

Figure 1. Lower panel shows normalized spectrum in Ly ll! forest with 2.95 ::; z" ::; 3.62; upper panel shows the distribution of C IV column densities as a function of redshift (same horozontal scale). Note the apparent clustering of C IV systems both on small and large-scales.

panel shows the distribution of C IV column densities as a function of redshift; these values were obtained by least-squares Voigt profile fitting with the minimum number of velocity components needed to get a satisfactory fit. It is visually apparent from this figure that the C IV systems are clustered. The tight groups of lines are clustered on a velocity scale less than 500 km s-1 ; this scale has already been well established by lower-resolution studies. It is also evident by eye, that the aggregate set is lumped on scales approximating several thousand km s-1 .

For this database of 146 C IV redshifts, we have evaluated the twopoint correlation function in comoving coordinates. Figure 2 (top) shows this correlation function for velocities less than 1000 km s-1, whereas the bottom panel extends out to red shift pair separations of 10000 km s-l. It can be seen that there is significant structure both below 500 km S-1 and in the vicinity of 4500kms-1 (at the Sa level). These structures correspond to comoving distances of::; 2.5h-1 Mpc and 25 h-1 Mpc, respectively. The structure seen on smaller velocities (Llv ::; 500 km s-1) is likely to stem from galaxy-galaxy correlations although some component is also certainly due to the motion of clouds within individual galaxy halos. On the much

HEAVY ELEMENTS IN THE LYMAN ALPHA FOREST 253

Comonnl Separation [h- I Mpc] 1~ 2 3

~ KIV System. In Q1422+2309 fO E 5 o u .. ...., 'i" 0 ..................................... . N

o 200 400 800 Velocity Separallon [km 8-']

5

800 1000

Comovlnl Separation [h- I Mpc] 150~~~-r~1~0,-~~~20~~-r~~30~-r~,-4TO~~,-~50

" ~ C IV Systems in Q1422+2309

<l 10

~ -

a; I:: 5 -

~ \ t1. 0 ... do - -~

.....

o 2000 4000 8000 8000 10' Velocity Separation [km 8-1]

Figure 2. Two-point correlation functions for 146 CIV red shifts shown in Fig. 1; pair separations denoted in both velocity and comoving distance for small and large scales.

larger scale, Sargent et al. 1988 saw some evidence for this structure in the distributions of redshifts in their lower-resolution C IV survey. The higher sensitivity of the present survey is probably responsible for the increased significance of such characteristics. It is interesting to note that the 25 h-1 Mpc scale length at z ~ 3 is comparable to the local cluster-cluster correlation length.

References

Cowie, L.L., Songaila, A., Kim, T.S., Hu, E.M. 1995, AJ, 109,1522 Dinshaw, N., Foltz, C.B., Impey, C.D., Weymann, R., Morris, S. 1995, Nature, 373, 223 Hammer, F., Rigaut, F., Angonin-Willaime, M.-C., Vanderriest, C. 1995, A&A, 298, 737 Lu, L. 1991, ApJ, 379, 99 Meyer, D.M., York, D.G. 1987, ApJ, 315, L5 Norris, J., Hartwick, F.D.A., Peterson, B.A. 1983, ApJ, 273, 450 Patnaik, A.R., Browne, I.W.A., Walsh, D., Chaffee, F., Foltz, C. 1992, MNRAS, 259, PI Sargent, W.L.W., Boksenberg, A., Steidel, C.C. 1988, ApJS, 68, 539 Sargent, W.L.W., Young, P.J., Boksenberg, A., Tytler, D. 1980, ApJS, 42, 41 Tytler, D., Fan, X.-M. 1994, ApJ, 424, L87 Tytler, D., Fan, X.-M., Burles, S., Cottrell, L., Davis, C., Kirkman, D., Zuo, L. 1995, in

"QSO Absorption Lines", ed. G. Meylan (Berlin: Springer-Verlag), p. 289

ABSORPTION LINES FROM COLD GAS IN EXTRAGALACTIC SUPERBUBBLES

Till and Call absorption towards the Superbubble LMC2 m the Large Magellanic Cloud

ADELINE CAULET ST-ECF, ESO, Karl-Schwarzschild-Str.2, Garching bei Miinchen, D-85748 Germany

1. Introduction

Today, there is considerable interest in studying Superbubbles. These gigantic bubbles are blown in galaxies by the action of very powerful phenomena (e.g. stellar winds and multiple supernovae explosions (SNE) in stellar associations, collisions of clouds with galactic disks). They represent the largest space laboratory to study the physics of a dynamic multiphase interstellar medium (ISM). As a hydrodynamical consequence of the star forming processes via interaction with the surrounding gaseous medium, superbubbles may have occupied a significant volume fraction in the ISM of galaxies at the time they formed their first stars, and during their evolution in conjunction with starburst episodes. In the present day universe, well known examples are the Galactic HI supershells, the supershells of the Large and Small Magellanic Clouds (LMC & SMC), those in nearby galaxies GR8, MI0l, NGC 4449, M82 identified either in H I (radio emission) and/or by their bright optical line emission (Ha). In the case of M 82, a powerful superwind emitted from the nuclear region is responsible for the observed superbubble.

Providing an enormous energy input to the ISM (typically rv 1053 ergs in 107 years), they have important effects on local dynamics and on galactic halos. The basic mechanisms of formation and evolution of superbubbles are connected to superwinds; a good review on the superwind phenomenon has been given by (Heckman et aZ., 1993). Ideas that link the superwind/superbubble phenomenon to galaxy formation and their evolution, and to high redshift galaxies (QSO absorption line systems) are being approached by theorists (Kauffmann, 1995). In the simulations, the inter-

255


256 ADELINE CA ULET

play of stars and gas in disk galaxies forming at the center of merging dark matter halos is considered with gas cooling and infall from the surrounding hot halo. Global observational properties of present-day disk galaxies are successfully reproduced such as gas and stellar surface density profiles, the relationships between luminosity, gas content and metaUicity. It has been clear for several years that a realistic description of the multi phase nature of the ISM is needed to avoid producing unrealistic disk galaxies, e.g. with a topology of "Swiss cheese" because the gas is blasted away by the multiple SNEs (McKee, 1995; Cox, 1995). Observing the superbubble phenomenon could lead to useful constraints on these theories where ejection of metals into the halo and the intracluster medium (ICM), gas accretion and infall from halos onto the galactic disk are predicted to occur during the blow-out of superbubbles. This conference paper reports on observations of cold and warm gas clouds in the "windy" multi-phase ISM of a superbubble.

2. Probing LMC2 gas in absorption

The Superbubble LMC2 is located eastwards of a huge complex of giant H II regions including 30 Dor Nebula in the nearby Large Magellanic Cloud. LMC2 has been investigated in the HI 21 em emission line, optically and in the soft X-ray emission (Caulet et at., 1982; Wang et at., 1991). The ionized Ha filamentary gas of LMC2 forms a half shell of radius 475 pc expanding with a velocity of 30 km s-1 into the ambient H I gas. As the superbubble expands above the disk into the halo, the lower density gas is best detected via absorption line spectroscopy. Many UV and optical absorption lines are expected to arise in the super bubble envelope, corresponding to the predicted temperature drop from 105 to 103 K across rv 50 pc thick funnel walls (Tomisaka et at., 1986). To probe the 3-D gas structure of LMC2, we have observed 7 LMC OB supergiants in the direction of LMC2 at high resolution with the GHRS onboard HST and CASPEC at LaSilla. Here we give the results on cold and warm gas from the ground-based data (details in Caulet & Newell, 1995). The observed ion Ti II (3~{83.8 A, ionization potential 13.6 eV) probes the diffuse H I gas; the observed ion Call (3933.7 and 3968.5 A, IP 11.9 eV) probes the warm clouds and the diffuse ICM. Ca and Ti are heavily depleted onto dust grains. In the Galactic halo (Albert et at., 1993), Ti seems to be depleted in denser regions; the distribution of Ti gas seems to be more uniform and perhaps more extended than that of Ca, although the latter property has been recently challenged by Lipman & Pettini (1995). These properties make Ti II and Ca II ideal to probe the cold and warm phases of the 3-D ISM of the superbubble, and to relate its properties to the halo and ICM cooling gas. Unlike Ca, Ti abundances can be derived directly without any assumption on the relative contribution of

ABSORPTION LINES FROM SUPERBUBBLES 257

SK-69290

1.5

0.5

0.0 H a 269-280 H

-200 -100 o 100 ~o 300 400 Hehocen1nc Velocl1y (kmls)

SK-69282 2.0

1.5

HI

Till 1.0

CaliK

0.5

0.0 -200 -100 o 100 200 300 400

Hellocentnc VeloCI1y (kmls)

Figure 1_ Spectra of 2 background stars behind LMC2j the H I emission line and HO' emission velocity range near the star positions are also plotted.

the specie at different ionization stages, since the ion Ti II is the dominant ionization stage in H I regions.

3. Main results

Sample spectra of two of our background stars are represented in Fig. l. The spectral resolution is 0.11 A and 0.16 A for Ti and Ca respectively. The range of signal-to-ratio is 16-84 and 58-143, giving equivalent width limits at 30' level between 6-30 rnA and 2-6 rnA for Ti and Ca respectively. The

258 ADELINE CAULET

GO Gascous dlsk OH Gaseous halo

FW Free wmd

SW Tlun Shell of shocked-WInd fiDld

CD Contact dlsconbnwty

So. Shocked clouds BS. . .J Bow shoclca 10 the wmd

., Multi-phase ISM clouds

o 08 stars. SN exploSIons 108(8" temp K)

Figure 2. Blow-out of a Superbubble in a multiphase ISM (modified after Heckman et al., 1993).

line profiles are plotted against heliocentric velocities. Line-profile fitting was made to derive the clouds parameters, namely Ti II and Ca II column densities, doppler-widths, velocities. Interstellar absorption occurs over a large heliocentric velocity range from -30 to +360 km S-1 which is attributed to local gas, the halo of the Galaxy, intermediate velocity gas and to the LMC large scale structures (Luks & Rohlfs, 1992), see details in (Caulet & Newell, 1995). The strengths of Ca and Ti lines seem to indicate reduced depletion in the low density shocked ambient gas of LMC2.

Figure 2 shows a schematic 3D representation of blow-out of a superbubble where theoretical ideas on superwind and multiphase halo gas are incorporated_ Inside the superbubble, the clouds are embedded in hotter shocked wind materiaL Figure 1 shows remarkable absorption profiles with broad asymmetric wings bluewards of the main disk components; the wings extend from +220 down to +150 km s-1, perhaps a signature of thermal-

ABSORPTION LINES FROM SUPERBUBBLES 259

ization (shock heating) in high speed collisions between material ejected in the shocked wind and multiple SNEs. It is interesting here to recall the explosion of the supernova 1987 A in the vicinity of LMC2. Similar extended broad interstellar absorption wings were first observed in the spectrum of the central star cluster of 30 Doradus (Blades et al., 1980), then towards more stars of the 30 Doradus Nebula (Songaila et ai., 1986), leading to the interpretation that the wings are generated by energetic winds from 30 Dar stars. Here the presence of such wings is observed as far as 23 arcminutes from 30 Doradus; this implies that the "wind" phenomenon is more extended.

By comparing with kinematical data at other wavelengths (Ha, HI), we conclude that the gas clouds in the velocity range 150-220 km s-1 are moving upwards, as pushed by the expansion of the superbubble above the disk (towards the observer). Downward motions within the 3D LMC2 are clearly identified by the high velocity absorption at 300 km s-1 located in front of the lower velocity LMC components, at '" 245 and 280 km s-1, the so called Land D H I components (Luks & Rohlfs, 1992). All together, these observations which reveal for the first time the interior of an extragalactic superbubble in the temperature range of cold and warm gas support the existence of galactic fountains, superwinds and falling high velocity clouds involved in the complex dynamics of superbubbles.

References

Albert, C. E., Blades, J. C., Morton, D. C., Lockman, F. J., Proulx, M., & Ferrarese, L. 1993, Astrophys. J. Suppl., 88,81

Blades, J. C., & Meaburn, J. 1980, Mon. Not. R.astr. Soc., , 190, 59p Caulet, A., Deharveng, 1., Georgelin, Y. M. & Georgelin, Y. P. 1982, A&A" 110, 185 Caulet, A., & Newell, R. 1995, Astrophys. J., submitted Cox, P. D. 1995, in Astronomical Society of the Pacific Conference Series 80, The Physics

of the Interstellar Medium and Intergalactic Medium, ed. A. Ferrara, C.F. McKee, C. Heiles, & P.R. Shapiro (Dordrecht: Kluwer), 317

Heckman, T. M., Lehnert, M. D., & Armus, L. 1993, in Astrophysics and Space Science Library 188, The Environment and Evolution of Galaxies, ed. J.M. Shull & H.A. Thronson (Dordrecht: Kluwer), 455

Kauffmann, G. 1995, this conference Lipman, K., & Pettini, M. 1995, Astrophys. J., 442, 628 Luks, Th., & Rohlfs, K. 1992, A&A, 263, 41 McKee, C. F. 1995, in Astronomical Society of the Pacific Conference Series 80, The

Physics of the Interstellar Medium and Intergalactic Medium, ed. A. Ferrara, C.F. McKee, C. Heiles, & P.R. Shapiro (Dordrecht: Kluwer), 292

Songaila, A., Blades, J. C., Hu, E. M., & Cowie, L. 1. 1986, Astrophys. J., 303, 198 Tomisaka, K., & Ikeuchi, S. 1986, PAS Japan, 38, 697 Wang, Q., Halmilton, T., Helfand, D. J., & Wu, X. 1991, Astrophys. J., 374, 475

GRAVITATIONAL LENSES AND DAMPED LYMAN ALPHA SYSTEMS

A. SMETTE Kapteyn Astronomical Institute Postbus 800, NL-9700 AV Groningen, The Netherlands

Abstract. We present two cases of gravitationally lensed quasars that have damped Ly 0: lines in the spectrum of one image and not in the spectrum of the second. We study how the statistical quantities derived from surveys for damped Ly 0: systems can be affected by gravitational lensing due to the absorbers themselves.

1. Introduction

Following the arguments given by Wolfe et al. (1986), the progenitors of present-day disk galaxies are thought to be responsible for the damped Ly 0:

systems (hereafter, D LAs) observed in quasar spectra. The extent ofthe gas associated with them thus appears as an important

parameter to constrain the evolution of spiral galaxies. However, only a couple of observations exists: the largest lower limit (16 kpc1 ) is set by the 21cm H I absorption towards the extended radio-sources PKS 0458-020 (Briggs et al. 1989). The first aim of this contribution is to present two cases of gravitationally lensed quasars whose spectra contain DLAs at z ~ 1.6.

Recent results show that the cosmological density of neutral hydrogen (OHI) present in the DLAs at z ~ 3.5 is very close to the cosmological density associated with stars (Ostars) at present epoch: this suggests that the material necessary to form the present day stars already lied in collapsed objects at z ~ 3.5. As a consequence, the variation of OHI with z appears to be of prime importance to study the overall star formation processes.

At first sight, the determination of OHI from a DLA survey is very simple as OHI(Z) ()( l:i Ni, where Ni are the observed column densities

lThroughout the paper, we use Ho = 50 hso km s-l Mpc- 1 , qo = 0.5, A = 0

261


262 A. SMETTE

in the survey; the proportionality constant depends on the redshift range considered. However, such determination of nHI(Z) strongly depends on the highest column density systems present in the survey, which, unfortunately, are also the most susceptible to be affected by biases.

Fall (these proceedings) has discussed the effects introduced by dust in the DLAs themselves. Here, we shall consider the mechanisms by which gravitational lensing due to the DLA absorbers can affect existing surveys.

2. Probing the damped Lya absorber extent with lensed QSOs

2.1. THE Z = 1.6616 DLA IN FRONT OF HE 1104-1805

The two images of HE 1104-1805 (the Double Hamburger) are separated by 3.0" and are both bright enough to allow reasonable SIN spectroscopy (Wisotzki et al. 1993). We have obtained AAT spectra for a study mainly devoted to the Lya forest (Smette et al. 1995). The spectrum of the component A shows a DLA line with a Ng1 = 6.3 X 1020 cm-2 at Z = 1.6616, which is absent in B; instead, there are 3 low equivalent width lines that can be attributed to Lya lines, so that we can infer that the total column density towards B is at most NliI :::::: 1.3 X 1019 cm -2 in this system.

Several low-ionization metal lines associated with the DLA in A are also present in B, but with a systematically lower equivalent width; there is no significant difference in velocity between them (within about 10 km s-I). On the other hand, the high-ionization (C IV and Si IV) lines split to form 3 sub-components in the spectrum of A, but only 2 in the spectrum of B. Only one of them has a similar equivalent width in A and in B, but it presents a velocity difference of:::::: 30 ± 14 km s-1 .

With the assumption that the so far unobserved lensing galaxy lies in the redshift domain between 1 and 1.6, the separation between the linesof-sight at the redshift of the DLA is in the range 8 to 25 kpc.

2.2. THE Z=1.662 DLA IN FRONT OF Q 1429-008

Q 1429-008 is a probable gravitational lens candidate (Hewett et al. 1989). The two quasar images, A & B, are separated by 5.14", present the same redshift (2.076), and have R-band magnitude mR = 17.7 and 20.8. No lensing galaxy has been observed so far.

We obtained 2 A resolution spectra of the Ly a forest for A & B with the CTIO 4m telescope, and additional high-resolution (R :::::: 33000) spectra for A with UCLES on the AAT. We observed a common high NHI Lya absorption line at Z = 1.662 in the spectra of both Q 1429-008 A & B: for the line in the A spectrum, we infer NHI = 3 X 1020 cm-2 • The measurement of the equivalent width ofthe corresponding line in the B spectrum critically

LENSES AND DLAS 263

depends on the definition ofthe continuum, a delicate operation for low SIN spectra as the one ofthe B component. A preliminary continuum placement leads to a value of 2.5 ± 1 A corresponding to NHI ~ 1019 cm -2. Further details shall appear in Smette et al. (1996).

We also observed a new absorption system at z = 1.42, for which we measure a velocity difference of ~ 580 km S-1 between the two images: this fact favors the hypothesis that this system is the one associated with the lens. One can thus compute the linear separation between the two lines-ofsight at the DLA redshift: we find 23 kpc.

2.3. CONCLUSION

These values are compatible with the current view that damped Lya systems arise in progenitors of present-day spiral galaxy disks with diameters of '" 40 - 60 hS'l kpc, embedded in '" 160 hS'OI kpc diameter halos giving rise to Lyman limit systems.

3. Do gravitational lenses affect damped Lya surveys?

Statistics of DLAs are based on the assumption that the lines-of-sight towards background quasars are uniformly distributed over the sky and thus are unaffected by the DLAs themselves. However, gravitational lensing (GL) provides at least the two following mechanisms that undermine this assumption:

the "by-pass" effect: a random line-of-sight has its effective impact parameter increased relatively to the case when no lensing is taking place; the amplification bias: a set of quasars selected on the basis of their (bright) apparent magnitude is likely to contain a significant fraction of quasars whose apparent brightness has been boosted by GL amplification due to the galaxy associated with DLAs.

We studied these effects and performed statistical tests devised to check whether existing surveys of DLAs are affected by GL. We assumed that the DLAs are the only GL agents, that they arise in spiral galaxies, immersed in dark matter halos which can be described by a simple model. We assume that absorption by dust is negligible.

3.1. THE BY-PASS EFFECT

The by-pass effect is a direct application of the lens equation; it corresponds to the second term in the left-hand part of the following equation. A Line Of Sight (LOS) towards a QSO can probe the central part of a galaxy up

264 A. SMETTE

to a radius Rmax if DIs

Os + -D Q < ORmax ' (1) os

where Os is the angular separation between the center of the lensing galaxy and the real QSO location, 0Rmax = Rmaxl Dol, and Q is the angular value of the deflection of light. Finally, Doh DIs, Dos are the angular-diameter distances between the observer and the lens, the lens and the source and the observer and the source, respectively. This effect (function of the background QSO redshift Zq) may decrease by more than 20% the effective cross-section of the galaxies for DLA absorption; furthermore, their central part is avoided.

3.2. THE AMPLIFICATION BIAS

The amplification bias factor B(bq ) gives the fraction of QSOs which appear brighter than a given apparent magnitude bq due to lensing, in this case, by the DLA itself:

(2)

where Nq(bq) is the (supposedly intrinsic) number of QSOs with magnitude bq per unit solid angle and p(A) dA is the probability that the amplification be in the range [A, A + dA]. Compared to the common use of the amplification bias (Fukugita & Turner 1991), we also consider cases that give rise to a single image, leading to a somewhat different expression for p(A) dA, as detailed in Smette, Claeskens & Surdej (1996). One must also remark that the amplification bias depends not only on bq , but, via A, depends also on Zq and ZDLA.

As a consequence of this effect, we find that DLAs produced by neutral hydrogen located at an impact parameter equal to twice the Einstein radius of the host galaxy are preferentially observed in DLA surveys.

3.3. RESULTS

We refer to Smette, Claeskens & Surdej (1996) for a more detailed description of the results. Basically, the combination of the two effects may lead to severely over-estimate the number of DLAs with high column densities at low ZDLA (say in the range 0.3-0.6) in front of bright (bq < 17), Zq > 1 quasars

However, the existing surveys for high-z DLAs have characteristics that preclude the detection of strong GL effects:

DLAs can be detected even in relatively low SIN spectra: thus faint QSOs can be used, for which the amplification bias does not work well;

LENSES AND DLAS 265

for high-zq QSOs, the redshift range in which DLAs are searched for is limited to a redshift domain just somewhat smaller than Zq.

Indeed, we find that the high-z (z > 1.6) survey (Lanzetta et al. 1991) is not significantly affected by GL. However, if the WE (z < 1.6) survey (Lanzetta et al. 1995) only detected 3 DLAs, they all lie in regions of the (bq , ZOLA, Zq) domain for which the strongest GL effects are expected. No definite conclusion can be drawn due to the paucity of data, but HST direct imagery of the 3 background QSOs should possibly reveal the signature of strong GL effects in the form of multiple lensed QSO images and should thus allow to set unique constraints on the mass of the DLAs.

If the H I profile of distant galaxies also present a 'hole' at their center as frequently observed in local spiral galaxies (cf. Rao & Briggs 1993), then the effect of lensing on the determination of !1HI is also decreased.

Bartelmann & Loeb (ApJ, in press) have independently presented a similar, but theoretical, work, suggesting that lensing may lead to stronger effects than described in our study. The reasons are that they use H I profiles for galaxies peaked at the center and, more importantly, consider conditions on bq , ZOLA and Zq that are more suitable for lensing, but not typical of existing DLA surveys.

Acknowledgments. It is a great pleasure to thank the collaboration of Gordon Robertson, Gerry Williger, Peter Shaver, Dieter Reimers, Lutz Wisotski and Thomas Kohler for the work on HE 1104-1805 and/or Q 1429-008, and Jean-FraIl!;ois Claeskens and Jean Surdej for the study devoted to the lensing effects on the statistics of the DLAs. This research was financially supported by grant no. 781-73-058 from the Netherlands Foundation for Research in Astronomy (ASTRON) which receives its funds from the Netherlands Organisation for Scientific Research (NWO).

References

Briggs, F.H., Wolfe, A.M., Liszt, H.S., Davis, M.M., Turner, K.1. 1989, ApJ 341, 650 Fukugita, M., Turner, E.1. 1991, MNRAS 253, 99 Hewett, P.C., Webster, R.L., Harding, M.E., Jedrzejewski, R.I., Foltz, C.B., Chaffee,

F.H., Irwin, M.J., Le Fevre, O. 1989, ApJ 346, L61 Lanzetta, K.M., Wolfe, A.M., Turnshek, D.A., Lu, 1., McMahon, R. G., Hazard, C. 1991,

ApJS, 77, 1 Lanzetta, K.M., Wolfe, A.M., Turnshek, D.A. 1995, ApJ 440, 435 Rao, S.M., Briggs, F.II. 1993, ApJ 419,515 Smette, A., Robertson, J.G., Shaver, P.A., Reimers, D., Wisotzki, 1., Kohler, Th. 1995,

A&AS 113, 199 Smette, A., Claeskens, J.-F., Surdej, J. 1996, A&A submitted Smette, A., et al. 1996, in preparation Wisotzki, L., Kohler, T., Kayser, R., Reimers, D. 1993, A&A 278, L15 Wolfe, A.M., TUfllshek, D.A.,Smith, H.E., Cohen, R.D. 1986, ApJS, 61, 249

HI 21 eM LINE OBSERVATIONS OF DAMPED LYMAN ALPHA SYSTEMS

C. CARILLI Smithsonian Astrophysical Observatory, Cambridge, MA and Leiden Observatory, Leiden, The Netherlands

w. LANE Smithsonian Astrophysical Observatory, Cambridge, MA and Rice University Department of Physics, Houston, TX

A.G. DE BRUYN Netherlands Foundation for Research in Astronomy, Dwingeloo, The Netherlands, and Kapteyn Astronomical Institute, Groningen, The Netherlands

R.BRAUN Netherlands Foundation for Research in Astronomy, Dwingeloo, The Netherlands

AND

G.K. MILEY Leiden Observatory, Leiden, The Netherlands

Abstract. We have used the Westerbork Synthesis Radio Telescope to search for HI 21 cm absorption associated with high redshift damped Lya systems. We calculate limits to harmonic mean spin temperatures, <Ts>, for these systems, and find typical lower limits ~ 103 K, consistent with previous results for other damped Lya systems. We then analyze the relationship between the total H I column density and the integrated 21 cm absorption line equivalent width, EW2I. for damped Lya systems. We compare these results with similar results for lines of sight through our own galaxy. A trend is seen in which the damped Lya systems show a systematically lower EW21 at a given total N(H I) than Galactic lines of sight, by a factor ~ 3 to 5. In the context of a two temperature phase model for the absorbing gas, these results suggest that for a given total neutral hydrogen column density, the damped Lya systems contain a larger percentage of warm gas than is seen in typical Galactic lines of sight.

267

M. N. Bremer et al. (eds.). Cold Gas at High RedshiJt. 267-278. © 1996 Kluwer Academic Publishers.

268 CARILLI ET AL.

We derive upper limits to HI 21 cm emission from the vicinity of the damped Ly 0: absorbers. The mass limits are adequate to rule out a substantial contribution by protoclusters of the type hypothesized by Dort (1984) to the observed statistics of damped Ly 0: systems at high redshift. Lastly, we derive a lower limit to the molecular hydrogen fraction in the cold gas phase for the absorption system towards 0528-2505 of F = 2N(H2)jN(HI) > 2%.

1. Introduction

Damped Ly 0: systems represent the high column density end of the distribution function for quasar absorption line systems (Lanzetta et al. 1991, Tytler 1987). These systems have a number of unique properties. The damped Ly 0: absorption line is easily identified among the many lines of the Ly 0: forest. Further, since the damped Ly 0: line profile is dominated by the radiation damping wings of the Voigt profile, it permits the derivation of an H I column density without dependence on the velocity width (Wolfe et al. 1986).

Damped Ly 0: systems are thought to indicate absorption by gas in large, flattened, rotating "proto-disk" galaxies, with radii roughly twice that of present day spiral disks. There are several reasons for this theory. First, the implied HI column densities (N(HI) 2:: 2 X 1020 cm-2 ), the velocity structure, and the multiplicity of lines are comparable to those expected for lines of sight through normal spiral galaxy disks (Wolfe 1988, Briggs and Wolfe 1983). Second, the cosmic mass density of H I found in damped Ly 0:

systems at z 2:: 2 is approximately the same as the cosmic mass density of luminous matter found in spiral galaxies at the present epoch (Lanzetta et al. 1991, Rao and Briggs 1993). And third, the "protogalaxy" aspect of these systems is based on the fact that the number of damped Lyex systems per unit redshift is roughly four times that predicted for normal spiral disks (Wolfe et al. 1986, Lanzetta et al. 1991). It is hypothesized that these protodisks are the uncollapsed precursors of present day spiral galaxies (Wolfe 1988).

Large H I column densities make Ly 0: systems good candidates for detection of HI 21 cm absorption. Studies of redshifted HI 21 cm lines associated with damped Ly 0: systems reveal detailed information about the physical conditions within the gas disks. Background radio sources can be spatially extended, so the spatial structures of the absorbing cloud can be mapped using radio-interferometry. The high velocity resolution obtainable with radio observations, typically one to two orders of magnitude better

21 eM OBSERVATIONS OF DAMPED LYMAN ALPHA SYSTEMS 269

than with optical observations, allows the accurate determination of kinematic properties. Lastly, comparison of column densities derived from HI 21 em observations with those found from the corresponding damped Ly 0'

absorption allows one to estimate the temperature of the absorbing gas. Herein we present HI 21 cm observations of three damped Ly 0' sys

tems using the recently upgraded Westerbork Synthesis Radio Telescope (WSRT). The upgraded WSRT allows us to probe redshift ranges which were previously unobservable using interferometers. In this paper we use h == HallOO km sec-1 Mpc-1 and qo = 0.5.

2. The Systems

The three damped Ly 0' systems observed are towards the radio loud quasars 0336-0142,0528-2505, and 2342+3417. The redshifts for the quasars and the absorbing systems are given in Table 1, along with the column density for the absorbing H I derived from the damped Ly 0' profile.

The damped Ly 0' absorption system towards 0;~36-0142 is described in detail in Lu et al. (1993). They find a redshift of Zabs = 3.0619 from the associated metal lines. The quasar itself is a Parkes radio source, and has a flux density at a frequency of 350 MHz of about 1 Jy.

The absorption system towards 2342+3417 is described in detail in White et al. (1993). They find a damped Ly 0' system of rest equivalent width of 32 A. The absorption system has a redshift of 2.9084, determined from associated metal lines. The quasar has a flux density of 300 mJy at 365MHz.

The redshift difference between the quasar and the absorbing systems in the above two cases are large enough to conclude that the absorbing gas is cosmologically intervening. However, in the case of the third system, 0528-2505, the absorbing gas is blue shifted (i.e. infalling in velocity) relative to the quasar by about 3200 km sec-I, and must therefore be associated with the quasar, either by gas surrounding the AGN itself, or by gas in a galaxy in a (hypothetical) associated cluster. This system is unique in that it shows absorption by molecular hydrogen with a column density of: N(H2) ~ 1 X 1018 cm- 2 (Foltz, Chaffee, and Black 1988). We discuss this point in detail below.


All the observations were made with the Westerbork Synthesis Radio Telescope (WSRT) in the spring of 1994, using the new broad band 92 cm system. As part of a series of upgrades to the WSRT the IF filters of the 92 cm receiver system have been adjusted to provide an RF passband matched to tha.t dictated by the response of the feed. The frequency range is now

270 CARILLI ET AL.

TABLE 1. Vital Statistics

0336-0142 0528-2505 2342+3417

Zem 3.197 2.770 3.01

Zabs 3.0619 ± 0.0003 2.8110 ± 0.0003 2.9084 ± 0.0005

flux density (mJy) 940 140 300 Channel width (kms-l) 16.8 7.8 4.0

RMS (mJy) 5.1 8.5 13.4

Optical depth (30-) 0.016 0.18 0.13

N(HI)Lya (cm-2 ) 1.5 x 1021 2.2 X 1021 2.0 X 1021

N(H Ihl (30- cm-2 ) 5.9 x 1017 T. 30 X 1017 T. 11 X 1017 T. T. (30- K) 2500 730 1800

0336-142: Lu et al. 1993; 0528-2505: Mf/lller and Warren 1993; 2342+3417: White et al. 1993

from 305 to 390 MHz, or almost twice that previously available (de Bruyn 1990, Braun 1993, Carilli, de Bruyn, and Boonstra 1994). The details of the observations and data reduction are summarized in Lane et al. (1996).

Spectra at full velocity resolution were extracted at the position of the target continuum source (note: for a uniform spectral taper, as employed herein, the effective velocity resolution is 1.2 times the channel spacing). Images and spectra of the sources are presented in Lane et al. (1996). In all cases the spectral RMS equaled the expected thermal noise. For sources observed on more than one day the spectrum from each day was velocity corrected to a heliocentric frame, and in the case of 0528-2505 hanning smoothed and resampled to matching velocity resolution, and summed. The parameters for each day's spectrum at full resolution are listed in Table 1. No absorption was detected at 30' or greater for any of the sources. Spectra for all sources were then smoothed and resampled to channel width ~ 16 km sec-I, and ~ 32 km sec-I, and searched for absorption. The spectral RMS values decreased with increasing channel width as RMS ex: i::l.v- 1/ 2 ,

as expected. Again, no absorption was detected at 30' or greater for any of the sources at these resolutions.

Lastly, the image cubes for all sources were smoothed and resampled to channel width of ~ 140 km sec-I, and visually inspected for emission. No signal was detected above 50' in any field. The RMS values at this resolution are listed in Table 3.

21 CM OBSERVATIONS OF DAMPED LYMAN ALPHA SYSTEMS 271

4. Discussion

4.1. THE N(H I) - EW21 RELATIONSHIP

The equation relating neutral hydrogen column density, N(H I), to observed optical depth in the 21 cm absorption line is:

where EW 21 is the integral of optical depth over velocity, / is the fraction of the radio continuum source covered by absorbing gas, and Ts is the spin temperature of the gas. For a single line with a Gaussian profile, EW21 = 1.06 X T~V, where T is the peak optical depth and ~V is the velocity FWHM (km seC!). For multiple Gaussian components the EW 21 is then summed over all components. In most astrophysical circumstances the spin temperature equals the kinetic temperature. A standard analytic tool in the study of damped Lya systems has been to compare the total N(H I) derived from damped Lya observations with N(H I) derived from 21 cm absorption observations in order to derive the spin temperature of the absorbing gas (Wolfe and Davis 1979). A likely physical situation is that the absorbing gas has multiple temperature phases. In this case, the Ts value derived from a naive comparison 21 cm and damped Lya absorption lines is a column density weighted harmonic mean temperature, which we denote <Ts>. It can also be considered an upper limit to the temperature of the coldest phase.

Table 1 lists lower limits to <Ts> values for the three systems in our study. A summary of all previous results on spin temperature calculations for damped Lya systems can be found in de Bruyn et at. (1995). Typical <To> values for most damped Lya systems are comparable to the limits in Table 1: <Ts> ;:::;J 103 K. For comparison, individual Galactic clouds with column densities greater than about 1021 cm- 2 typically have <Ts> values near 100 K (c/. Braun and Walterbos 1992). This difference between <Ts > values for Galactic clouds and damped Lya systems was first pointed out by Wolfe and Davis (1979), based on the observation of a single system. The data in Table 2, and in de Bruyn et al. (1995), confirm their conclusion as a general trend for damped Lya systems.

While the <Ts> analysis above suggests a physical difference between damped Lya systems and Galactic clouds, a proper comparison between the Galactic <Ts> values and those for damped Ly tY systems remains problematic for the following reason. The Galactic <Ts> analysis uses 21 cm emission and absorption data on individual "clouds" - i.e. isolated velocity components (Dickey, Salpeter and Terzian 1978, Payne, Sa.lpeter, and Terzian 1982, Colgan, Salpeter, and Terzian 1988). This is not possible for the damped Lya systems, due to the lack of velocity information for the

272 CARILLI ET AL.

0

20 - -0

0 0

0

15 r- -0

0

0 0

I I 0

" 0 OJ

'- 0

e C- o ~;; 10 - 0

0 -~ 0

0 0 T

0 ! 0

0

0

cP 0 0

5 - 0 0 -0 0 0

0 ~ 0 0

oD 0

Q:&q,o 0 ty I

fJk ~i I or- -

I I I I I

0 20 40 60 80 100 N(H!) (xl0" em-Z)

Figure 1. The open circles show the N(H I) - EW21 relat.ionship for Galactic H I, The results for the damped Ly Cl' systems are shown as points with error bars. The arrows indicate limits (317) for the three new systems in this study.

damped Lya line. Hence, the <Ts> values for the damped Lya systems correspond to integrated measurements.

To remedy this situation, we have reanalyzed data on Galactic absorption and emission (from Dickey, Salpeter and Terzian 1978, Payne, Salpeter, and Terzian 1982, Colgan, Salpeter, and Terzian 1988), with the purpose of deriving integrated line of sight measures of total N(H I), and the integrated EW21. The results for N(HI) and EW21 for damped Lya systems are given in Table 2, and are plotted against Galactic data in Fig.!. There is a general trend for increasing EW 21 with increasing N (H I) for the Galactic data. There is also a clear trend for the damped Ly (1' systems to be situated


along the lower edge of the envelope defined by the Galactic data, i.e. for a given total N(H I), the EW21 values for damped LyO' systems are typically less than expected for lines of sight through our galaxy, by a factor ~ 3 to 5. The only exception is the lowest redshift system in our study, towards 0235+164, which falls within the Galactic relation in Fig. 1. Vanny et ai. (1990) find that the line of sight towards 0235+164 passes through the optical disk of a starburst galaxy at z = 0.524.

Systematically lower EW21 values for given total N(H I) for damped Ly 0' systems versus Galactic lines of sight could indicate an overall warmer temperature for the neutral gas in the damped Ly 0' systems. An alternative, perhaps more physically reasonable, model is to assume two temperature phases for the H I: a cold phase at ::; 100 K and a warm phase at ~ 8000 K (Braun and Walterbos 1992). In this case, the results in Fig. 1 suggest that for a given total neutral hydrogen column density, the damped Ly 0'

systems contain a larger percentage of warm phase gas than is seen in typical Galactic lines of sight, by a factor of about 3 to 5.

TABLE 2. The N(H I) - EW21 R.elationship

Source a Zqua.sa.r Z!b::; N(H I)C EW~1 rer

1021 cm-2 kms-1

0235+164 0.94 0.524 3.1 ± 1.5 13 ± 0.6 1,2

3C286 0.849 0.692 2.0 ± 0.12 0.91 ± 0.09 3,4

1331+170 2.081 1.776 0.76±0.16 0.4 ± 0.1 5

1157+014 1.986 1.944 6.3 ± 1.0 2.3 ± 0.2 6 0458-020 2.286 2.038 8±4 7.8 ± 0.4 7,13

2342+3417 3.01 2.908 2.0 ± 0.5 ~ 1.4' 9,10

0528-2505 2.770 2.811 2.2 ± 0.1 ~ 2.5 9,11

0336-0142 3.197 3.062 1.5 ± 0.3 ~ 0.34 9,12 0201+113 :l.61 3.388 3.0 ± 0.9 1.5 ± 0.5 8

a Heliocentric redshift of the background quasar. bHeliocentric redshift. of t.he absorpt.ion line system. "Total neutral hydrogen column density derived from t.he damped Ly (Y line. dEW 21 = the integral of 21 cm absorption line optical depth over velocity. eCode for references for the original data: 1. Wolfe et al. 1978 ,2. Snidjers et al. 1982, 3. Davis and May 1978,4. Cohen et al. 1994, 5. Wolfe and Davis 1979, 6. Wolfe, Briggs, and Jauncey 1981, 7. Wolfe et al. 1985, 8. de Bruyn et al. 1995, 9. this work ,10. White et al. 1993, 11. M(2Siler and Warren 1993, 12. Lu et al. 1993, Briggs et al. 1989. 'Limits are 3<T at :=:::: 16 km sec- 1 channel- 1 .

There are number of uncertainties when interpreting the data in Table 2 and Fig. 1. First is the uncertainty in the covering factor. The problem arises if the optical and radio lines of sight sample significantly different gas, as could be the case if the spatial extent of the radio source is substantially

274 CARILLI ET AL.

larger than the optical quasar, and the absorbing cloud is smaller than the background radio source. We have used f = 1 in all cases. The most important observation in this regard was the direct measurement of the HI 21 cm covering factor towards 0458-020 by Briggs et al. (1989) using very long baseline interferometry (VLBI) line and continuum observations. They found f = 1, with an implied lower limit to the absorbing cloud size of 8h-1 kpc. A similar observation was made by Wolfe et al. (1976) of the absorbing system towards 3C 286. They also found f = 1 for this system, with a lower limit to the absorbing cloud size of 0.1h-1 kpc. For 1331+170 and 1157+014, Briggs and Wolfe (1983) used VLBI continuum observations to derive the fraction of the continuum emission at the frequency of the redshifted 21 cm line coming from a source smaller than a couple hundred parsecs. They then argue for f > 0.5 in both cases, on the assumption that the absorbing clouds are kpc-scale structures. Wolfe et at. (1978) present VLBI observations of 0235+ 164 which show a continuum source size less than 6 mas at the redshifted H I line frequency. Also, their VLBI H I line spectrum of 0235+ 164 shows the same optical depths for the absorption lines as seen in the Arecibo spectrum. Lastly, for 0201+113, de Bruyn et al. show that the continuum source is likely to be smaller than 6 pc at 330 MHz, which again argues strongly for f = 1, unless the absorbing cloud is smaller than a few pc.

The cases for high covering factors for 2342+3417 and 0336-0142 are less certain. The radio spectrum for 2342+3417 between 1.5 GHz and 5 GHz is flat, arguing for a compact (pc-scale) radio source. However, the spectrum between 1.5 GHz and 0.3 GHz is slowly rising (index = -0.5), suggesting a contribution from a more extended component. We estimate that perhaps 50% of the 350 MHz flux density from the source may come from this component, suggesting a lower limit to f 2: 0.5. The source 0336-0142 is a GHz-peaked spectrum radio source, and 5 GHz VLBI observations of this source reveal a very compact source of size less than 7 mas (Gurvits et at. 1994). Whether this small size applies to the low frequency emission remains to be determined. Higher resolution observations at low frequency are required to test the f = 1 assumption for these two sources. Overall, we do not feel that covering factor values substantially less than unity are responsible for the results in Table 2, although a few cases require verification.

The second uncertainty is the fact that the total N (H I) values for Galactic data are derived from 21 cm emission observations, while those for damped Ly 0' systems are derived from the optical spectra. A direct comparison between results from the two methods is reasonable, as demonstrated by Dickey and Lockman (1990), who show a linear relation of slope unity between Galactic column densities derived from damped Ly 0' absorp-

21 CM OBSERVATIONS OF DAMPED LYMAN ALPHA SYSTEMS 275

tion with those derived from 21 cm emission. The third uncertainty is the possibility of the quasar emission affecting

the absorbing gas. Most of these systems in Table 2 can be considered cosmologically intervening absorption systems, since the absorption red shift is substantially lower than the quasar emission redshift (implied separations ~ 1 Mpc). In these cases it is safe to assume that the hyperfine level populations in the absorbing gas are not affected by continuum radiation from the background source (Wolfe, Davis, and Briggs 1982, de Bruyn et al. 1995, Bahcall and Ekers 1969, Field 1959a,b). The only exception is the case of 0528-2505, for which the absorption redshift approximately equals the emission redshift. Whether the absorption towards 0528-2505 is by gas in a galaxy in a (hypothetical) associated cluster, or by gas directly associated with the quasar, remains to be determined.

4.2. DAMPED LYa SYSTEMS AS PROTOCLUSTERS?

The WSRT data on damped Lya systems can be used to search for HI 21 cm emission from "protoclusters" in the field. Detecting HI 21 cm emission from protoclusters remains one of the most important, most difficult, and thus far unsuccessful, observations in radio astronomy. There are a number of predictions for the expected 21 cm emission profile from protoclusters (see Wieringa et al. 1992). As a rough guide we use the simple model of Oort (1984), who suggests a proto cluster will be composed of about 1014 Me of hydrogen, of which abDut 1013 Me will be neutral. The expected emission profile may have a velocity width of a few hundred km seC t, and a size of an arcminute or so (the typical beam size at the observing frequencies was ~ 50").

TABLE 3. H I Emission

0336-0142 0528-2505 2342-3417

Channel Width (km/s) 141 144 147 Redshift Range Searched 0.021 0.0091 0.0095 Volume Searched (h-3 Mpc3 ) 4.4 X 104 1.9 X 104 2.0 X 104

500 (mly/beam) 7 7.5 7 MHI (500 h-2 M 0 ) 1.4x1013 1.0 x 1013 1.0 X 1013

To search for HI 21 cm emission in the vicinity of the damped Ly 0-

clouds in this study, the spectral image cubes were smoothed and resampled to channel width of ~ 140 km seC1 , and then visually inspected for emission. Table 3 summarize the results from the emission search listing

276 CARILLI ET AL.

the 5a mass limit per channel, the total velocity range searched, and the total (comoving) volume searched. No emission was seen in any of the fields at any velocity. Typical limits are:::::: 1013 M0 per channel.

The deepest search, over the largest volume, for HI 21 cm emission from proto clusters was by Wieringa et al. (1992). Their lack of detection of emission in any field allowed them to set an upper limit to the comoving space density of HI 21 cm emitting protoclusters of the type outlined by Oort of ::; 1000h3 Gpc-3 .

Our data reach similar limits to that of Wieringa et a.l., however, the total volume searched is considerably smaller (about 10% or so). Hence, our data are consistent with, but do not alter substantially, the conclusions of Wieringa et al ..

The one important difference between our emission search and that of Wieringa et al. is that the fields presented herein aU contain damped Ly 0'

systems. Wieringa et al. point out that the column densities derived from damped Ly 0' absorption lines are not radically different than those predicted for lines of sight through the neutral gas in protoclusters, thereby leaving open the possibility that at least some of the damped Ly 0' systems may represent absorption by protocluster gas (as opposed to proto-disk galaxy gas). Our limits are deep enough to rule out any proto cluster emission of the type hypothesized by Oort in three damped Ly 0' systems. Although the sample is smail, it seems that the contribution to the statistics of Ly 0' absorption clouds by protocluster gas is not dominant.

4.3. THE MOLECULAR GAS CONTENT OF DAMPED LY 0' SYSTEMS

An important question concerning high redshift quasar absorption line systems is their molecular content (see Lanzetta 1993). The absorption system towards 0528-2505 is the only high redshift system with detected H2 absorption (Foltz et al. 1988). Model fitting to the Lyman- and Werner-band molecular hydrogen lines resulted in a narrow velocity width for the molecular gas, FWHM = 8 km sec-I, implying a temperature of 1400 K if the width is thermal. This temperature is higher than the derived H2 excitation temperature of 100 K, suggesting that the line width is due to turbulence.

Foltz et a.l. (1988) derive a small molecular fraction for the z = 2.811 absorption system towards 0528-2505: F = 2N(H 2 )/N(H I) = 0.2%. However, Foltz et al. point out that this fraction is calculated from the integrated H I column, as derived from the damped Lya line. Again, it is likely that the absorbing gas has multiple temperature phases, and that the molecular gas is associated with the coldest phase. In this case, a comparison of N (H 2) with the N(H I) derived from the 21 cm line (which is also weighted towards the coldest phase), provides a better estimate of the molecular fraction in


the cold phase. Our spectra of 0528-2505 are at 8 km sec-1 resolution, reasonably well

matched to the observed line width ofthe cold gas determined by the H2 line fitting. We can calculate a lower limit to the molecular fraction in the cold absorbing gas towards 0528-2505 by assuming that the spin temperature in the cold phase = the H2 excitation temperature = 100 K. Our upper limit to 21 cm absorption then implies: N(H I) < 1 X 1020 cm-2 (10"), which leads to: F = 2N(H2)/N(HI) > 2%, for the cold phase gas in this system.

Support for W. Lane was provide by the REU summer research program at SAO. We acknowledge support by a programme subsidy provided by the Dutch Organization for Scientific Research (NWO). CLC acknowledges support from a NOVA research fellowship, and from the AXAF science center at the Smithsonian Astrophysical Observatory under NASA contract NAS8-39073.

References

Bahcall, J.N. and Ekers, R.D. 1969, Ap.J., 157, 1055 Bahcall, N.A. 1977, A.R.A.A., 15,505 Braun, R. and Walterbos, R. 1992, Ap.J., 386, 120 Braun, R. 1993, NFRA Newsletter, 5, 3 Briggs, F. and Wolfe, A. 1983, Ap.J., 268, 76 Briggs, F. 1988, in QSO Absorption Lines, eds. J.C. Blades, D. Turnshek and C. Norman,

Cambridge Univ. Press, p. 275 Briggs, F. et al. 1989, Ap.J., 341, 650 Brown, R.L., and Roberts, M.S. 1973, Ap.J. (letters) 184, L7 de Bruyn, A.G. 1990, NFRA Newsletter, 1, 1 de Bruyn, A.G., Baum, S., and O'dea, C. 1995, A.A., in press Carilli, C.L., de Bruyn, A.G., and Boonstra, A.-J. 1994, NFRA Newsletter, 6, 1 Colgan, S.W., Salpeter, E.E., and Terzian, Y. 1988, Ap.J. 328, 275 Cohen, R.D., Barlow, T.A., Beaver, E.A., Junkkarinen, V., Lyons, R., and Smith, H.

1994, Ap.J., 421, 453 Cornwell, T.J., Uson, J.M., and Haddad, N. 1992, A.A 258, 583 Dickey, J. and Lockman, F.J. 1990, A.R.A.A., 28,215 Dickey, J., Salpeter, E.E., and Terzian, Y. 1978, Ap.J.Supp. 36, 77 Draine, B.T. 1978, Ap.J. Supp., 36, 595 Field, G. 1959a, Ap.J., 129, 551 Field, G. 1959b, Ap.J., 129, 536 Foltz, C., Chaffee, F., and Black, J. 1988, Ap.J., 324, 267 Gurvits, L.L, Schilizzi, R.T., et al. 1994, A.A., 291, 737 Lane, W., Carilli, C., de Bruyn, A.G., Braun, R., and Miley, G. 1996, A.J., submitted Lanzetta, K.M., Wolfe, A., Turnshek, D., Lu, L., McMahon, R., and Hazard, C. 1991,

Ap.J. (Supp.), 77, 1 Lanzetta, K.M. 1993, in The Environment and Evolution of Galaxies, eds. J.M. Shull and

R.A. Thronson, Kluwer Academic Publishers, Dordrecht, p. 237 Lu, L., Wolfe, A., Turnshek, D.A., and Lanzetta, K.M. 1993, Ap.J. (Supp.), 84, 1 MlIlller, P. and Warren, S.J. 1993, A.A., 270, 43 Oort, J.R. 1984, A.A., 139, 211 Payne, H.E., Salpeter, E.E., and Terzian, Y. 1982, Ap.J. Supp. 48,199

278 CARILLI ET AL.

Rao, S. and Briggs, F. 1993, Ap.J. 419. 515. Snijders, M.A., Boksenberg, A., Penston, M.V., and Sargent, W.L. 1982, M.N.R.A.S.,

201, 801 Taramapolous, A., Garwood, R., Briggs, F., Wolfe, A. 1995, A.J., 109, 480 Tytler, D. 1987. Ap.J., 321, 49 White, R.L, Kinney, A.L., and Becker, R.H. 1993, Ap.J. 407, 456 Wieringa, M., de Bruyn, A., and Katgert, P. 1992, A.A., 256, 331 Wolfe. A. 1988, in QSO Absorption Lines, eds. J.C. Blades, D. Turnshek and C. Norman,

Cambridge Univ. Press, p. 300 Wolfe, A.M., Broderick, J.J., Condon, J.J., and Johnston, K. 1976, Ap.J. {letters}, 208.

L47 Wolfe, A.M., Brown, R.L., and Roberts, M.S. 1976, Pllys. Rev. Lett., 37,179 Wolfe, A.M., Broderick, J.J., Condon, J.J., and Johnston, K.J. 1978, Ap.J., 222, 752 Wolfe, A.M., Briggs, F.H., Turnshek, D.A., Davis, M., Smith, H., and Cohen, R. 1985,

Ap.J., 294, L67 Wolfe, A.M., Briggs, F.H., and Jauncey, D.L. 1981, Ap.J., 248, 460 Wolfe, A.M., Davis, M, and Briggs, F. 1982, Ap.J. 259, 495 Wolfe, A.M. and Davis, M.M. 1979, A.J., 84, 699 Wolfe, A.M., Turnshek, D.A., Smith, H., and Cohen, R. 1986, Ap.J. Supp. 61, 249 Yanny, B., York, D., and Williams, T. 1990, Ap.J., 351, 377

A RADIO SEARCH FOR HIGH REDSHIFT HI ABSORPTION

J. N. CHENGALUR, A. G. DE BRUYN, R. BRAUN

NFRA, P.O. Box 2, 7990 AA Dwingeloo, The Netherlands

AND

C. CARILLI

Center for Astrophysics, Cambridge, MA, USA

1. Introduction

The rare damped Lya systems seen in absorption against the UV continuum of high redshift QSOs nonetheless dominate the observed neutral gas at high redshift. HI 21 cm absorption studies of these systems are challenging and only a handful of have been detected hitherto (Carilli et al., these proceedings). HI 21 cm observations are however of great value because they potentially yield physical information (velocity dispersion, spin temperature, size), that cannot be obtained by other means.

If some damped Lya systems contain a significant amount of dust, then QSOs behind such systems will be considerably extincted. Further, since high resolution optical spectroscopy is only practical for bright QSOs, such extincted QSOs will be excluded from observing samples, leading to an underestimate of the number count of damped Lya systems, and hence also ofthe gas content of the early universe. If the dust obscuration is sufficient, the number counts of QSOs themselves could be strongly biased, and if the damped Lya population is sharply divided into two halves, one with little or no dust and another with substantial dust, (if for example systems above a certain redshift are much dustier than those at lower redshifts) this bias could in fact be completely undetectable in purely optical studies (Heisler & Ostriker 1988). Such a bias however would be apparent when trying to find optical counterparts to radio sources, and the fact that optical identification was possible for a complete sample of radio sources (Shaver, 1994) makes this unlikely.

Although the effect of obscuration by dust may not be as substantial as envisaged by Heisler & Ostriker, it could nonetheless be important. Two independent observations in fact provide complementary evidence for the

279


280J. N. CHENGALUR, A. G. DE BRUYN, R. BRAUN AND C. CARILLI

presence of dust in high redshift damped Lya systems, (i) QSOs with damped Lya systems along the line of sight have redder UV continuum than those without, (Pei et al. 1991) (ii) The gas phase Cr abundance in these systems is much lower than that of Zn, (Steidel et al. 1995). In the galaxy Cr is depleted onto dust grains while Zn is not. Thus the count of both damped Lya systems and that of QSOs is biased, however since the spectral properties of the dust are poorly constrained, the severity of the bias is correspondingly uncertain. Fall & Pei (1995) estimate that up to 70% of high red shift quasars could be missed.

Blind radio searches (against the lobes of radio galaxies which are often well removed from the central core, however note that since one requires to detect the Lya line to determine the redshift, the line of sight to the central object itself must be relatively dust free), are less subject to the bias introduced by dust extinction and are hence of particular value in resolving this controversy. Such searches have traditionally been difficult because of the challenging instrumental requirements and the hostile radio interference environment at low frequencies. In this paper we describe a pilot radio search for damped Lya systems made using a novel observing mode at the WSRT and present preliminary results.

2. Observations and Data Reduction

2.1. OBSERVATIONS

An absorber with column density NH > 2 X 1020 cm-2 is encountered on the average 0.25 times along a line of sight a unit red shift long and centered at z'" 2.5 (Lanzetta et al. 1991). For a column density NH > 1.0 X 1021 cm- 2

the corresponding probability is 0.09. Note that this probability estimate includes only those absorbers seen against optically bright QSOs and is hence a lower limit if dusty systems do indeed exist. Further, the estimate is based largely on QSOs with z < 3, and there is some evidence that the probability of encountering a high column density absorption system increases (by a factor of 2-3) at a redshift of c::: 3 (White, Kinney & Becker 1993). However, even in the most optimistic scenario, the number count is low enough to make it essential to search a large redshift path interval before the probability of detecting a damped Lya system becomes meaningful. The velocity width of the HI 21 cm absorption signal is probably small, 10 - 100 km s-l (however, any object which large enough to cover the entire radio emitting region is unlikely to have very small velocity width). This combination of high spectral resolution and simultaneously high instantaneous bandwidth is quite challenging to achieve. Further the interference environment at these frequencies (200 - 300 MHz) is often hostile. The new Compound Interferometry (CI) observing mode at the WSRT, along with the newly

HIGH Z H I ABSORPTION 281

commissioned broadband 92 cm system (Carilli et al. 1995) however does make such observations feasible.

In CI mode, the WSRT is split into two phased arrays, and the summed signal from these arrays is cross-correlated, i.e. one has a two-element interferometer, with each element being a phased array. The reduction in the number of measured spatial baselines (from 40 to 1) allows one to achieve high spectral resolution, up to 8192 channels across an instantaneous bandwidth of 20 MHz. In practice this 20 MHz bandwidth is obtained by using 4 contiguous 5 MHz bands, and after allowing for overlap between the bands the usable instantaneous bandwidths is 16.4 MHz. There is reduced sensitivity to interference because of the interferometers rejection of terrestrial signals, and further unlike single dish radio spectroscopy there is no need to spend large amounts of time calibrating the total power induced spectral band pass shape. A first round of observations were conducted in March 1995, when a total redshift interval of 3.5 was observed towards 4 objects. The typical integration time per frequency setting was'" 8000 seconds. Software limitations prevented us from attaining the highest possible resolution; we were instead restricted to a resolution a factor of two worse (Le. '" 25 km s-1).

2.2. DATA REDUCTION

The shape of the spectral baseline (or equivalently the frequency dependence of the visibility) is a function of the distribution of background sources and the hour angle. (For example, a bright source at the 10 dB point of the primary beam would lead to baseline structure on the scales of '" 2 MHz). Figure 1a shows the observed visibility during a single 80 s integration towards the radio source 8C1435, showing the dramatic influence of background sources in determining the shape of spectrum. However, this spectrum can be easily modeled (Fig. 1 b) if one has a map of the sky (as seen by the same telescopes). Modeling is done by special purpose software produced by us, and is in general quite successful although there are occasionally residuals which might be attributable to imperfect knowledge ofthe shape of the primary beam, and also perhaps to some low level cross-talk in the adding stage.

The data reduction proceeds along the following steps: (i) the raw spectra are calibrated to an absolute flux level using observations of calibrator sources interspersed throughout the observation; (ii) the spectra are then corrected for the instrumental bandpass using observations of these same calibrator sources; (iii) model background sources (obtained from an independent continuum map of the field, sometimes from the WENSS survey, and sometimes from other projects) are subtracted from the spectrum; (iv)


325 330

325 330

8C143

335

Frequency (MHz)

8C143 Model

335

Frequency (MHz)

340

340

Figure 1. Upper panel: Observed visibilities during a single 80 sec integration towards tbe radio source 8e 1435. Lower panel: Modeled spectrum.

any residual large scale baseline features are removed by low order polynomial fitting; (v) RFI is flagged and the spectra are co-added to yield the final spectrum.

o <0

o ....I

g I

340

HIGH Z H I ABSORPTION

360

Frequency (MHz)

380

Figure 2. Final spectrum towards 8C1435

3. Preliminary Results

283

Figure 2 shows the final spectrum towards 8C1435, which is a radio galaxy at z rv 4.25 (Lacy et al. 1994). It has a flux at 350 MHz of rv 2.7 Jy and a spectral index a rv -1.2. The low frequency flux is presumably dominated by the two hot spots, which have a separation of rv 5", or rv 20 kpc in a flat n = 1 universe. The total red shift range observed is ~z rv 0.63, which is about 65% of the available red shift (using the WSRT broadband 92 cm system) towards this object. The noise is rv 13 mJ y, which is the expected thermal noise limit. The spectral resolution is rv 25 km s-1.

We do not detect any narrow linewidth absorption in this redshift interval. At the center of the band the 40" upper limit to the optical depth (assuming that the object covers the entire radio emitting region) is T ~


2 X 10-2, or 20 ~V Ts

NH ~ 8.8 X 10 X 25 kms-1 X 1000 K (1)

The data to the remaining objects is being reduced. A further redshift interval of '" 3.5 will be observed shortly (with a frequency resolution a factor of two better than that for these observations). Continuum mapping observations of the same field with the broadband 92 cm should enable us to model the spectra within the thermal noise and thus obviate the necessity to fit low order polynomials and dramatically improve our ability to detect both weak broad lines, (for example from larger scale structure along the line of sight), and also recombination lines from ionized gas associated with the radio galaxy itself.

The WSRT will soon have a UHF(high) system covering the frequency range between 1200 MHz and 700 MHz, with a system temperature'" 75 K at the upper end of the band, which will make it practical to search for HI 21 cm absorption towards complete samples of radio galaxies. Predictions that the maximum effect due to obscuration is at z'" 1, (Fall & Pei 1995) makes such a search specially interesting.

Acknowledgments. These observations would not have been possible without the substantial and enthusiastic support of the WSRT staff, in particular A. J. Boonstra, A. Bos, J. Bregman, H. Butcher, H. v. Someren Greve and the telescope operators. We are also grateful for software and insightful comments from F. Briggs.

References

Carilli C. L. et al. 1996, J.Atsrophys. and Astro., Supplement, 16, 163. Fall M. S. & Pei, Y. C. 1995, in QSO Absorption Lines, ed. G. Meylan, Springer Heisler J. & Ostriker J. P. 1988, Ap.J. 332, 543 Lacy, M. et al. 1994, M.N.R.A.S. 271, 540 Lanzetta K. M., Wolfe A. M., Turnshek D. A., Lu L., McMahon R. G. & Hazard C. 1991,

Ap.J.S. 77, 1 Pei Y. C., Fall, S. M., & Bechtold J. 1991, Ap. J. 378, 6 Shaver P. A. 1995, in 17th Texas Symposium on Relativistic Astrophysics & Cosmology Steidel C. C., Bowen D. V., Blades J. C. & Dickinson M 1995, Ap. J. 440, L45 White R. 1., Kinney, A. L. & Becker R. H. 1993, Ap.J. 407,456

TESTING Z ~ 0 ANALOGS FOR THE DAMPED LYMAN ALPHA ABSORBERS

The H I content and distribution in the local universe

F.R. BRIGGS Kapteyn Astronomical Institute Postbus 800, 9700 A V Groningen, The Netherlands

AND

E. SORAR Department of Physics & Astronomy University of Pittsburgh, Pittsburgh, PA 15260

1. Introduction

One of the striking findings of QSO absorption-line studies is that there appears to be substantially more neutral hydrogen at redshift z ~ 2.5 than there is at present in nearby galaxies. This result comes from the statistics of the "damped Lyman a" (DLa) class of absorption system (Wolfe et al. 1986, Lanzetta et al. 1995). Pushing the observations and analysis to redshifts greater than 4, Storrie-Lombardi et al. (1995) find evidence for a maximum in the integral H I content, nHI, at z ~ 2 to 3 (see review in this volume by Norman & Braun 1996). In addition to uncertainty in the conclusions due to small number statistics, uncertainties may enter the interpretation due to possible biases and selection effects. Fall and Pei (1993) have argued that dust in galaxies will skew the statistical arguments against inclusion of dusty lines of sight through spiral galaxies, causing the DLa statistics to under-estimate the true H I content. Gravitational lensing may also influence the QSO absorption statistics (Thomas & Webster 1990, Bartelmann & Loeb 1995, Smette et al. 1995), since the amplification provided by the intervening galaxy may preferentially bring dim background QSOs into the sample selected for spectroscopic observation, thus overemphasizing the DLa systems. On the other hand, the bending of the light path may cause the line of sight to "by-pass" (Smette et al. 1995) the highest column density regions within the main body of spiral galaxy progenitors, thus underestimating the quantity of neutral gas. Clearly, there

285


286 F.R. BRIGGS AND E. SORAR

are a number of competing effects that will need to be evaluated before a clear picture emerges.

At low redshift the H I content must be evaluated differently. Pushing the QSO absorption-line techniques to low z has proved difficult, in part because the Lyman-a line is not redshifted to optical wavelengths and must be studied in the UV from space. The IUE and HST studies conducted so far (Lanzetta et al. 1995, Rao et al. 1995) are consistent with a smooth decline in nH I from z ~ 1. 7 to the present, but construction of an unbiased catalog of QSOs for studying lines of sight through very low z, nearby galaxies is difficult, since QSOs that lie within the boundaries of an optical image are less likely to be identified than isolated QSOs. Fortunately, a great deal is already known about neutral gas in nearby galaxies through observations in the 21cm line, and several authors have evaluated nH I of nearby galaxies with an eye toward comparing the nHI(Z ~ 0) with the values measured at higher redshift QSO absorption-line techniques (Fall & Pei 1989, Rao & Briggs 1993). There is agreement that nHI(Z ~ 0) ~ 0.0004 for Ho = 50 km s-l Mpc- 1 , but a major concern is whether the value for the local H I content is complete when it is computed for galaxies that are selected and catalogued optically.

The H I content nH I inferred for Z ~ 3 is five to ten times that at z ~ 0, with much of the spread resulting from choice of cosmological model. If five times the H I identified with optically selected galaxies was missing from the local census, then no evolution would be required from Z ~ 3 to the present, and the apparent drop registered by the space-based, DLa statistics might be instead be the result of selection effects caused by dust and the "by-pass effect ."

2. The H I Line Strip Surveys

In order to measure directly the H I content of the nearby Universe, we have performed blind surveys in the H I line to select gas-rich galaxies without regards to their optical properties. The details of the project are given by Sorar (1994).

The surveys were made with the Arecibo 305 m Telescope and were designed to take advantage of periods during the current upgrade when the telescope would be immobilized. Thus, the data was taken in a driftscan mode, and the telescope beam traced a strip of constant declination in the sky. The same strip was retraced on many days, allowing an integration time of more than five minutes to be accumulated for each beam area along the strip. For the principal survey strips, the redshift coverage was from -400 to 7500kms-1 and was sensitive at the 5a level to rv2x105h-2 M8 at 3 h-1 Mpc, to rv2x 106h- 2 M8 at 10 h-1 Mpc, and to rv2x 108h- 2 M8 at

H I CONTENT OF THE LOCAL UNIVERSE

Current Dec.=23d09m

7h

••

.... ---•

. . " " " .'

\"

287

llh .'.:/: ''.~

\ Ih

5h

• Cataloged Gal. .. Uncataloged Gal.

Current Dec.= 14d14m

.... /. :,-.

/ . .'.

/. / / ' .. " .. /> ~:/~~ ';~.-I ;1<..... .~;

I / ;.' / •

/ .1

"..,- .. : "---..:-.1 ...... /. "

...... '- . ~ . .~ .' . ., . \

/ / .', " I vj:~ \

./ \ . A--

I . I'~~ "

• Cataloged Gal. .. UncatEiloged Gal.

~

o o o

g o

\ \

\"

C]\ o o o

...J o o o

~

IBh

~ ~ ~ -til -til -'" -III Figure 1. Slice diagrams comparing the locations of the H I selected galaxies with galaxies in the efA catalog (Geller & Huchra 1989). In order to include a sufficient number of optical galaxies to define the large scale structure, all galaxies within ±4 ° of declination of the survey strip are plotted. The strips intersect the Galactic plane at RA ~ 6h (Dec. 23°) and RA ~ 19h (Dec. 14°).

288 F.H. BRIGGS AND E. SORAR

75 h-1 Mpc. The first sidelobe of the telescope beam pattern could detect ",2x109 h-2 Me;) at the full depth of the survey. The H I surface brightness sensitivity was ",1018 cm- 2 (50') for gas filling the telescope beam, which su btends ",3 kpc at 3 M pc and 70 kpc at 75 M pc.

The principal survey strips yielded a total of 61 detected signals. Of these, 29 could be associated with catalogued galaxies in the NASA Extragalactic Database (NED). Another 18 were readily identified on the old PSS prints. Nine of the remaining detections fall at low Galactic latitude where extinction is strong; the last 5 are so dim that deeper optical observations are required, but these same 5 objects are among the lowest in H I mass. The bulk of the uncatalogued detections are sufficiently distant that optical catalogs are expected to be incomplete. There is a strong correlation between H I flux and optical brightness, implying that optically selected catalogs will also constitute the bulk of the local H I content. The observation did detect all galaxies with previously measured redshifts listed in NED with coordinate within the main beam along the survey strip. Approximately one half the H I-selected galaxies were detected by the sidelobe.

The locations of the H I selected objects are indicated on the "slice diagrams" in Fig. 1, where their positions are compared with optically catalogued galaxies. There is a strong tendency for the H I selected systems to fall in the same groupings as the optical galaxies. This result is consistent with findings that (1) low surface brightness galaxies and gas-rich dwarfs lie on structures delineated by high surface brightness, normal galaxies (Bothun et al. 1986, Timan et al. 1987, Eder it et aI, 1989, Binggeli et al. 1990, Maia et al. 1993) and that (2) H I selected galaxies are found in selected fields with optical galaxies but not in selected void fields (Weinberg et al. 1991, Szomoru et al. 1994).

A less intensive, blind survey was conducted over the range 19,000 to 28,000 km s-l. Approximately 104h-3 Mpc3 were searched with sensitivity to objects with MHI > 1Q1°h-2 Me;); four objects were found in the range 0.6 to 1.7x1010h-2 Me;). The survey was also sensitive to objects MHI > 1011 Me;) in a volume of ",105 Mpc; none were found.

3. Integral H I Content of Local Universe

The distribution of the integral H I content of the local Universe among objects of different H I mass is summarized in Fig. 2. Although no objects with H I mass below 107h-2 or greater than 2x 101Oh-2 Me;) were detected, the surveys do place constraints on the integral mass content that objects in these ranges can contain. The surveys confirm that the integral mass content is dominated by the gas-rich segment of the optically-identified galaxy population and that f2HI(Z ~ 0) must be significantly less than the

H I CONTENT OF THE LOCAL UNIVERSE 289

H' II '<I 7 ~ .. II 1>0

'L I 1>0

;:JI I ..

1 II I 1>0

Lrr ::I 6.5 ]

r ~ .. .2

6

6 7 8 9 10 11 1°1(".)

Figure 2. The integral mass content of the local Universe as a function of object mass. Two histograms are plotted to indicate two alternate approaches to computing the object masses from the survey data. The upper limits are set to represent 95% confidence limits.

nHI determined from high z QSO absorption-line statistics.

Acknowledgements The National Astronomy and Ionosphere Center is operated by Cornell University under contract with the National Science Foundation. This research has made use of the N ASAjIPAC Extragalactic Database (NED), which is operated by the Jet Propulsion Laboratory, Cal tech, under contract with the National Aeronautics And Space Administration. This work has been supported by NSF Grant AST 91-19930 and NSF Grant AST 88-2222.

References

Bartelmann, M., & Loeb, A. 1995, submitted to Ap.J. Binggeli, B., Tarenghi, M., Sandage, A. 1990, A&Ap, 228, 42. Bothun, G.D., Beers, T.C., Mould, J.R., Huchra. J.P. 1986, ApJ, 308, 510. Eder, J., Schombert, J., Dekel, A., Oemler, A. 1989, ApJ, 340, 29. Fall, S.M., & Pei, Y.C. 1989, Ap.J., 337,7. Fall, S.M., & Pei, Y.C. 1993, Ap.J., 402,479. Geller, M.J., & Huchra, J.P. 1989, Science, 246, 897. Lanzetta, K.M., Wolfe, A.M., Turnshek, D.A. 1995, Ap.J., 440, 435.

290 F.H. BRIGGS AND E. SORAR

Mala, M.A.G., Da Costa, N.L., Giovanelli, R., Haynes, M.P. 1993, AJ, 105, 2107. Norman, C., & Braun, R. 1996, these proceedings. Rao, S., & Briggs, F.H. 1993, Ap.J., 419,515. Rao, S.M., Turnshek, D.A., Briggs, F.H. 1995, Ap.J., 449, 488. Smette, A., Claeskens, J.-F., Surdej, J. 1995 in Astrophysical Applications of Gravita

tional Lensing, Proc. ofIAU Symp. 173, eds. C.S. Kochanek and J.N. Hewit Kluwer Academic Publ), p 99.

Sorar, E. 1994, Ph.D. Thesis, University of Pittsburgh. Storrie-Lombardi, 1.J., McMahon, R.G., Irwin, M.J., Hazard, C. 1995, to appear in "ESO

Workshop on QSO Absorption Lines." Szomoru, A., Guhathakurta, P., van Gorkom, J.H., Knapen, J.H., Weinberg, D.H.,

Fruchter, A.S. 1994, Ap.J., 372, L13. Thomas, P.A., & Webster, R.L. 1990, Ap.J., 339, 437. Wolfe, A.M., Turnshek, D.A., Smith, H.E., & Cohen, R.D. 1986, Ap.J.S., 61, 249. Weinberg, D.H., Szomoru, A., Guhathakurta, P., van Gorkom, J.H. 1991, Ap.J., 372,

L13.

INTERSTELLAR MEDIUM IN DISTANT GALAXIES

MOLECULAR GAS IN HIGH REDSHIFT GALAXIES

SIMON J. E. RADFORD

National Radio Astronomy Observatory, Tucson, AZ, USA

1. Introduction

Study of molecular gas in distant galaxies during the last twenty years has followed the steady progress in mm wave receiver sensitivity. In 1975, CO was detected in M 82, NGC 253, and several other galaxies with redshifts of a few hundred km S-1 (Rickard et al. 1975; Solomon & de Zafra 1975). Over the next fifteen years, the CO detection horizon increased steadily, reaching z ~ 0.22 by 1990 (Downes et al. 1991). The discovery that the large population of infrared luminous galaxies detected by IRAS are very gas rich (e. g., Sanders, Scoville, & Soifer 1991) was especially significant. In the last few years, there has been a breakthrough; CO has been observed in two high redshift objects, IRAS FSC 10214+4724 at z = 2.3 (Brown & Vanden Bout 1992b) and the Cloverleaf quasar (H 1413+ 117) at z = 2.6 (Barvainis et al. 1994; Barvainis 1996). These objects offer glimpses of galaxies' properties when the Universe was only about 15% of its present age. The presence and conditions of molecular gas in galaxies at such an early epoch are clues to understanding galaxy formation and evolution in the early Universe. Despite much observational effort, however, no other high redshift sources have confirmed detections of CO. Indeed molecular gas in both 10214+4724 and the Cloverleaf is visible only because they are gravitationally lensed. In intrinsic molecular content, gas distribution, and IR luminosity, these galaxies resemble gas rich, ultraluminous IR galaxies in the local universe. They may be powered by ongoing bursts of massive star formation in the gas rich interstellar medium or by active nuclei fueled by accretion of the gas. Whatever the power source, substantial processing by massive stars must have already taken place to create the observed heavy elements, dust, and molecular gas. These galaxies indicate the existence of a high redshift parent population similar to nearby ultraluminous galaxies.

293


294 SIMON J. E. RADFORD

2. IRAS FSC 10214+4724

2.1. APPARENT PROPERTIES

The apparent luminosity of 10214+4724, about1 1014h-2 L0 emitted primarily in the far infrared (Rowan-Robinson et ale 1991, 1993), places it among the most luminous objects known. It exceeds the quasar luminosity threshold by two orders of magnitude. Although 10214+4724 is not strictly speaking a primordial galaxy, because there has been sufficient processing of material through stars to produce metals, it obviously has not had much time to convert its gas to stars. It clearly formed at least a few X 107 yr, and more probably> 108 yr, earlier.

To date, CO in 10214+4724 has been detected with six different telescopes. Since the first observations (Brown & Vanden Bout 1991), however, the CO line flux, precise redshift, and source size have all been disputed. Follow up observations with the IRAM and Nobeyama telescopes (Fig. 1) have all indicated a much smaller line flux, 4 ± 1 Jykms-t, than first measured with the NRAO 12m telescope, 21 ± 5Jykms-1 (Brown & Vanden Bout 1991). The lower measurements imply an apparent mass M(H2 ) ~ 1011 h-2 M0 (Solomon, Downes, & Radford 1992), as much gas as the total mass of stars in a large spiral galaxy or the core of a giant elliptical galaxy. The discrepancy with the original measurement led to suggestions (Brown & Vanden Bout 1992b; Sakamoto et ale 1992; Tsuboi & Nakai 1992) that the molecular gas is extended over a diameter as large as 60" (240h- 1 kpc) with an apparent molecular mass of about 1012 M0 . With smaller beams, the large telescopes and interferometers would see, therefore, only the central peak, while the smaller telescope would detect the more extended component. Although it cannot be ruled out a priori, such a large source seems unlikely since only a few molecular clouds have been observed > 20 kpc from the center of the Milky Way and CO has never been detected> 100 kpc from the center of any other galaxy.

To resolve the observational quandry, then, we reobserved CO(3 ~ 2) from 10214+4724 with the NRAO 12m telescope (Radford et ale 1996). We found an integrated line flux of 6.7 ± 1.4kms-1 • This is 3 ± 1 times smaller than the first measurement and is consistent, within its uncertainties, with observations at other telescopes. No evidence remains for an extended source larger than a couple of arcseconds. Molecular gas in 10214+4724 is concentrated in a small central region.

In high redshift objects, submm lines usually difficult to study can be observed at mm wavelengths through the standard atmospheric windows. In 10214+4724, the CO(3~2), (4~3) (Brown & Vanden Bout 1992b), (6--?5)

1 Ho = lOOh km 8-1 Mpc-1 , qo = 0.5

MOLECULAR GAS IN HIGH REDSHIFT GALAXIES 295

20

20

IRAS 10214+4724 CO(3~2) Observed Frequency [GHz]

105.3 105.1 105.3 105.1

12 m (1991)

O~-----~~~-+-

~ 20 ! s

45 m

......... 0 1------iIiF----H-t:---_i_ ... '-"'ltfl--t-----H--+ fI)

20

OHr-----+;-+P~~

20

-500 o +500 -500 0 Velocity Offset [km s -1]

= 2.286

12 m (1993)

+500

2.9

2.3

1.7

1.1 ,........, ...... N

0.6 s .........

o ~ * ,........,

S -0.6e

1.3

0.6

o

Figure 1. Spectra of CO(3-+2) emission at z = 2.286 from IRAS 10214+4724 observed with the NRAO 12 m telescope at 16 MHz resolution in 1991 (Brown & Vanden Bout 1991) and in 1993 (Radford et al. 1996), the lRAM 30 m telescope at 16 MHz resolution (upper: Brown & Vanden Bout 1992bj lower: Solomon et al. 1992), the Nobeya.ma 45 m telescope at 10.5 MHz resolution (Tsuboi & Nakai 1992), the Nobeyama Millimeter Array at 32.5 MHz resolution (Kawabe et a.l. 1992j Sa.ka.moto et al. 1992), and the IRAM interferometer (Bure) at 50 MHz resolution (Radford et al. 1993). The ±lu error bars represent the per channel uncertainty.

(Solomon, Downes, & Radford 1992), and, tentatively, (7 -+6) (Solomon private communication) lines have all been detected. The line ratios are consistent with a single component LVG model that indicates the gas is warmer, Tkin ::::J 5DK, and denser, n(H2) ::::J 5DDDcm-3 , than the bulk of


the gas in the Milky Way (Solomon, Downes, & Radford 1992), where CO(6-5) is only observed in molecular cloud cores near sites of massive starformation (Jaffe et al. 1989). An upper limit to the CO(I-0) line flux (Barvainis 1995) is also consistent with this excitation model.

Searches for C IePI _3PO) and ep2 _3PI) emission from 10214+4724 have been made, but a claimed detection of C Iep2- 3PI) (Brown & Vanden Bout 1992a) remains unconfirmed and controversial. In the Cloverleaf quasar, on the other hand, C IePI _3PO) has been detected (Barvainis 1996) with a C I/CO level similar to the nearby galaxy IC 342 (Biittgenbach et al. 1992).

The observed mm and submm spectral energy distribution peaks near 150 pm, corresponding to a rest frame peak of 46 pm and a dust temperature of about 80 K. In addition, the 60/Lm flux observed by IRAS corresponds to 18/Lm in the rest frame and indicates the presence of hotter dust at about 200 K. These components each have an apparent luminosity of 4 X lOI3h-2 L0 . The apparent gas-to-dust ratio, inferred from the CO line flux and the 350/Lm (rest frame) continuum flux, is about 500, similar to the Milky Way and to nearby ultraluminous galaxies. This suggests the metal abundance is already approximately solar (Downes et al. 1992).

2.2. GRAVITATIONAL LENS AND INSTRINSIC PROPERTIES

Optical and infrared images suggest 10214+4724 is magnified 5-50 times by an intervening gravitational lens (Matthews et al. 1994; Elston et ai. 1994; Januzzi et ai. 1994; Broadhurst & Lehar 1995; Graham & Liu 1995; Eisenhardt et al. 1996). This means, of course, the galaxy's intrinsic luminosity and mass of H2 are proportionally smaller, but it still ranks among the most luminous and gas rich IR galaxies (Downes, Solomon, & Radford 1995). There is still copious molecular gas, comparable to that in an ultraluminous IR galaxy, to fuel vigorous star formation.

The 2.2/Lm (Matthews et at. 1994; Graham & Liu 1995) and red HST (Eisenhardt et al. 1996) images of 10214+4724 show a 1.5" long arc about 1.2" south of the intervening galaxy. At the center of the arc is a bright, compact, 0.7" diameter core that has a faint counterimage just north of the intervening galaxy. This apparent morphology suggests the intrinsic source has at least two components: a compact source almost coincident with a cusp of the lens caustic and an extended envelope or disk that appears as the extended arc (Broadhurst & Lehar 1995). Because the lens magnification depends on the intrinsic source size, the lens may be chromatic since different spectral components are emitted by sources with different intrinsic sizes. Differential magnification may alter, then, the observed spectrum so it no longer accurately represents the intrinsic spectrum.


Current mm wave interferometers do not have the sub arcsecond resolution available in the near IR and optical and necessary to see details of the CO distribution in 10214+4724. Nevertheless, it is possible to gauge its overall extent.

In a CO(3-2) image made with the IRAM interferometer, the source is clearly more extended east-west than the beam, but it is no wider northsouth. Convolving the 2.3" x 1.6" synthesized beam with the 15% contour of the 2.2 I'm arc reproduces the observed CO distribution well (Downes, Solomon, & Radford 1995). Convolving the beam with a small compact source, such as the bright, compact 2.2 I'm core, on the other hand, produces a much more condensed distribution than observed in the CO image.

A CO(6-5) image made with the OVRO interferometer (Fig. 2) offers better resolution and shows somewhat more directly the east-west extent of the CO distribution. Regardless of the weighting of the visibility data, natural or uniform, this image indicates the apparent CO distribution is 1.5" long. We identify the CO distribution with the extended arc in the 2.2 I'm image and, hence, with an extended part of the intrinsic source.

An upper limit to the magnification of the CO image can be determined because the gravitational lens stretches the source image in one dimension, preserving surface brightness, and we can estimate the intrinsic CO brightness temperature from the line ratios (Downes, Solomon, & Radford 1995). The observed CO(3 - 2) luminosity Leo = 2.6 x 101Oh-2 K km s-1 pc2 • For the smallest possible intrinsic CO distribution, an optically thick thermal source, Leo = meo1rr2Tb~V, where r is the intrinsic source radius, n is the rest frame brightness temperature of the line, ~v = 220 ± 30 km S-1 is the linewidth, and meo is the source magnification. Since the magnification is one dimensional, meo = a/ r, where 2a is the apparent extent of the CO distribution. The observed CO line ratios clearly indicate the molecular gas is warm. If the gas kinetic temperature is 60 K, slightly cooler than the 80 K dust, an LVG excitation model that fits the observed CO line intensities and ratios indicates the CO(3-2) and (6-5) brightness temperatures are 43 ± 7 and 27 ± 5 K, respectively, and the opacities are 6, 37, and 41 for the C0(1-0), (3-2), and (6-5) lines. At z = 2.3, the apparent extent ofthe CO distribution 2a = 1.5" x (4h- 1 kpc arcsec-1) = 6h-1 kpc, so meo ~ 10, independent of h.

With this modest lens magnification, 10214+4724 still has a molecular gas content comparable to the most CO rich, IR luminous galaxies (Graham & Liu 1995), but is not extraordinary for that class. The galaxy's intrinsic CO(3 - 2) luminosity is 2.6 X 109h-2 K km s-1 pc2 and the true radius of the CO distribution r ~ 300h-1 pc. For comparison, Arp220 has a C0(1-0) luminosity of 4 x 109h-2 K km s-1 pc2 (Solomon, Radford, & Downes 1990), with 2/3 of that concentrated within a radius of 240h-1 pc (Scoville et al.


IRAS 10214+4724

30" . ,,",

........ "

25"

~ 20" C\2 •• ::0', a l'-.q. ........ 0 l!) en ..... 15" ......-I=l 30" 0 ..... ..., ttl I=l ..... -() (IJ

0

25"

31.5" 31" 30.5" Right Ascension (1950) 10h 21m

Figure 2. Integrated emission maps of CO(6-+5) at z = 2.286 from IRAS 10214+4724 observed with OVRO interferometer at 210GHz. The contour interval is 2mJybeam-1

and the synthesized beam (insert) is 2.7" X 2.1" for natural weighting (top) and 1.4" xLI" for uniform weighting (bottom).. The source is more extended than the beam; the deconvolved source size is about 1.5 ' east-west.


1991). In a sample of 37 ultraluminous IR galaxies out to z = 0.27, the highest C0(1-0) luminosity is 9 x 109 h-2 Kkms-1 pc2 and the median is 5 X 109h-2 K km S-1 pc2 (Solomon et al. 1996).

The far IR magnification can be determined by a similar argument, although in this case we have no direct measurement of the source size (Downes, Solomon, & Radford 1995). In 10214+4724, the apparent far IR luminosity of 4 x 1013h -2 L0 , observed at 450 to 130011m, is emitted by optically thick dust at ~ 80 K (Downes et al. 1992). In nearby, non-lensed ultraluminous galaxies, the far IR source is 0.6-1.0 times the the size of the CO source. This suggests the 80K dust source in 10214+4724 has a radius of 200-300h- 1 pc, the far IR magnification is 10-13, and the intrinsic far IR luminosity is 3-4 X 1012h-2 L0 . Again, this is similar to nearby ultraluminous galaxies, albeit at the high end of the distribution. Arp 220, for example, has a far IR luminosity of 6 x 101lh-2 L0 . In our sample of 37 galaxies, the most luminous has 2 X 1012h-2 L0 . Since the CO and far IR have similar magnifications, their ratio, which indicates the gas-to-dust ratio and the metal abundance, is largely unaffected.

The mid IR radiation, on the other hand, is emitted by a hotter, 200 K source that may be much smaller. If it corresponds to the compact, 0.7" diameter core in the 2.211m images, it may be magnified 50 times (Broadhurst & Lehar 1995). Then the intrinsic mid IR luminosity would be 8 x 101lh-2 L0 and the true radius of the core would be 30h-1 pc. This is a typical radius for an AGN Narrow Line Region (NLR), but is much too small to account for the thermal far IR emission at 80 K. Hence the far IR dominates the intrinsic luminosity of 10214+4724 as it does in nearby ultraluminous galaxies.

Although 10214+4724 has a Seyfert 2 spectrum lacking the usual signatures of star formation (Rowan-Robinson et at. 1991, 1993; Lawrence et at. 1993, Elston et al. 1994; Januzzi et al. 1994; Soifer et al. 1995; Goodrich et al. 1996), this may be an artifact of differential magnification by the intervening lens rather than an intrinsic property. If the galaxy harbors both an extended starburst and an AGN, the small nucleus will be magnified much more than the larger region of star formation where the H II regions are. Since the observed optical spectrum is dominated by the compact 0.7" core, which is the magnified image of the nucleus, the AGN characteristics overshadow the starburst signatures.

Even with true bolometric and CO luminosities ten times lower than earlier estimates (Downes et al. 1992; Solomon, Downes, & Radford 1992), the nature ofthe energy source remains a problem. Is 10214+4724 powered by star formation in the molecular region itself, or are the gas and dust just part of a massive envelope heated by the AGN (Sanders et al. 1989)7 The amount of dense molecular gas in ultraluminous IR galaxies indicates


they harbor huge starbursts. To explain even the intrinsic luminosity of 10214+4724 with a starburst, however, time scale constraints (Heckman 1994) imply there must be an IMF of high mass stars only and a high star formation efficiency - 20% of all the gas converted to stars in 107 yr. Nevertheless, in intrinsic molecular content and IR luminosity, 10214+4724 resembles a typical nearby ultraluminous IR galaxy. Whatever its current power source, the heavy elements, dust, and molecular gas must all have been produced by massive stars.

It has been a pleasure to collaborate with Dennis Downes and Phil Solomon. The NRAO is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.

References

Barvainis, R., et al. 1994, Nature 371, 586 Barvainis, R., 1995, AJ 110, 1573 Barvainis, R., 1996, this volume Broadhurst, T., & Lehar, J. 1995, ApJ 450, L41 Brown, R. L., & Vanden Bout, P. A. 1991, AJ 102, 1956 Brown, R. L., & Vanden Bout, P. A. 1992a, ApJ 397, Lll Brown, R. L., & Vanden Bout, P. A. 1992b, ApJ 397, L19 Biittgenbach, T. H., Keene, J., Phillips, T. G., & Walker, C. K. 1992, ApJ 397, L15 Downes D., et al. 1991, in IAU Symp. 146, ed. Combes & Casoli (Kluwer) p.295 Downes, D., et al. 1992, ApJ 398, L25 Downes, D., Solomon, P. M., & Radford, S. J. E. 1995, ApJ 453, L65 Eisenhardt, P. R., et al. 1996, ApJ in press Elston, R., et al. 1994, AJ 107, 910 Goodrich, R. W., et al. 1996, ApJ 456, L9 Graham, J. R., & Liu, M.C. 1995, ApJ 449, L29 Heckman, T. M. 1994, in Mass Transfer ... , ed. Shlossman (Cam. U. P.) p. 234 Jaffe, D. T., et al. 1989, ApJ 344, 265 Jannuzi, B.T., et al. 1994, ApJ 429, L49 Kawabe, R., Sakamoto, K., Ishizuki, S., & Ishiguro, M. 1992, ApJ 397, L23 Lawrence, A., et al. 1993, MNRAS, 260, 28 Matthews, K., et al. 1994, ApJ, 420, L13 Radford, S. J. E., Brown, R. L., & Vanden Bout, P. A. 1993, A&A 271, L71 Radford, S. J. E., et al. 1996, AJ 111, in press Rickard, L. J., et al. 1975, ApJ 199, L75 Rowan-Robinson, M., et al. 1991, Nature 351, 719 Rowan-Robinson, M., et al. 1993, MNRAS, 261, 513 Sakamoto, K., Ishizuki, S., Kawabe, R., & Ishiguro, M. 1992, ApJ 397, L27 Sanders, D. B., et al. 1989, ApJ, 347, 29 Sanders, D. B., Scoville, N. Z., & Soifer, B. T., 1991, ApJ 370, 158 Scoville, N. Z., Sargent, A. I., Sanders, D. B., & Soifer, B. T., 1991, ApJ 366, L5 Solomon, P. M., & de Zafra, R., 1975, ApJ 199, L79 Solomon, P. M., Downes, D., & Radford, S. J. E. 1992, ApJ 398, L29 Solomon, P. M., Downes, D., Radford, S. J. E., & Barrett, J. W., 1996, in preparation Solomon, P. M., Radford, S. J. E., & Downes, D. 1990, ApJ 348, L53 Solomon, P. M., Radford, S. J. E., & Downes, D. 1992, Nature 356, 318 Tsuboi, M., & Nakai, N. 1992, PASJ 44, L241

co, C I, AND (POSSIBLY) HCN IN THE CLOVERLEAF QUASAR

RICHARD BARVAINIS

MIT Haystack Observatory Westford, MA 01886

1. Introduction

The Cloverleaf (H1413+ 117) is a quasar at redshift 2.56 which derives its name from its optical image, which is gravitationally lensed into a pattern of four points with separations of about I". The lensing galaxy is unseen, but models assuming an elliptical potential at intermediate redshift give total magnifications of roughly a factor of 10 (Kayser et al. 1990). The Cloverleaf's optical spectrum has classical broad emission lines, plus broad absorption troughs - it is classified as a BALQ (broad absorption line quasar).

We initially selected the Cloverleaf for molecular line studies because of its strong submillimeter emission, detected during a survey of radio quiet quasars and BALQs at the JCMT (Barvainis, Antonucci, & Coleman 1992). It turns out that the Cloverleaf was also detected by IRAS, and has a far-IR/submm spectrum identical to that of the luminous infrared galaxy IRAS F10214+4724 at z = 2.28 (Barvainis et at. 1995). F10214+4724 was of course the first high-z object to be convincingly detected in CO (Brown & Vanden Bout 1992; Solomon, Downes & Radford 1992), with the Cloverleaf being the second (Barvainis et at. 1994). Several new lines of evidence link F10214+4724 closely with the Cloverleaf: spectropolarimetry of F10214+4724 shows broad optical emission lines in reflected light, meaning that it harbors a hidden quasar (Miller 1995); FI0214+4724 now appears to be gravitationally lensed (e.g., Broadhurst & Lehar 1995); and similar far-IR and CO emission, but different optical emission, in the two objects suggests that they differ primarily in orientation, with F10214+4724's nucleus obscured by a surrounding dusty torus (Barvainis et at. 1995).

301


Figure 1. Four transitions of CO in the Cloverleaf quasar. For ease of comparison, the spectra have been put on a brightness temperature scale (in Kelvins), arbitrarily assuming a source size of 0.3"(::::: 1.5 kpc at the source).

Here I present new millimeter observations of line emission from the Cloverleaf. My collaborators in this work are Ski Antonucci, Danielle Alloin, Linda Tacconi, Paul Coleman, and Phil Maloney.


We observed the Cloverleaf using the IRAM 30 m telescope on Pieo Veleta in June 1994 and May 1995. Because of the unique ability of the 30 m to collect data in the 3 mm, 2 mm, and 1 mm bands simultaneously, we were able to observe a rather large suite of lines. This has resulted in definite detections in CO J = 3-2, 4-3, 5-4, and 7-6 (Fig. 1), and an upper limit on 13CO(3-2) emission. In addition, we obtained what appears to

LINES IN THE CLOVERLEAF QUASAR

TB«(J = 0.3")

10

5

O~+~~~--------------~~-++

-5

-500 o Velocity (km/s)

500

303

Figure 2. Spectrum of the carbon fine-structure line C Ie Pl _ 3 Po). The line was detected on two separate observing runs, and this is the combined result.

15

10

5

TB«(J = 0.3")

o

-5

-10

,

rl ) L

U

I

-500

s

Lr-~

o Velocity (km/s)

L

500

r-

Figure 3. Spectrum of HCN(4-3) in the Cloverleaf. We consider the apparent detection to be likely but not ironclad.

304 RICHARD BARVAINIS

be a solid detection of the atomic carbon fine structure line C Ie p1 _3 Po) (Fig. 2), and an upper limit on CIep2 -3pt). We also have a possible detection of HCN( 4 - 3) at the T1 = 1 mK level (Fig. 3), which required some 7 nights of integration. Finally, we have an upper limit on CO(1-0), at Vobs = 32.4 GHz, using the Haystack 37m telescope in January 1995. This limit is consistent with CO(1-0) being no more than about a factor of two brighter than the thermalized value relative to CO(3-2).

These results will be discussed in detail in a future paper (Barvainis et al., in prep). A few general statements can be made here. First, the CO excitation is interesting, in that CO( 4-3) appears to have a higher brightness temperature than either CO(3-2) or CO(5-4). This probably means that conditions are such that the critical density for the CO( 4-3) transition obtains, and that the lines are not too optically thick. Second, the (possible) detection of H CN means that there is probably a large quantity of high density gas (n rv 105 cm-3 ) present in the Cloverleaf. The HCN/CO brightness temperature ratio is 0.2, very much like the values found in luminous IRAS galaxies. Finally, the CI/CO ratio is also about 0.2, which is similar to the value in IC 342 (the only other extragalactic source where both species have been clearly measured).

3. Conclusion

Although there have been many attempts to detect high-z CO emission, only the Cloverleaf and F10214+4724 have been definitely confirmed. It appears that the boosting derived from lensing, uncertain but perhaps an order of magnitude, is ~ssential for studying gas in the early universe, given current instrumentation. Therefore we need to derive as much information as we can from molecular line studies of the two objects where such studies are now possible. Detailed modeling of the various mm lines are providing constraints on physical conditions in the ISM of the Cloverleaf quasar, at an epoch when the universe was only one-seventh its current age.

References

Barvainis, R., Antonucci, R., & Coleman, P. 1992, ApJ, 399, L19 Barvainis, R., Tacconi, L., Antonucci, R., Alloin, D., & Coleman, P. 1994, Nature, 371,

586 Barvainis, R., Antonucci, R., Hurt, T., Coleman, P., & Reuter, H. 1995, ApJ, 451, L9 Broadhurst, T., & Lehar, J. 1995, ApJ, 450, L41 Brown, R., & Vanden Bout, P. 1992, ApJ, 397, L19 Kayser, R., et al. 1990, ApJ, 364, 15 Miller, J. 1995, talk given at the National Academy of Sciences Colloquium "Quasars

and AGN, High Resolution Imaging," March 24-25, Irvine, CA Solomon, P.M., Downes, D., & Radford, S.J.E. 1992, ApJ, 398, L29

SEARCHING FOR MOLECULAR GAS IN A RADIO GALAXY AT RED SHIFT 3.8

ROBERT J. IVISON Royal Observatory, Blackford Hill, Edinburgh EH9 3HJ, U.K.

PETER PAPADOPOULOS AND ERNEST R. SEAQUIST Department of Astronomy, University of Toronto, 60 St George Street, Toronto M5S 1A 7, Canada

AND

STEVEN A. EALES Department of Physics, University of Wales, College of Cardiff, P.O. Box 913, Cardiff CF4 3TH, U.K.

Abstract. There is reason to suspect that 4C 41.17, a radio galaxy at z = 3.8, contains a large quantity of enriched molecular gas. We have searched for CO in 4C 41.17, taking advantage ofthe fact that the J = 1-0 transition is redshifted into the K radio band for 3.80 < z < 4.25. There is no sign of CO line emission in our resulting spectrum (which covers 2000kms-t, or 3.785 < z < 3.821), nor in the channel maps. Our limit for the integrated line intensity is Sco/).v < 0.17Jykms-t, for a line width of 500kms-1 .

Naively assuming a Galactic conversion factor gives M(H2 ) < 4 X 1011 M0 ,

similar to the stellar mass of a present-day giant galaxy, and consistent with gas-to-dust ratios as high as 4400.

1. Introduction

The number of detections of thermal emission at z > 2 recently reached double figures (Ivison 1995), although only a handful have been independently confirmed by several observers at more than one frequency; CO detections at z > 2 have been less common, and those of FI0214+4724 and the Cloverleaf quasar were probably aided by lensing (e.g. Serjeant et al. 1995).

The dust detections have shown that even at z f'V 4 there must have been sufficient stellar lifecydes to enrich the ISM with metals, so it is perhaps

305


306 ROBERT J. IVISON ET AL.

surprising that CO detections are so rare. Although the CO flux falls rapidly between z = 0 and 1, thereafter it varies quite slowly; add to this the optical thickness of the sub millimetre lines, and CO should become a potent probe of molecular gas in very distant galaxies.

It is natural to use high redshift radio galaxies as probes of the early stages of galaxy formation. The low redshift examples are a fairly homogeneous group, in the sense that their stellar populations are relatively coeval, so comparing high and low redshift radio galaxies may give good insights into the process of galaxy formation. An effective method of determining the evolutionary state of a radio galaxy might be to measure its gas mass and compare it with that of older examples in the near Universe, which are known to contain around 1012 M0 of stars.

For 3.80 < z < 4.25, CO(1-0) emission falls in the K radio band (e.g. Barvainis & Antonucci 1996). If Tb(4-3)/Tb(1-0) is around 0.4 (as observed towards IRAS F10214+4 724) then the detection capability of CO( 1-0) measurements with the VLA is similar to that of the current generation of millimetre arrays working on CO( 4-3) measurements. Even after taking into account the overheads associated with building up adequate velocity coverage, the VLA's low Tsys and 13,000m2 collecting area make it extremely competitive.

4C 41.17 is a distant radio galaxy where there is strong evidence for vast quantities of dust-rich gas (Dunlop et al. 1994). Here, we describe a search for CO(1-0) from 4C41.17, using the VLA.

2. Data and Results

We used the VLA in its C configuration, which matched the size of the synthesized beam ('" 1 arcsec) to the minor axis of the suspected dust lane.

Although the redshift of 4C 41.17 appears to have been well determined (z = 3.800 ± 0.003, Chambers et at. 1990), redshifts as low as 3.794 have been reported. It is vital to obtain the widest possible velocity coverage: because of the aforementioned inconsistency; because of possible offsets between the high excitation lines used to determine z and the systemic velocity; and because of the unknown and conceivably large CO linewidth.

We observed six overlapping 50 MHz IFs, giving a total velocity coverage of 2000 km s-1 (or a total redshift coverage of z = 3.803 ± 0.018). We recorded seven 6.25 MHz channels (78 km S-1 at z = 3.8) for each IF band, with around fifty 9 min scans for each IF pair. Several advantages accrued from this approach: no band was assigned better weather than any other, and the uv coverage for each IF was virtually identical.

SEARCHING FOR MOLECULAR GAS AT REDSHIFT 3.8 307

41 34 14

12

10

0- 08 1ft

'" e 06 z ()

0

fi 04 0 z :::i o u 02 w 0 c

00

33 58 • @;

o a

56

06 47 21.5 21.0 20.5 20.0 RIGHT ASCENSION (81950)

Figure 1. 24.0GHz continuum map of 4C41.17. The visibilities were given natural weightings, resulting in a symmetrical 1 arcsec FWHM beam and an rms of 5511Jy beam-to Contours are at -3, 3,6, 9, 15, 21, 27 and 36 x 50I1Jy.

3. Results and discussion

In Fig. 1, the core of the galaxy dominates, with a peak 24.0 GHz brightness of 2.03 ± 0.06mJybm-1 and a total flux density of 2.72 ± 0.15mJy.

Also visible in Fig. 1 is a weaker feature exactly coincident with feature 'A' in the low frequency maps of Chambers et al. (1990) and Carilli, Owen & Harris (1994), and 1 arcsec SW of a continuum feature in Keck 2pm images (Graham et al. 1994). Its peak is around 9 times above the noise level of the map; we estimate its integrated flux density to be 0.7±0.lmJy, and its position (BI950) to be a = 06h 47m 20.5 00, 0 = +410 33' 59.3".

We have estimated the total flux densities of feature 'A' at 0.33, 1.45, 4.54,8.09 and 14.9 GHz from the Carilli et al. maps and, together with our 24.0 GHz datum, they fit a tight power law with (a) ~ -1.7, as expected for optically thin synchrotron emission. For the core, the spectral index shows a larger dispersion, possibly due to more than one prominent synchrotron component; nevertheless, it is quite steep (a 2: -1). We can see in Fig. 2 that the observed 24.0 GHz continuum flux density agrees, more or less, with an extrapolation of the steep spectrum observed at lower frequencies, and gives additional support to the idea that the emission observed at 800 pm originates from ~ 3 X 108 M0 of dust (Dunlop et al. 1994).

308 ROBERT J. IVISON ET AL.

Rest wavelength fcrn 101 10'

1019 II' . • ,

~ loll r T • ~ • ,e.l017 r t • • 1ij Q

.9 16 Eho ••• , = ~ • CORE •• 10%5 • LOBE 1 .. _. ,.

10.1 10.1 10°

Observed frequency fGHz Figure 2. SED of the 4C41.17 core (diamonds), and offeature 'A' (circles).

,.-., 3.0

i '-' 2.5

.e-.; 2.0

! ~ 1.5

1.0

Velocity for z=3.800 (kmls) Figure 3. Spectrum of 4C 41.17, centred in velocity at the expected position of the CO(1 ...... 0) line for z = 3.800.

One-, two- and three-channel maps were created from the spectral line databases; in the former case, the rms was 0.2-0.3mJy beam-I, consistent with theoretical predictions.

The maps were scrutinized for blobs of emission covering more than one channel, as might be expected for gas-rich systems comoving with 4C 41.17, and for evidence of extended zones of emission near the radio galaxy's core. Neither was evident. The spectrum is shown in Fig. 3. The spectral mean

SEARCHING FOR MOLECULAR GAS AT RED SHIFT 3.8 309

after averaging the overlap regions is 2.10 ± 0.28mJy, consistent with the continuum image. Although the spectrum is marginally affected by baseline offsets, the dominant source of noise is thermal.

For a line width of 500 km s-t, we derive a 30' upper limit of Scofl.v < 0.17 Jy km S-I, which is consistent with the prediction ofthe Scofl.v/ SlOOp.m

= 3.5 km S-1 correlation for ultraluminous galaxies (S.~=94p.m < 56 mJy from Dunlop et al., hence Scofl.v < 0.20Jy kms- l ).

The H2 mass is thus limited to ~ 4 X 1011 M0' comparable to the stellar mass of a present-day giant galaxy; for a gas-to-dust ratio of 500, it is consistent with the estimated dust mass of 3 X 108 M0 in 4C 41.17.

4. Concluding remarks

The low noise achieved here, and the successful knitting together of six IF bands, show that the VLA is quite capable of seeking out CO at high redshift; furthermore, once CO is found, it has the potential to obtain maps with subarcsec (kpc-scale) resolution.

With a high bandwidth correlator, with more Q-band receivers, and with more sensitive K-band receivers, the VLA will hold the key to detecting and mapping C0(1-0) and CO(2-1) in even low mass galaxies.

More promising objects for VLA observations in the near future are galaxies with M(H2 ) '" 1012 M0' (roughly the total mass of a giant elliptical). Such an a quantity of gas has been detected in the case ofthe damped Ly Ci system at z = 3.1 towards PC 1643+4631A, which gives us hope that some objects, possibly the progenitors of present-day giant ellipticals, may lie within the VLA's detection (and imaging) capabilities.

Acknowledgments. This work was supported by an operating grant to ERS from the Natural Sciences and Engineering Research Council of Canada. NRAO is operated by Associated Universities Inc., under a cooperative agreement with the National Science Foundation.

References

Carilli, C.L., Owen, F.N. and Harris, D.E. (1994), AJ, 107, 480 Barvainis R. and Antonucci R. (1996), PASP, in press Chambers, K.C., Miley, G.K. and van Breugel, W.J.M. (1990), ApJ, 363, 21 Dunlop, J.S., Hughes, D.H., Rawlings, S., Eales, S.A. and Ward, M.J. (1994), Nat, 370,

347 Graham, J.R., Matthews, K., Soifer, B.T., Nelson, J.E., Harrison, W., Jernigan, J.G.,

Lin, S., Neugebauer, G., Smith, G. and Ziomkowski, C. (1994), ApJ, 420, L5 lvison, R.J. (1995), MNRAS, 275, L33 Serjeant, S., Lacy, M., Rawlings, S., King, L.J. and Clements, D.L. (1995), MNRAS, 276,

L31

THERMAL EMISSION FROM DUST IN HIGH-Z GALAXIES

DAVID H. HUGHES Astrophysics, Oxford University, U.K.

Abstract. A measure of the FIR luminosity in galaxies, assuming optically thin re-radiated emission from dust grains heated by young massive stars (and no contribution from an AGN), provides an estimate of their current starformation rates and dust masses. The fundamental reason for making millimetre and submillimetre continuum observations ofhigh-z (z '" 2 ---+ 5) galaxies, which measure the rest-frame FIR emission, is to determine their evolutionary status. This is achieved in a relatively model-independent way by inferring the molecular gas mass, through a measure of dust mass and assuming a gas-to-dust ratio, available for further starformation and to compare this mass with the expected baryonic mass of a present-day counterpart. If this mass fraction is ~ 0.9 then, together with the extreme FIR luminosities (LFIR ~ 1013 L0 ) implying starformation rates ~ 1000M0 yr-1,

there is persuasive evidence that high-z galaxies are seen early in their evolution, whilst undergoing a significant (and possibly first) burst of massive starformation.

1. Introduction

In this paper I shall review the recent submillimetre and millimetre wavelength continuum observations towards high-redshift radio galaxies and quasars. It is important to point out the uncertainties in the physical parameters (e.g. dust grain models, dust temperature, FIR opacity, gas-todust ratio) that currently hinder our ability to unambiguously interpret the available observational data in the context of galaxy formation and their subsequent evolution. The major purpose of this paper is to summarise the detections of all high-redshift (z > 2) sources at wavelengths between 350 pm - 1300 pm, and describe the methods used to determine their star-

311


312 DAVID H. HUGHES

formation rates (SFRs), dust and gas masses, and hence place constraints on their evolutionary status. A few comments will be made regarding the future opportunities to make sensitive rest-frame FIR-submillimetre continuum observations of high-redshift galaxies using new ground-based and space-borne instrumentation.

As galaxy formation models develop it becomes increasingly unclear at what epoch we can describe a high-redshift galaxy as primceval. For latetype galaxies whose spectral evolution is well described by a gently declining or even constant star-formation rate (Guiderdoni & Rocca-Volmerange 1987) attempting to identify a unique epoch of formation may be meaningless, and the progenitors of normal discs may have already been found in the form of damped Lya absorbers (Wolfe 1993, Fall & Pei 1995 and references therein).

The properties of present-day elliptical galaxies are harder to interpret, namely low molecular gas and dust masses, < 108 M0 (Lees et al. 1991, Knapp & Patten 1991, Wiklind et al. 1995), enormous stellar masses ,...., 1011 _1012 M0 (Sandage 1972), uniform optical-IR colours that are dominated by a well-evolved stellar population, but which also require a bluer population of intermediate-age stars, 0.1 - 1 Gyr (O'Connel 1987), ongoing star-formation, albeit at a low-level (e.g. Keel & Windhorst 1991, Mazzei & de Zotti 1994), and counter-rotating cores (Kormendy 1984, de Zeeuw & Franx 1991 and references therein). Taken together, it is difficult to reconcile these observational data with a model in which the bulk of their stars are formed in a single, relatively short-lived « 1 Gyr), starburst at high-redshift during the free-fall collapse of a massive, isothermal gaseous halo (> 1011 M0).

The favoured formation model for elliptical galaxies is that they grow from the hierarchical clustering of lower-mass gas-rich clumps (discs or spheroids), in which the oldest stars (~ 10 Gyr) have already formed and therefore are not associated with any subsequent starformation episodes that ultimately form the elliptical structures. In this scenario the progenitor of a present-day giant-elliptical galaxy may still be in pieces at z > 2 or at least in the process of merging. N-body hydrodynamic simulations (Hernquist 1989) show that dynamical instabilities introduced during a merging event between galaxies can drive large masses of gas (2:: 109 M 0 )

into the central few hundred parsecs of the resultant galaxy on timescales of < 108 yrs. CO interferometric measurements of nearby ULIRGS (LFIR 2:: lOll L 0 ) confirm that strongly interacting and merging galaxies have extremely high gas surface densities in their nuclei (~H2 > 5 X 103 M0 pc-2 -

Sargent & Scoville 1991, Scoville et al. 1991) and are undergoing a high level of starformation (SFR > 200 - 500M0 yr- 1 ). Recent high resolution optical studies of distant clusters (Dressler et al. 1994, Cowie et al. 1995) are now

THERMAL EMISSION IN HIGH-Z GALAXIES 313

beginning to provide observational support for merging and starformation at higher redshifts (z > 1).

A characteristic signature of massive star formation is intense FIR emission from dusty, molecular material, where the rate of dust production is proportional to the star formation rate. The dust is heated primarily by the embedded 0 and B stars which evolve quickly and disperse their surrounding material on similarly short timescales ('" 1 07yrs , Wang 1991). Hence the FIR luminosity provides a measure of the current star formation rate,

where W = 0.8 - 2.1 (Scoville & Young 1983, Thronson & Telesco 1986).

In the Milky Way and local disc galaxies a significant fraction (30%) of the bolometric luminosity is re-radiated at FIR wavelengths, and hence the SFR (Miller & Scalo 1979, Kennicutt 1983) and the ratio LFIR/ Lbol cannot have evolved much with look-back time. However the situation is very different for elliptical galaxies. Mazzei, de Zotti & Xu (1994) have modelled the photometric evolution of elliptical galaxies and show that LFIR/ Lbol rv 0.3 within the first 1 - 2Gyr (see Fig. 1), whilst at the current epoch ellipticals emit < 1% of their bolometric luminosity at FIR wavelengths. The details of the evolution are sensitive to the assumed initial mass function (IMF) and SFR, where the steeper IMF and higher SFR produces a more luminous, but shorter, burst of starformation.

No matter what, the formation of elliptical galaxies, whether they grow through merging or form via the collapse of a single gaseous halo, is expected to be a spectacular and luminous phenomenon at FIR wavelengths in the rest-frame, suggesting that the discovery of high-z proto-ellipticals at submillimetre wavelengths is a realistic prospect.

Throughout this paper I assume Ho = 50kms-1 Mpc-1 and qo = 0.5.

2. Submillimetre continuum emission from AGN at high z

All radio-quiet galaxies (both quiescent and those hosting an AGN) at low red shift exhibit a spectral energy distribution with a sharp turnover at rv 100 JL due to thermal emission from dust (Sanders et al. 1989, Chini et al. 1989a, Barvainis & Antonucci 1989, Hughes et al. 1993). However there is increasing evidence that an underlying FIR thermal component also exists in radio-loud galaxies (Gear et al. 1985, Knapp & Patten 1991, Antonucci et al. 1990). At redshifts z > 2 this FIR spectral peak moves to wavelengths > 300 JLm and we therefore expect galaxies undergoing massive starformation at high-z to be bright at submillimetre wavelengths,

314

1

..10 o .... °-1 ..-"-,....... 6 ....J-2 t<

01 .2-3

-4 1000

DAVID H. HUGHES

100 10 1 0.1

Wavelength Vtm)

Figure 1. Evolution of the UV to millimetre wavelength rest-frame spectral energy distribution of elliptical galaxies. Models corresponding to galactic ages between 2 and 15 Gyr are shown, assuming a Salpeter IMF with a lower mass limit, ml = 0.01 Me;). The SFR (..p(t) = (mg ... /mgadt/to) is proportional to the fractional mass of gas in the galaxy and assumes an intial SFR t/to = 100 Me;) yr-1 • The data are taken from Mazzei, de Zotti & Xu (1994).

providing the opportunity to make ground-based observations in the few atmospheric windows between 350 - 1300 pm.

Figure 2 illustrates how the observed flux density varies with redshift at 800 pm for a typical starburst galaxy (e.g. M82, LFIR = 3 X 1010 L0 ), and demonstrates that between z = 1 - 10, assuming the rest-frame spectrum peaks at '" 100 pm, (which corresponds to dust temperatures of 30-60 K), the increase in the intrinsic brightness of the source as the rest-frame wavelength climbs the steep (Fil ex vn , n = 3-4) Rayleigh-Jeans tail, is sufficient to offset the dimming (assuming n = 1) due to the increasing cosmological distance. The dependence of the observed flux density on redshift is similar at other submillimetre and millimetre wavelengths.

If it is assumed the submillimetre continuum (Arest > 200 pm) is due to optically-thin emission from heated dust grains with no additional contribution from bremsstrahlung or synchrotron radiation, a measure of the dust mass Md can be determined directly from the relationship,

o o o ..-I

o o ..-I

..-I

o

..-I

o o

THERMAL EMISSION IN HIGH-Z GALAXIES

MB2. \ \ \

z

\ . \

•

" \

315

Figure 2. The dependence of flux density at 800 I'm on redshift for a spectral energy distribution similar to that of the starburst galaxy M82 (Hughes et al. 1994), assuming no = 1. The curves shown represent xl (solid), xlO (dashed), x100 (dashed-dotted) and x1000 (dotted) the rest-frame FIR luminosity of M82 (LFIR = 3 X 1010 Le). Also shown are measured 800 I'm flux densities of various high-z and low-z active galaxies; starbursts (solid-stars), Seyferts (dotted-circles), ULIRGS (crossed-circles), radio galaxies (open-stars), radio-quiet quasars (solid-circles), taken from Hughes, Ward & Davies (1996a), Hughes et al. (1993) and data included in Table 2 (this paper). The flux densities of IRASF10214+4724 and the Cloverleaf quasar (H1413+117) have been corrected for their amplification due to lensing, see Sect. 4. The error-bar associated with IRAS FI0214+4724 represents the uncertainty in the amplification factor at FIR wavelengths.

316 DAVID H. HUGHES

M _ 1 SobsDl d - (1 + z) k~estB(vrest, T)

where z is the red shift of the source, Sobs is the observed flux density, k~est is the rest-frequency mass absorption coefficient, B(vrest, T) is the rest-frequency value of the Planck function from dust grains radiating at a temperature T, and DL is the luminosity distance, given by

for no = 1, and by

cz z DL = -(1 +-)

Ho 2 for no = o.

Thus, for a given cosmology and excluding the inherent uncertainty of the continuum measurement, the robustness of dust mass determinations from submillimetre photometry depends on the uncertainty in kd(V) and T. Our understanding of each of these parameters will be briefly discussed in turn.

3. Uncertainties in the calculation of the dust and molecular gas masses from continuum observations

A reasonable estimate of the maximum fractional uncertainty in kd at 800 fLm is ~ 7, with the values of kd(800 fLm) ranging between 0.04m2 kg-1

(Draine & Lee 1984) and 0.3m2 kg-1 (Mathis & Whifffen 1989) with intermediates values of 0.15 m2 kg-1 (Hildebrand 1983) and 0.12 m 2 kg-1 (Chini et al. 1986). I have adopted an average value of kd(800 fLm) = 0.15 ± 0.09m2 kg-1 and assumed that kd ex >.-2. A different choice of kd should therefore only be expected to result in estimates which differ from the dust masses calculated here by at most a factor of ~ 2.

The fractional uncertainty in the dust mass, which results from our ignorance of the dust temperature within the range Tl < T < T2 , is given by

Ml ehvre.l/kTt - 1 M2 = ehVreSI/kT2 - 1

which increases rapidly as Vrest moves above the Rayleigh-Jeans tail of the thermal dust emission. Dust mass estimates derived from rest-frame submillimetre photometry therefore have the benefit of being relatively insensitive to uncertainties in temperature, as compared to dust masses derived from FIR data (i.e. >'rest < 200 fLm).


The temperature of dust grains that radiate at submm-FIR wavelengths and dominate the FIR luminosity in nearby starburst galaxies, low metallicity dwarf galaxies, ellipticals, ULIRGs, Seyferts and radio-quiet quasars is typically 50 ± 20 K (Sanders et al. 1989, Chini et al. 1989a,b, Barvainis et al. 1992, Hughes et al. 1993, Hughes, Ward & Davies 1996a, Wiklind et al. 1995). There is no a priori reason to believe that the temperature of dust grains that dominate the rest-frame luminosity of starforming regions in high-z galaxies should be significantly different at early epochs.

At z ~ 0, where submillimetre photometry samples the Rayleigh-Jeans tail of the thermal dust emission, the error in the estimated dust mass is proportional to the uncertainty in the dust temperature. Unfortunately the uncertainty in Mdust increases rapidly with increasing redshift unless the observing frequency is reduced appropriately. This is shown clearly in Fig. 3 where the increase in the fractional uncertainty in the dust mass with redshift is calculated for photometry taken through filters that match the ground-based atmospheric windows available to current and future receivers at 350 f-Lm ~ 2mm.

0 N

..-... / 450JMT1 75Ol'rr/ .' 850JMT1 ~

0 / ,..... LO II ...... / 1100JMT1 l- I ......,

/ "0 I / ~ ./ "'- 0 / / ..-... ......

6ooJMT1 ~ ./ / 0 / . . ./ t') . ..... II ./. . ./

I- ..... ' .. - -LO -......, .. ~ :. ..... --.:-- -"0

! C. :.-.-: "--~

0 0 1 2 3 4 5 6 7 8 9 10

z

Figure 3. Fractional uncertainty in dust mass as a function of source redshift, based on a single photometric measurement represented by the curves shown, assuming the dust temperature lies in the range 30 -> 70 K. For example, if a fractional error in dust mass of greater than 5 is considered to be unacceptable, then a single photometric observation at wavelengths:::; 800 JIm is worthless for galaxies at z > 3.

318 DAVID H. HUGHES

4C41.17 BR1202-0725

100 100

1 10 f 10 ......

~ ~ , • c • c • ., • c 0.1 c 0.1 )( )(

'" '" ii: 0.01 I II ii: 0.01

10-3 10-3 105 10 105 10

Observed Wavelength CIun) Observed Wavelength CIun)

Figure 4. Spectral energy distributions of the radio galaxy 4C41.17 (z=3.8) and the radio-quiet quasar BR1202-0725 (z=4.68). The radio emission in 4C41.17 is for the core only. The dashed, solid, and dashed-dotted lines represent isothermal grey-body emission, with an emissivity index j3 = 2, from dust at temperatures of 30 K, 50 K and 70 K respectively. Data are taken from Dunlop et al. 1994, Chini & Kriigel 1994, Yun & Scoville, in prep., Isaak et al. 1994 and McMahon et al. 1994

For example, at z ~ 3 dust masses determined from SSOOJLm photometry assuming T = 30 K are typically six times greater than dust masses calculated assuming T = 70 K. This level of uncertainty in the temperature makes an estimate of the dust mass based on a single measurement at ::; 800 I'm almost useless, and the most appropriate value of Tdust in high-redshift galaxies is therefore the dominant source of uncertainty in the derived dust mass. Hence it is essential to constrain, or place an upper limit on, the dust temperature of the coolest component (which dominates the mass) by obtaining multi-wavelength data that measures the rest-frame FIR spectral turnover between 50 I'm - 450 I'm (e.g. see Fig. 4, Barvainis et al. 1992 and Rowan-Robinson et ai. 1993). Such data will be readily available within a few years for large numbers of high-z galaxies using new continuum instruments on ground-based and space-borne telescopes (e.g. SCUBA on the JCMT, ISOPHOT on ISO).

In order to say something about the evolutionary status of galaxies at high-z it is necessary to convert the estimates (including upper-limits) of the dust mass into a measure of molecular gas still available for future starformation, since it has proved particularly difficult to detect the molecular gas in high-z galaxies directly (see papers by Radford, and Israel and Van der Werf - these proceedings). Unfortunately the H2 gas-dust ratio is not a well determined quantity in galaxies at low-redshift, let alone at high-z. Studies of damped Lyman-a systems (DLAAS) currently provide the only opportunity to directly measure the dust content of the universe at early


TABLE 1. Summary of all published submillimetre and millimetre continuum data measuring thermal emission from dust in galaxies at z < 2

Source name z 'ype Sl.25mm SaOOpm S.SO,.m S350p,m Ref. (mJy) (mJy) (mJy) (mJy)

IRASI02H+4124 2.286 IRAS 2H:5 50:5 273:45 53W002 2.390 RG 6.9:2.3 2 H1413+117 2.546 BALQSO < 35 44:8 224:38 189:56 3 2132+0126 3.194 RQQ 11.5:1. 7 < 12 2,4 B20902+34 3.391 RG 3.1:0.6 <14 2,5 0345+0130 3.638 RQQ 6.1:2.0 < 25 2,4 4C 41.17 3.800 RG 2.5:0.4 17.4:3.1 < 56 5,6 PC2047+0123 3.800 RQQ 1.9:0.5 7 8C1435+643 4.26 RG 2.6:0.4 7 0307+0222 4.379 RQQ 6.6:1.7 4 BRI033-0327 4.51 RQQ 12:4 8,9 BR1202-0725 4.69 RQQ 10.5:1.5 50:7 92:38 8,9

1. Rowan-Robinson et al. 1993; 2. Hughes et al.1996b; 3. Barvainis et al. 1992; 4. Andreani et al. 1993; 5. Chini et al. 1994; 6. Dunlop et al. 1994; 7. Ivison 1995; 8. Isaak et al. 1994; 9. McMahon et al. 1994

epochs. At z rv 3 it has been suggested that the gas-dust ratio in DLAAS is 400-2000 (Fall, Pei & McMahon 1990, Pettini et al. 1994), a value significantly higher than the galactic value of 100-160 (Hildebrand 1983, Savage & Mathis 1979), and on average higher than that found in nearby spirals ('" 500, Devereux & Young 1990), ellipticals ('" 700, Wiklind et al. 1995) and ULIRGS (540 ± 290, Sanders et al. 1991). In the calculation of the molecular gas masses (in Table 2) a conservative value of MH2/Md '" 500 has been assumed.

4. Current status of rest-frame FIR-sub millimetre wavelength observations of galaxies at z > 2

Table 1 summarises the observations of all radio galaxies and radio-quiet quasars with redshifts z > 2 that have detections at one or more wavelengths between 350 j.£m - 1300 j.£m. This table excludes the unpublished detections of 6 additional radio-quiet quasars at z > 2 (see Omont, these proceedings). The FIR luminosities, dust masses and SFRs in Table 2 are determined from an optically-thin, isothermal 50 K greybody consistent with the submillimetre data (see Fig. 4) and assume the value of kd described in Sect. 3.

Given the dust masses are in the range 6 X 107 - 6 X 108 M0' allowing for amplifications of order'" 11 (Barvainis et ai. 1994) and", 10 - 30 (Eisenhardt et al. 1996) for the FIR luminosity in the lensed sources HI413+117 and IRASI0214+4724 respectively, then we can infer that the molecular gas mass (using MH2 / Md '" 500, see above) in high-z radio galaxies and quasars

320 DAVID H. HUGHES

TABLE 2. Dust masses, FIR luminosities and starformation rates (SFRs) of high-z galaxies. Corrections for the amplification of the FIR emission due to lensing have been applied to the values for IRAS10214+4724 and H1413+1l7 (see Sect. 4).

Source name z log Md/M0 log LFIR/ L0 SFR (M0 yr-1 )

IRAS F10214+4724 2.286 7.76 12.96 903 53W002 2.390 8.19 13.40 2511 H1413+117 2.546 7.97 13.17 1509 B20902+34 3.391 8.40 13.67 4677 4C41.17 3.800 8.42 13.79 6165 PC 2047+0123 3.800 8.07 13.43 2691 8C 1435+643 4.26 8.15 13.44 2754 BR 1033-0327 4.51 8.20 13.63 4265 BR 1202-0725 4.69 8.81 14.25 17782

is ~ 5 X 1010 MG , i.e. a significant fraction of the stellar mass observed in their present-day counterparts.

5. Concluding remarks

If it is correct to assume that at z > 2 the rest-frame LFIR still provides a measure of the starformation rate then, in the absence of any contribution from an AGN or amplification due to lensing, the rest-frame luminosities in Table 2 (LFIR > 1013 LG)' imply SFRs > 1000MG yr-1 and suggest that the entire molecular gas content of a primceval galaxy (1011 - 1012 M G ) could be converted into stars in < 1 Gyr. Evidence for extreme SFRs and young galaxy ages « 1 Gyr) has also been found in the rest-frame UV-optical morphologies and SEDs of 53W002 and 4C41.17 (Windhorst et al. 1992, Chambers et al. 1990, Mazzei & de Zotti 1996).

Despite the uncertainties described in Sect. 3 that affect the absolute measure of the physically interesting quantities (mass, luminosity, SFR), we can still conclude that the high-z radio galaxies and radio-quiet quasars, which have been detected at submm-mm wavelengths, are extremely dusty (Table 2), with dust masses> lOx larger than observed in their low-z (z < 0.5) counterparts (Chini et al. 1989a, Knapp & Patten 1991, Hughes et al. 1993), and with levels of starformation and starforming efficiencies similar to, or exceeding those observed in low-z ULIRGS. This comparison is illustrated in Fig. 5 where the masses and SFRs are calculated for each AGN by scaling the dust mass (Md = 9.6 X 105 M G , Hughes et al. 1994) and the current SFR (3 MG yr- 1 ) of M82 by ratio of the measured 800 /-lm flux of the AGN compared to the 800 /-lm flux density of M82 if observed


°0=1

1010 "I "I "1

-: 104

• 109 "" -- - - - ------ -- -- -- -- - --- -----e.

(fCr -: 1000 0

108 ... -- -- --~ -:- - ~ - -- -1- -- - ,....... ~ ~ "'-fIl

100 "'-fIl @ • 0 as • ~ ~ 107 ~- ..........

.- -- - -- --+I

~0 ~

fIl ~ ~ -: 10

Q 0rB Ul

106 ,-0

•• 0 -: 1

105 iiJ 0

.. I u.d 0.1 10-3 0.01 0.1 1 10

z

Figure 5. Dust masses and SFRs of high- and low-z AGN calculated according to the method described in Sect. 5. The symbols and lines have the same meaning as in Fig. 2.

at the redshift of the AGN concerned (as shown previously in Fig. 2). The discrepancy, of order a factor f'V 3 at the highest redshifts, between this method and the values in Table 2 is due entirely to the difference in the SED of M82 and the assumptions regarding kd (Sect. 3) in this paper.

Without the benefits of lensing to amplify the continuum fluxes the current submillimetre and millimetre bolometers are only sensitive to SFRs ~ 1000M0 yr-1 . However the next generation of bolometer arrays,e.g. SCUBA on the JCMT (Gear & Cunningham 1995), on single-dish telescopes will have not only a 10 times improvement in the sensitivity but also the ability to image a field of f'V 900 X 900 kpc2 in a single snap-shot at redshifts of z f'V 3 with a spatial resolution of 50 kpc (HPBW f'V 7" at 450 /.lm for a 15 m submillimetre telescope).

The long-term future of high-z astronomy at submillimetre wavelengths undoubtably lies with interferometric observations where further gains in

322 DAVID H. HUGHES

sensitivity and angular resolution will make it possible to distinguish unambiguously whether the redshifted FIR emission currently detected towards high-z sources is due to a single protogalactic structure, or whether the emission arises from a number of interacting and merging lower mass disc galaxies and spheroids, that formed at still higher-redshifts, and currently occupy a volume larger than that of the final galaxy. A comprehensive description of future submm-mm interferometers is given in a previous conference (see IAU Coll. 140).

References

Andrew, P., LaFranca, F., Christiani, S. 1993, MNRAS, 261, L35 Antonucci, R.R, Barvainis, R., Alloin, D. 1990, ApJ, 353, 416 Barvainis, R., Antonucci, R.R. 1989, ApJS, 70, 257 Barvainis, R., Antonucci, R.R., Coleman, P. 1992, ApJ, 399, L19 Barvainis, R. et al. 1994, Nature, 371, 586 Chambers, KC., Miley, G.K, van Breugel, W. 1990, ApJ, 363, 21 Chini, R., Kreysa, E., Biermann, P.L. 1989a, A&A, 219, 87 Chini, R., Kriigel, E. 1994, A&A, 288, L33 Chini, R., Kriigel, E., Kreysa, E. 1986, A&A, 167, 315 Chini, R., Kriigel, E., Kreysa, E., Gemiind, H.-P. 1989b, A&A, 216, L5 Cowie, L.L, Hu, E.M., Songaila, A. 1995, AJ, 110, 1576 de Zeeuw, P.T., Franx, M. 1991, Ann. Rev. Astron. Astrophys., 29, 239 Devereux, N., Young, J.S. 1990, ApJ, 359, 42 Draine, B.T., Lee, H.M. 1984, ApJ, 285, 89 Dressler, A., Oemler, A., Sparks, W.B. 1994, ApJ, 435, L23 Dunlop, J.S., Hughes, D.H., Rawlings, S., Eales, S.A., Ward, M.J. 1994, Nature, 370, 347 Eisenhardt, P.R., Armus, L., Hogg, D.W., Soifer, B.T., Neugebauer, G., Werner, M.W.

1996, ApJ, to appear 10 April 1996 Fall, S.M., Pei, Y.C. 1995, in QSO Absorption Lines, ed. G.Meylan, Spinger-Verlag Fall, S.M., Pei, Y.C., McMahon, R.G. 1989, ApJ, 341, L5 Gear, W.K., Gee, G, Robson, E.L, NoIt, LG. 1985, MNRAS, 217, 281 Gear, W.K, Cunningham, C. 1995, in Mtlltifeed systems for radio telescopes, P.A.S.P.

Conf. Ser., Vol. 75, p.215, eds. D.T. Emerson, J.M. Payne Guideroni, B., Rocca-Volmerange, B. 1987, A&A, 186, 1 Hernquist, L., 1989, 340, 687 Hildebrand, R. 1983, QJRAS, 24, 267 Hughes, D.H., Robson, E.I, Dunlop, J.S., Gear, W.K 1993, MNRAS, 263, 607 Hughes, D.H., Gear, W.K, Robson, E.L 1994, MNRAS, 270, 641 Hughes, D.H., Ward, M.J., Davies, R. 1996a, in preparation Hughes, D.H., Dunlop, J.S, Rawlings, S. 1996b, in preparation IAU Colloquium 140, 1994, Astronomy with Millimetre and Stlbmillimetre Wave Inter-

ferometry, A.S.P. Conf. Ser., Vol 59, eds. M. Ishiguro and Wm. J. Welch Isaak, K, McMahon, R.G., Hills, R.E., Withington, S. 1994, MNRAS, 269, L28 Ivison, R.J. 1995, MNRAS, 275, L33 Keel, W.C., Windhorst, R.A. 1991, ApJ, 383, 135 Kennicutt, R. 1983, ApJ, 272, 54 Knapp, G.R, Patten, B.M. 1991, AJ, 101, 1609 Kormendy, J. 1984, ApJ, 287, 577 Lees, J.F., Knapp, G.R., Rupen, M.P., Phillips, T.G. 1991, ApJ, 379, 177 Mathis, J.S., Whiffen, G. 1989 ApJ, 341, 808 Mazzei, P., de Zotti, G. 1994, ApJ, 426, 97

THERMAL EMISSION IN HIGH-Z GALAXIES

Mazzei, P., de Zotti, G., Xu, C. 1994, ApJ, 422, 81 Mazzei, P., de Zotti, G. 1996, MNRAS, in press

323

McMahon, R.G., Omont, A., Bergeron, J., Kreysa, E. Haslam, C. 1994, MNRAS, 267, L9

Miller, G.E., Scalo, J.M. 1979, ApJS, 41, 513 O'Connell, R.W. 1987, in Stellar Populations, STScI Symp. Ser., Vol. 1, p.167., eds. Nor-

man, Renzini, Tosi Pettini, M., Smith, L.J., Hunstead, R.W., King, D.L. 1994, ApJ, 426, 79 Rowan-Robinson, M. et al. 1993, MNRAS, 261, 513 Sandage, A. 1972, ApJ, 178, 25 Sanders, D.B., Scoville, N.Z., Soifer, B.T. 1991, ApJ, 370, 158 Sanders, D.B., Phinney, E., Neugebauer, G., Soifer, B., Matthews, K. 1989, ApJ, 347, 29 Sargent, A.I., Scoville, N.Z. 1991, ApJ, 366, L1 Savage, B.D., Mathis, J.S. 1979, Ann. Rev. Astron. & Astrophys., 17, 73 Scoville, N.Z., Sargent, A.I, Sanders, D.B., Soifer, B.T. 1991, ApJ, 366, L5 Scoville, N.Z., Young, J.S. 1983, ApJ, 265, 148 Thronson, H., Telesco, C. 1986, ApJ, 311, 98 Wang, B. 1991, ApJ, 374,456 Wiklind, T., Combes, F., Henkel, C. 1995, A&A, 297, 643 Windhorst, R.A., Mathis, D.F., Keel, W.C. 1992, ApJ, 400, L1 Wolfe, A. 1993, in Relativistic Astrophysics and Particle Cosmology, Texas/PASCOS'92,

eds. Akerlof, C. & Srednicki, M., Ann. N.Y. Acad. Sci., 668, 281

SUB MILLIMETRE OBSERVATIONS OF QSOS AT RED SHIFTS Z>4

K.G. ISAAK, R.E. HILLS, S. WITHINGTON Mullard Radio Astronomy Observatory Cavendish Labomtory, Madingley Road Cambridge, CB3 OHE, U.K.

AND

R.G. MCMAHON Institute of Astronomy Madingley Road Cambridge, CB3 OHA, U.K.

Abstract. We present an interim summary of the results of a program to search for submillimetre-wave line, and continuum emission from the host galaxies of QSOs at z > 4. To date, we have observed 18 sources, and have detected 800 J.Lm continuum emission from BR 1202-0725, BR 1033-0327 and BR 1335-0417. The submillimetre spectral index, a~~:::m' for each of these sources is greater than 2.4, and thus the (sub ) millimetre continuum spectrum is consistent with thermal emission from warm, optically thin dust.

1. Background

The identification of IRAS F10214+4724 as an ultraluminous object at z = 2.286 (Rowan-Robinson et al., 1991) confirmed the speculations of some that the submillimetre waveband would prove to be important to studies of early galaxy formation. The subsequent discovery of submillimetre continuum (eg. Downes et al., 1992), and CO line emission (eg. Solomon et al., 1992) suggested that IRAS FI0214+4724 was in fact a young galaxy undergoing its first bursts of massive starformation. Since then, this object has been shown to be gravitationally lensed (e.g. Broadhurst and Lehar, 1995), and accordingly, the interpretations of observations listed above are

325


326 K.G. ISAAK, R.E. HILLS, S. WITHINGTON AND R.G. MCMAHON

less extreme. Nonetheless, lRAS F10214+4724 is an inspirational object, from which one is led to ask the question of whether similar ultraluminous objects exist at even higher redshifts.

Starburst galaxies contain large amounts of warm dust and molecular gas which emit strongly at far-infrared (FIR) wavelengths. The dust absorbs the UV flux from OB stars and is thus heated to temperatures of 10-100 K. The rest-frame thermal emission ofthe dust at these temperatures peaks in the FIR/IR, however as one moves to higher redshift, the peak shifts from the observed IR, to submillimetre wavelengths. The emission spectrum of dust in the Rayleigh-Jeans region is a steeply rising function of frequency, and so the observed flux at 800 I'm actually increases with redshift between the range z = 1 to z f'V 5. In contrast, the fluxes at lRAS wavelengths (100 I'm, 60 I'm) fall, and so the sub millimetre waveband becomes that in which to search for high-redshift ultraluminous objects (e.g. Blain and Longair (1993), Blain (this volume)).

Continuum and spectral line observations provide complementary information on sources: with a continuum emission spectrum one can determine dust mass and temperature, while line observations, particularly the line width, may be used to determine kinematic information and the virial mass of the emitting source. One of the strongest lines in regions of high starforming activity is the FIR (158 I'm) fine-structure cooling transition of singly ionized carbon (Crawford et ai., 1985; Stacey et ai., 1991); the line may contain up to 1% of the total FIR luminosity in local starbursts, and is the strongest FIR line observed in our own Galaxy (Wright et ai., 1991).

2. Observations

Existing instrumentation does not yet have the required sensitivity to undertake large-scale surveys of the sky at sub millimetre wavelengths. We therefore chose to use QSOs to pinpoint their less luminous host galaxies. Our observed sample consisted of 18 QSOs of redshifts z > 4, 13 of these were taken from the APM BRI survey (Irwin et ai., 1991), and the remainder from a selection of different sources including a similar high redshift survey by Schneider et al. (1991). Where possible, radio-quiet QSOs were chosen to minimize the likelihood of contamination of the submillimetre waveband by synchrotron emission.

2.1. CONTINUUM OBSERVATIONS

Each source was observed using the single-pixel continuum bolometer UKT14 at the 15m diameter James Clerk Maxwell Telescope (JCMT) on Mauna Kea, Hawaii. Observations were made at 800J.tm as a compromise between atmospheric stability and transmission (best at longer A), and high intrin-

SUBMM OBSERVATIONS OF HIGH REDSHIFT QSOS 327

o o

1000 100 10

1000

/'-'\J. J,. ._\ '.

\ \ \ \ \ \ \ \ \ \ \ \ \ , , i , i \ i , i , i , i , . , , ,

100 10

Observed Wavelength/ Ikm

Figure 1. The continuum spectrum of BR1202-0725 (taken from Isaak et al. (1994). The lower horizontal scale denotes the observed wavelength, while the upper horizontal scale indicates the rest-wavelength. The filled circles (with lIT error bars) denote fluxes measured at the JCMT (Isaak et al. 1994); the open circle denotes the 1.25 mm flux measured at IRAM (McMahon et al. 1994); the arrows denote IRAS upper limits, and the near-infrared points are observations made at UKIRT (Storrie-Lombardi et al. in prep.). The lines denote fits as given in the text.

sic source luminosity (increases at shorter wavelengths close to the thermal emission peak).

1. Three objects have been detected at 800 Jlm, two of which (BR 1033-0327 and BR 1202-0725) have already been reported in Isaak et al. (1994). BR 1335-0417 was detected on two different observing runs (> 5cr on first, > 2.4cr on second due to poor weather conditions). A tentative value of 30 ± 6mJy is assigned to the 800 Jlm flux, however this value needs to be confirmed with further observations.

2. The submillimetre spectral indices (5 oc VOl) have been evaluated using our 800 Jlm fluxes and those measured at 1250 Jlm by Omont et al. (priv. comm., also this volume) and McMahon et al. (1994): a~~~t:m ~ 2.4 for each of the three sources, thus consistent with thermal emission from warm dust.

3. By combining observations made at 1100 Jlm, 800 Jlm and 450 Jlm, we have modeled the continuum emission from BR 1202-0725 using a grey body fit. There are insufficient points to fully constrain such a model, however if we assume that the dust emissivity is parametrized by a spectral index of (3 = 2, (3 = 1.5 or (3 = 1 (and thus for optically

328 K.G. ISAAK, R.E. HILLS, S. WITHINGTON AND R.G. MCMAHON

thin thermal emission from dust; a = 4, a = 3.5 and a = 3), then the observations are consistent with dust at a physical temperature of approximately 58 K (solid line), 68 K (dashed line) and 98 K (dasheddotted line) respectively (Fig. 1). The dust mass is estimated to be rv 108 M(!h and depends strongly on the chosen dust temperature.

4. The continuum emission is consistent with thermal emission from warm dust, however we have no indication of whether the dust is heated by a starburst or by the central AGN (Sanders et al. (1991».

2.2. SPECTRAL LINE SEARCH

A search was made for C+ line emission from BR 1202-0725 on the basis of the strong continuum emission observed from this source. At z = 4.69, the 1.9 THz rest-frequency line emission moves down to 334 GHz. Observations were made using the Schottky dual-channel heterodyne receiver RxB2 at the JCMT. The combination of a broad anticipated line width (dv ~ 300km s-I), instrumental baseline ripples and the difficulty in determining precisely the redshift ofthese high-redshift objects, meant that it was necessary to search for the line over a 3 GHz interval. To achieve this, the receiver was tuned at a number of different overlapping frequencies to cover the required bandwidth.

No line was detected during either of two observing trips. It should have been possible to detect the presence of a line at around the level of the continuum, and so the lack of line may be attributed to: (a) A lower line-to-continuum ratio in BR 1202-0725 than that seen in local starbursts (b) The difficulty in deriving intrinsic redshifts from broad QSO emission lines (c) A broader C+ line width consistent with a cluster rather than galactic dispersion velocity. The heterodyne data is, however, consistent with an interpolated value of the continuum flux determined from measurements made with UKT14, and thus provides independent confirmation of the submillimetre continuum emission.

3. Summary

We have observed 18 QSOs at z > 4 to try to detect sub millimetre emission from the underlying host galaxies. Three sources were detected at 800 J.Lm, and, combined with observations by Omont et al. (in prep. - priv. comm.), the results suggest that in each case the continuum emission is consistent with that from warm dust. Model fitting ofthe 1100 J.Lm, 800 J.Lm and 450 J.Lm fluxes from BR 1202-0725 is in agreement with this, and further suggests that the continuum emission from BR 1202-0725 is consistent with dust

SUBMM OBSERVATIONS OF HIGH REDSHIFT QSOS 329

at temperatures of around 50-60 K. No clear sign of the C+ line was seen around the best estimate of the source redshift.

We plan to continue both the continuum and line emission searches with the advent of more sensitive and stable instrumentation.

References

Blain, A.W. and Longair, M.S (1993), MNRAS, 264, 509 Broadhurst, T. and Lehar, l. (1995), Ap.J, 450, L41 Downes, D., Solomon, P.M. and Radford, S.l.E. (1994), ApJ., 414, L13 Crawford, M.K., Genzel, R., Townes, C.H. and Watson, D.M. (1992), ApJ., 291, 755 Irwin, M., McMahon, R.G., and Hazard, C. (1991), in The Space Distribution of Quasars,

ed. D. Crampton, ASP Conference Series 21, 117 Isaak, K.G., McMahon, R.G., Hills, R.E. and Withington, S. (1994), MNRAS, 269, L28 McMahon, R.G., Omont, A., Bergeron, l., Kreysa, E. and Haslam, C.G.T. (1994),

MNRAS, 267, L9 Rowan-Robinson, M., Broadhurst, T., Lawrence, A., McMahon, R.G., Lonsdale, C.l.,

Oliver, S.l., Taylor, A.N., Hacking, P.B., Conrow, T., Saunders, W., Ellis, R.S., Efstathiou, G.P. and Condon, l.l. (1991), Nat., 351, 719

Sanders, D.B., Phinney, E.S., Neugebauer, G., Soifer, B.T. and Matthews, K. (1989), ApJ., 347, 29

Schneider, D.P., Schmidt, M. and Gunn, J.E. (1991), A.J., 101, 2004 Solomon, P.M., Downes, D. and Radford, S.l.E. (1992), ApJ., 398, L29 Stacey, G.J., Geis, N., Lugten, J.B., Poglitsch, A., Sternberg, A. and Townes, C.H. (1991),

ApJ., 373, 423 Wright, E.L., et al. (1991), ApJ., 381, 200

1.25 MM DETECTION OF 7 RADIOQUIET QSOS WITH VERY HIGH Z

A.OMONT Institut d'Astrophysique de Paris, C.N.R.S.

R.G. MCMAHON Institute of Astronomy, Cambridge

P.cox Observatoire de Marseille and Max-Planck-Institut fur Radioastronomie, Bonn

E. KREYSA Max-Planck-Institut fur Radioastronomie, Bonn

AND

J. BERGERON Institut d'Astrophysique de Paris, C.N.R.S. and ESO, Garching

Abstract. We have performed a systematic study of the 1.25mm continuum emission of radio-quiet QSOs with z > 4, with the IRAM 30 m telescope. In addition to the case of BR 1202-0725 previously reported, five new sources with z > 4 have been detected, as well as one with z = 2.7. Their fluxes range from 2.5 to 10mJy. In addition 18 other sources with z rv 4 were searched for but not detected with fluxes probably smaller than 3-4mJy.

1. Introduction

The detection by IRAS of strong far infrared emission by bright quasars has revealed the presence of very large amounts of dust. Their far IR luminosity can be comparable to their huge UV luminosity. The very steep submillimeter emission spectrum of dust in the rest frame can considerably rise the detect ability of high redshift sources in the submillimeter and millimeter ranges, since the observed flux increases with red shift for constant

331


332 A. OMONT ET AL.

luminosity (see e.g. Fig. 1 of McMahon et al. 1994). We have already reported the success of our first search, with the detection of the bright QSO BR 1202-0725, with z = 4.7, at 1.25 mm at IRAM (McMahon et al. 1994), and in the submm range at JCMT (Isaak et al. 1994). The mass of dust inferred, > 108 M0' is probably among the largest known in a single object.

We report here the results of the continuation of our program of systematic search for 1.25 mm detection of QSOs with z > 4. The new detections show that the case of BR 1202-0725 is not an exception, but that the rate of millimeter detections among such sources is relatively high. A more detailed report of this work is given in Omont et al. (1996).

2. Results

The observations were performed with the IRAM 30m telescope equipped with MPIfR bolometer arrays (Kreysa 1993, see McMahon et al. 1994)

The results are displayed in Table 1. Each of the detected sources was consistently detected at least at a 2 or 3 sigma level in several different days. Alltogether the combination of these observations warrants a 5 sigma level for the six new detections.

TABLE 1. 1.25 mm detections

Source z 1.25mm flux (mJy)

BRI0952-0115 4.43 2.7S±0.63 BR 1033-0327 4.51 3.45±0.65 BRl117-1329 4.00 4.09±0.Sl BR 1144-0723 4.15 5.S5±1.03 BR 1202-0725 4.69 12.6±2.29 BRI 1335-0417 4.40 10.3±1.04 LBQS 1230+1627 2.70 7.5 ±1.4

Our main goal was a systematic study of the Cambridge APM sample of radio-quiet QSOs with z > 4 (Irwin et al. 1991, 1996). We observed 16 of them, i.e. about half the sample, with an r.m.s. noise ~ 1.5 mJy. In addition to the 5 new detections (plus BR1202-0725, McMahon et al. 1994), there are thus 10 non detections among this sample with a 30' upper limit of 5mJy (4mJy for most of them). In addition, we observed with a comparable sensitivity, without any detection, 8 radio-quiet QSOs with z ;::: 4 detected in the visible by various authors, mainly Schneider et al. (1991). Among these sources, are those reported as detected by Andreani et al. (1993). The

1.25MM DETECTION OF VERY HIGH Z RADIO QUIET QSOS 333

non detection of PC 2132+0126 (z = 3.18) is quite puzzling since Andreani et ai. reported 11.5± 1.7mJy, while we measured 0.01 ± 1.20mJy.

Since the detection rate was relatively high among the z > 4 QSOs, one is naturally prompted to address the question of the millimeter detect ability of the many radio-quiet QSOs of comparable luminosity in the redshift range 1-3. We have begun an exploratory programme to observe luminous radio-quiet QSOs with z rv 1 - 3.5, We have detected a strong source, QI230+1627, with z = 2.7, and we have three 30" tentative detections which should be reobserved with a better sensitivity.

3. Discussion

Our new results raise the number of 1.25 mm detections of radio-quiet QSOs with z > 4 from 1 to 6, and for those with z > 1 from 3 to 9. For z > 4, our study is relatively systematic since we observed with a good sensitivity about half of the objects known at the time of our observations. Accordingly, the general trends of the millimeter emission of the optically known z > 4 QSOs can be inferred. Among color identified samples such as APM, the detection rate with an r.m.s. rv 1 - 1.5mJy, i.e., with a detection limit rv 3 - 5 mJy, should be in between 20% and 30%. Sources with S1.25 > 10 mJy and even> 5 mJy are rare, with proportion rv 7% and 10-15%, respectively.

The question of the frequency of a strong amplification by gravitational lensing of such objects remains a major issue. There is only one clear case known of strong lensing among the six mm detected objects with z > 4, namely BR0952-0115. However, sensitive visible and near-IR searches are needed to definitely discard systematic effects of lensing.

Some trends begin to emerge from the relations of the strength of millimeter emission with other characteristics of the sources. All the millimeter detected QSOs are among those which have the largest blue-UV luminosities. Indeed, all the QSOs we have detected pertain to the color selected APM sample which privilegiates high luminosities. Not independent of a large luminosity is the fact that the visible spectra (Storie-Lombardi et al. 1996) of the millimeter detected QSOs have weak and broad emission lines. The presence of broad absorption lines (BAL) seems to increase the chance of a positive millimeter detection, but does not warrant it. Most of the detected sources have a visible spectrum somewhat reddened, as expected from the presence of dust. However, there is no case of very strong reddening among the known QSOs with z > 4.

Waiting for submillimeter observations, we do not know the millimetersubmillimeter spectral index of the new detected sources. However, as for BR 1202-0725 (McMahon et al. 1994, Isaak et al. 1994), it is probable that

334 A. OMONT ET AL.

it is large, 3-4, and characteristic of dust emission. The mass of dust Md is probably at least"" 108 M0 for all the detected sources in the absence of lensing. If such an absence oflensing is confirmed in the strongest millimeter sources, they are among the objects with the largest mass of dust known. There is no indication about the dust temperature, and hence the far infrared luminosity. However, it is likely that the latter is at least comparable to that of the strongest IRAS hyperluminous galaxies. Sub millimeter and ISO observations will be essential to confirm that and to bring information about the dust temperature and heating, likely performed by the QSO UV radiation.

In addition to ISO, it is clear that this new field will soon strongly benefit offurther observations in the millimeter and sub millimeter ranges, especially with the advent of new facilities. All the sources detected at 1.25 mm at IRAM should be detectable at 0.8 mm with the present equipment of JCMT in very good weather conditions, thus providing the spectral index. It is hoped that a systematic study at IRAM on the whole z > 4 sample will be soon completed. In addition, it is hoped that a more sophisticated data analysis could significantly reduce the sky noise. Both points will improve the information on the millimeter luminosity function, and in particular it should give a reliable value for the average flux of the undetected sources. The advent of lower temperature receivers, in particular SCUBA, will bring another large gain of sensitivity, allowing probably to detect most presently known QSOs with z > 4.

The impact of such technical developments could be still more important at smaller redshift, in the range z "" 1 - 3.5, where the number of bright radio-quiet QSOs known is much larger. We have begun to explore this redshift range, using criteria which seem to favor mm detections, established from our z > 4 detections. The first results on weak line sources are encouraging as reported above. However, it seems that, as expected, the detection is not easier at z = 2 than at z = 4; it could even be more difficult. A vigourous systematic programme should be pursued in this range with a priority towards weak line sources, red spectra and BALs.

It is rather certain that such large amounts of dust imply giant starbursts, at least comparable to the most hyperluminous IRAS galaxies. The search for CO is clearly essential, to confirm the properties of the interstellar medium. One should also expect to be able to detect the C+ line, when z > 4.3 to allow the redshifted line to be in the 0.8 mm atmospheric window . Both searches have been presently mostly negative (see e.g. Isaak et al. 1994, Omont et al. in preparation), except for CO in the exceptional lensed objects H1413+1143 and IRAS F10214+4724 (see e.g. Barvainis, Radford and Scoville et al. in these proceedings). Such searches should be actively pursued when accurate values for the redshift are available. It should be

1.25MM DETECTION OF VERY HIGH Z RADIOQUIET QSOS 335

noted that, given the weakness and the broadness of the visible lines in the best candidates, the near-IR should be the best domain for the accurate determination of z. Some redshift ranges could be given a particular priority because at least one main IR lines is then in an atmospheric window. Let us also note that the ratio of the CO to the continuum millimeter intensity should slightly decrease with z.

References

Andreani P., La Franca F., Christiani S. (1993), MNRAS, 261, L35 Irwin, M.J., McMahon, R.G. & Hazard, C., (1991), In: The Space Distribution of Quasars,

ASP Conference Series, Vol. 21, p117 (ed.) D. Crampton Irwin, M.J., McMahon, R.G. & Hazard, C., (1996), in preparation Isaak K.G., McMahon R.G., Hills R.E., Withington S. (1994), MNRAS, 269, L28 Kreysa, E.,(1993), In: Proc.Int.Symp.on Photon Detectors for Space Instrumentation,

edited by T.D.Guyenne, ESA/ESTEC Noordwijk McMahon R.G., Omont A., Bergeron J., Kreysa E., Haslam C.G.T. (1994), MNRAS,

267,L9 Omont, A., McMahon, R.G., Cox, P., Kreysa, E., Bergeron, J., Pajot, F. and Storie

Lombardie, L.S., (1996), to be submitted to Astron. Astrophys. Schneider D.P., Schmidt M., Gunn J.E. (1991), Astron. Journal, 101, 2004 Storie-Lombardi, L.S. et al. (1996), in preparation

RADIATIVE TRANSFER MODELS FOR IRAS FI0214+4724 AND OTHER HYPERLUMINOUS GALAXIES

S. M. GREEN AND M. ROWAN-ROBINSON Imperial College of Science, Technology and Medicine, Blackett Laboratory, Prince Consort Road, London SW7 2BZ

1. Introduction

At the time of its discovery, IRAS F10214+4724 was thought to be the most luminous object known (Rowan-Robinson et al. 1991), but it has recently become clear that gravitational lensing is at least in some part responsible for its extremely high luminosity of 5 X 1014hs02 L0 (Q = 1, Ho = 50 hso km s-1 Mpc -1). Models for the gravitational lensing suggest magnifications ranging from""' 10 (Graham & Liu 1995) to greater than 50 (Broadhurst & Lehar 1995). The higher magnification would make the luminosity and mass of IRAS F10214+4724 typical of ultraluminous IRAS galaxies, whereas the lower magnifications would still make IRAS F10214+4724 an extreme object, and a member of the hyperluminous class of galaxies. These galaxies have a bolometric luminosity in excess of 1013 L 0 , which is concentrated almost entirely in the far-infrared. Attempts have been made to explain IRAS F10214+4724 in terms of either a massive galaxy in the process of formation, or a quasar heavily enshrouded in dust. In this paper we present radiative transfer models for these two scenarios, to investigate both the primary source of the luminosity, and the implications of lensing magnifications of""' 10 or ""' 100.

2. Starburst Models

IRAS F10214+4724 shows many of the signatures of a starburst. The optical and radio emission are extended with similar morphologies in each band (Lawrence et al. 1993; Elston et al. 1994). The gas mass from CO observations (Solomon et al. 1992) is M(H 2 ) = 4 x 1010(10/ M)hsi M0 after allowing for gravitational lensing with magnification M. The Ha, radio and bolometric luminosities can all be explained by starburst models with ages

337


338

1000

12

S. M. GREEN AND M. ROWAN-ROBINSON

100

1000

10

, •.... -. -::r',·,. t T -·----'~·<:<i··-·T\

. , , , , , '- \ \, \ . , , , . , . ,

, \

' ..

10

, , , , , , , , , , , , , . ,

.. '

Figure 1. Starburst models for IRAS F10214+4724. The dashed line represents Model 'B' of Rowan-Robinson et al. (1991) with T1 =1000K and Tuv = 100. The dot-dash line shows the effect of reducing Tl to 400K with the other parameters the same. The model shown by the solid line again has Tl = lOOOK, but has an optical depth of Tuv = 800.

of 107 to a few X 108 years (Elbaz et al. 1992; Rowan-Robinson et al. 1993; Mazzei & De Zotti 1994). Rowan-Robinson et al (1993) considered detailed radiative transfer models for the infrared spectrum of IRAS FI0214+4724. A pure starburst model (their Model 'B') fitted the submillimetre and infrared observations very well, but subsequent observations by Telesco (1993) at 20pm have provided an upper flux limit below the predictions of this model (See Fig. 1). We now consider new models which fit this upper limit.

The spectrum of the starburst clouds is determined by an exact solution of the radiative transfer equation which treats absorption, scattering and thermal dust emission in a multi-grain dust medium at radiative equilibrium. This code has previously been applied to starbursts by RowanRobinson & Efstathiou (1993). The main parameters for the models which determine the infrared spectrum are the radial optical depth (T uv) at 100A, the ratio of the source radius to the outer radius of the cloud (rs/r2) and the temperature (TI) above which dust is destroyed at the edge ofthe cloud cavity.

Initially, a value of Tl = 1000K was adopted, as has successfully been used in previous starburst models. The parameter rs/r2 determines the wavelength at which the Rayleigh-Jeans tail turns over, and is therefore constrained by the 350 pm detection and 100 pm upper limit to lie between 10-6

RADIATIVE TRANSFER MODELS FOR IRASF10214+4724 339

and 10-7• This constrained the dust mass to (1.2-3.1)X108(1O/M)hsJ M0 •

The parameter Tuv influences the amount of near-infrared emission. By increasing Tuv to f'V 400 or more, the near-infrared upper limits can all be satisfied but the 60 I'm emission then falls below the detection. Thus, no model with Tl = 1000 K can simultaneously fit all the near-infrared to millimetre observations. However, if Tl is reduced to 400K, the near-infrared emission from the hot grains is suppressed below all the near-infrared upper limits, and all the observations can be fitted (See Fig. 1).

In the starburst model, the radio emission is due to the relativistic electrons released in supernova explosions in the starburst clouds. The high farinfrared to radio ratio found for IRAS FI0214+4724 (Rowan-Robinson et al. 1993) and the young age suggested by stellar evolution starburst models suggest that cosmic ray electrons from supernovas have not yet diffused into the general galactic magnetic field. The radio emission is therefore expected to be cospatial with the starburst clouds. If the model starburst clouds are arranged to be cospatial with the radio extent observed by Lawrence et al. (1993), it is found that the models with optical depth Tuv = 100 require large numbers of clouds (f'V 70) along the line of sight. In such a case extinction of one cloud by another would be very significant, and the 60 I'm flux would fall below the detection. Such a dense distribution of clouds is more accurately modelled by the high optical depth starburst models. These models, such as the model shown in Fig. 1 with Tuv = 800, require only f'V 3 clouds along the line of sight to match the radio extent, but fail to fit the 60 I'm detection. Comparing the Tuv = 800 model to the CO observations, it is found to be consistent with the gas density measured by Solomon et al. (1992), if there are a total of f'V 30 star forming complexes. Therefore, starburst models can be reconciled with the submillimetre, radio, near-infrared and CO observations but fail to fit the 60 I'm detection (Green & Rowan-Robinson 1995).

3. Flared Disc Seyfert Models

The near-infrared to ultraviolet emission-line spectrum (Rowan-Robinson et ale 1991; Elston et ale 1994), the high ultraviolet polarisation (Lawrence et ale 1993) and the X-ray emission from IRAS F10214+4724 (Lawrence et ale 1994) are all consistent with an obscured quasar. However, Seyfert models investigated by Rowan-Robinson et ale (1993) provided very poor fits to the infrared observations, and were incapable of fitting the submillimetre fluxes. A reasonable fit was provided by an embedded quasar model using an anisotropic sphere of dust, but this implied an unfeasibly large size for the 'disc'. Our new Seyfert models have the 'flared disc' geometry described by Efstathiou & Rowan-Robinson (1995). The models are parameterised

340 S. M. GREEN AND M. ROWAN-ROBINSON

by the temperature (TI) above which dust is destroyed, the ratio of the inner radius to the outer radius of the disc (rl/r2) and the radial optical depth (T uv ). The opening half angle of the disc was taken to be 30°. The optical depths considered were consistent with the ROSAT observations of Lawrence et at. (1994) and the source optical spectrum was matched to observations.

With the models normalised to the optical observations, all the disc models failed to produce sufficient submillimetre emission to match the observed fluxes. Decreasing rdr2 increased the submillimetre emission, but this also increased the size of the disc. The extent which would be observed for a model fitting the submillimetre observations would be f'V 12", which is greater than the CO extent. HST observations (Eisenhardt et at. 1996) suggest that the optical-ultraviolet emission is gravitationally lensed by more than the far-infrared. This increases the intrinsic submillimetre to optical flux ratio, and so requires even larger disc sizes. Decreasing the maximum dust temperature to 400 K, so that dust at larger distances is illuminated directly by the source (as in warped discs), and decreasing the opening half angle of the disc can reduce the size of disc by factors of a few. A uniform density dust profile which does not fall with radius (See poster by Granato & Danese) also decreases the disc size, but such discs still exceed the CO size. The f'V 80pc torus with a magnification factor as high as 50, proposed by Broadhurst & Lehar (1995), cannot account for the observed submillimetre emission.

It remains to be considered what proportion of the mid-infrared flux a flared disc can contribute. By increasing Tuv and decreasing TI , the nearinfrared emission can be reduced below the upper limits. The viewing angle can then be varied to fit simultaneously the optical and 60 J-Lm detections. The Seyfert disc model shown in Fig. 2 can account for all the observations at observed wavelengths below 80 J-Lm.

4. IRAS 09104+4109

Of the other hyperluminous galaxies, only IRAS 09104+4109 has sufficient infrared observations (Kleinmann et at. 1988) for comparison to these models. The Rayleigh-Jeans tail for this object is constrained by the IRAS 60 J-Lm and 100 J-Lm observations. All the observations can be fitted by the flared disc Seyfert models described above (Green & Rowan-Robinson 1995). This supports the ASCA observations (Fabian et ai. 1994) which indicate an embedded quasar.

RADIATIVE TRANSFER MODELS FOR IRAS F10214+4724 341

.'

: I

12

1000 100

--, , ,

, , , , , , , , , , , , , , , , " " " " " " ,

10

. .

Figure 2. A composite starburst and AGN disc model for IRAS F10214+4724. The dashed line which fits the submillimetre observations is the high optical depth (Tuv = 800) star burst model shown in Fig. 1. The dot-dash line, with the higher flux at observed wavelengths below 80 Jtm, is a flared disc Seyfert model with Tuv = 100, r1/r2 == 0.01 and T1 = 500 K. The solid line shows the resultant emission from the two components.

5. Conclusions

The observed submillimetre emission in IRAS F10214+4724 cannot be produced from within the'" 80 pc region required for gravitational lensing magnifications of ;::50. Instead more extended emission and a lensing magnification of '" 10 is required. To account for the 60/-Lm detection and the sub millimetre emission simultaneously requires a composite starburstAGN model (See Fig. 2). This model has a bolometric luminosity of 5 x 1013hso2(10/ M) L0' 40% of which originates in the starburst. However, IRAS 09104+4109 can be fitted by a Seyfert disc model alone.

References

Broadhurst, T. and Lehar, J. (1995) Astophys. J. Lett. 450, L41 Efstathiou, A. and Rowan-Robinson M. (1995) Mon. Not. R. Astra Soc., 273, 649 Eisenhardt, P.R. et al. (1996) Astophys. J. 461, 72. Elbaz, D. et al. (1992) Astr. Astraphys.,265, L29 Elston, R., et al. (1994) Astron. J., 107, 910 Fabian, A.C., et al. (1994) Astrophys. J. 436, L51 Graham, J.R. and Liu, C. (1995) Astraphys. J. 449, L29 Green, S.M. & Rowan-Robinson, M. (1995) Mon. Not. R. Astra Soc., submitted. Kleinmann, S.G. et al. (1988) Astophys. J. 328, 161

342 S. M. GREEN AND M. ROWAN-ROBINSON

Lawrence, A. et al. (1993) Mon. Not. R. Astro Soc., 260, 28 Lawrence, A. et al. (1994) Mon. Not. R. Astro Soc., 266, L41 Mazzei, P. and De Zotti, G (1994) Mon. Not. R. Astro. Soc., 265, L5 Rowan-Robinson, M. et al. (1991) Nature, 351,719 Rowan-Robinson, M. et al. (1993) Mon. Not. R. Astro. Soc., 261, 513 Rowan-Robinson, M. and Efstathiou, A. (1993) Mon. Not. R. Astro Soc., 263, 675 Solomon, P.M., Downes, D. and Radford, S.J.E. (1992) Astrophys. J. Lett. 398, L29 Telesco, C.M. (1993) Mon. Not. R. Astro. Soc., 263, L37

IR AND X-RAYS FROM IRAS F10214+4724: A HIDDEN AGN?

G.L. GRANATO, A. FRANCESCHINI

Osservatorio Astronomico di Padova, Italy

AND

L. DANESE

International School for Advanced Studies, Trieste, Italy

Abstract. Under the assumption that a hidden AGN contributes a major part of the bolometric flux of the primeval object IRAS FI0214+4724, we argue that sensitive X-ray observations would provide crucial information on the metal enriched gas present in the circumnuclear region.

1. Introduction

The origin of the prodigious luminosity of IRAS FI0214+4724 ('" 3 X

1014h-2 L0 ) has been debated since its identification as an high-redshift primeval object at z = 2.286. However it recently has been recognized that the source is gravitationally lensed by an intervening galaxy, a circumstance which decreases its true power by a factor of'" 10 to 50 (see Eisenhardt et al. 1996 and references therein).

The AGN origin for the bulk of the luminosity is supported by the resemblance of its optical and UV (rest frame) emission-line spectrum to those of Seyfert 2 galaxies, by the small size (:s 0.5h- 1 kpc) implied by lensing, and by the extremely large barionic mass required if the luminosity were produced by stellar processes.

Altogether, the lensing makes IRAS FI0214+4724 a less extreme object than implied by the first observations, both in terms ofluminosity as well as of mass in stars and in a diffuse medium. The parameters, after correction for lensing, would be closer to those of typical ultraluminous IR galaxies in the local universe.

343


344 G.L. GRANATO, A. FRANCESCHINI AND L. DANESE

In the context of unified theories for AGNs this object might be the high redshift and high power counterpart of the type-2 Seyfert galaxies in which our direct view of the nucleus is hidden by a torus-like distribution of gas and dust, responsible for the IR emission. In the following we will assume that the emission of a dusty torus dominates the observed IR spectrum.

2. IR and X-Rays as Probes of Matter around the AGN

Dusty tori around active nuclei reprocess also the nuclear X-ray emission by photoelectric absorption and Compton downscattering (Krolik et al. 94, Granato et al. 1995). Therefore one can expect that X-ray and IR emission may at some extent be used to probe the torus structure. To do this we will compare available data and models.

We have reproduced the observed spectral energy distribution (SED) of IRAS F10214+4724 using the torus-like model described in Granato & Danese (1994), which successfully fits the available data of both broad and narrow-lined local AGN (Seyfert Is and 2s) in the context of unified schemes. A possible fit is shown in Fig. 1. The model parameters are poorly constrained, due to the paucity offlux detections in the IR broad-band spectrum. However, the optical thickness Te of the torus at 3000 A is bound to be within say 100 and 400 to fit the IR portion of the SED. Correspondingly, the total dust mass of the torus is of order of 10778 h-2 m- I M0' where m is the assumed lensing amplification. The inner boundary of the structure is at ~ 5 m- I / 2 pc.

In the following we will estimate the X-ray flux of IRAS F10214+4724 under the assumption that this object is a "type-2 QSO", the high z analogue of the Seyferts 2, in the context of unified models. We therefore consider the effects on the X-rays transmission of a torus-like gas distribution with hydrogen column density NH surrounding the nucleus. This gas may be entirely associated with the dusty torus responsible for the IR bump, but can also be completely or in part independent. In addition, the dust to gas ratio in the torus is likely to be different than the typical Galactic value, thus NH is not so obviously linked to the inferred dust optical thickness of the torus.

We have recently developed a code which solves for the radiative transfer equation for X-ray photons in a torus-like gas distribution, taking into account the anisotropy and incoherence of Compton scattering. We checked our results against those provided by more time-consuming Monte Carlo simulations, finding an excellent agreement up to energies of hundreds of keY. The torus absorbs the lower energy photons along obscured directions, and if NH is large produces a reflection bump along unobscured directions. The observed flux between VI and V2 is given by

IR AND X-RAYS FROM IRAS FI0214+4724

.... .... o ......

104

N 10~ I

..c:: ........

110

1

10~

101~

10

\ \ \

1

\ xX

0.1

\.-.-.-.-.~~ 'So:

1013 1014 1015

Vern [Hz] 1016

345

Figure 1. A fit to the IR rest frame SED of IRAS FI0214+4724 with a dust enshrouded QSO model. The dusty torus has an equatorial optical thickness at 3000 A Te = 250, the ratio between the outer and inner radius is 700, and the density depends only on the polar angle (ex: exp( -6.2 cos2 El). Note that the same model observed face-on reproduces the SED of the Cloverleaf QSO.

(1)

where Ve = v(l+z) and L~~~2e is the luminosity observed along unobscured lines of sight, and PVl e ,V2e (NH) is the thickness depending ratio between the flux emerging along lines of sight with column density NH and the flux coming along unobscured directions, as derived from the solution of the radiative transfer equation. X-ray luminosity can be expressed in term of the luminosity at 5 keY as L~s~ = f(NH)LsQkSOv, where f(NH) is also

Ie, 2e e

derived from the solution of the radiative transfer (the dependence on NH arises from the reflection bump).

The monochromatic flux observed at ground at 60!-lm is given by

(2)


TABLE 1. X-ray fluxes expected for different values of the hydrogen column density.

NH F(0.5 - 2 keY) F(2 - 10 keY) cm-2 ergs-1 erg S-l

0 8.9 X 10-13 1.2 X 10-12

1023 .1 2.7x10-13 1.1 X 10-12

1024 .1 1.5 X 10-14 6.5 X 10-13

1024 .6 3.5 X 10-15 4.6 X 10-13

1024 .8 1.5 X 10-17 1.5 X 10-13

1025 .1 2.4 X 10-18 2.3 X 10-14

1025 .6 2.6 X 10-15

where P18(Te , 8) is the ratio in the rest frame between the monochromatic flux at ~ 18 j.lm (60/ (1 + z)) emitted by the dusty torus with optical thickness Te along the obscured direction we see it (defined by the polar angle 8) and that emitted along a typical unobscured direction.

Combining the above equations we get finally

F( ) _ FII (60j.lm)Rf(NH)P lIl e ,1I2.(NH) VI, V2 -

P18(Te , 8) (1 + z) (3)

where R = L~SO(5 keV)/ L~SO(18 j.lm). Barcons et al. 1995 find an average fll(5 keV)/ fll(12j.lm) = 1.4~~:! X

10-6 for a sample of 54 Seyferts 1 (error ranges corresponding to 95% confidence). Adopting the mean IR SED derived by Granato and Danese (1994), this translates into an X-ray to IR ratio of R ~ 0.8 X 10-6 . The observed IR SED is reproduced by models having 100 ~ Te ~ 400 and the required combination of the parameters Te and 8 yelds P18(Te ,8) rv 0.5 with possible deviations never exceeding 20%. The monochromatic flux FII (60j.lm) is 1.9 X 10-24 ergs- l Hz-I.

The expected soft and hard X-ray fluxes are reported in Table 1 as a function of the gas column density, assuming solar abundances. Lawrence at al. (1994) measured with ROSAT a flux upper limit (or marginal detection) corresponding to F(0.5 - 2keV) ~ 8 X 10-15 ergs- l , which requires NH 2:: 2 X 1024 cm-2 . At higher energies, long exposure observations have been planned with ASCA. In particular an observation is planned with 40 Ksec exposure, corresponding to a flux limit of F(2 - 10 keY) rv 10-13 erg S-l.

From Table 1 it is apparent that with NH implied by the 0.5 - 2 keY we may expect detection of IRAS F10214+4724 by ASCA, unless the nucleus is covered by a quite larger hydrogen column density (NH 2:: 7 X 1024 cm-1 ).

IR AND X-RAYS FROM IRAS FI0214+4724 347

Since the typical dimensions of the X-ray emitting regions are much smaller than the mid-IR emitting regions (around 200 - 500m-1 pc in our models), the magnification may be more effective for the X-ray than for midIR emission by a factor of 2 to 5. As a consequence using the value found for Seyfert 1 galaxies for the ratio R of the X-ray to 18 J-Lm luminosity, may result in an underestimate of the fluxes in Table 1 by the above mentioned factor. Hence the values of the column density inferred by using upper limits to the X-ray fluxes may be slightly underestimated.

The inferred value of column density, though very large, is not unusual in local narrow-line AGNs. A similar column density has been claimed by (Iwasawa et al. 1993) for NGC 4945 and a lower limit of the same order holds for NGC 1068. However, it is easy to see that it cannot be provided by gas associated with the IR emitting dust torus, if a standard gas-to-dust ratio is adopted. For a galactic mixture of gas and dust we have in fact NH ~ 1.2 X 1021 Te cm-2, which translates the optical depth Te < 400 of the dusty torus found above into a column density of NH < 5 X 1023 cm -2.

On the other hand, the warm molecular gas responsible for the CO emission has NH '" 3 x 1013M(H2)/r2[kpc] cm-2, where M(H2) is the total mass of molecular gas in solar units and r the radial extent of the gas distribution. Solomon et al. (1992) derived M(H2 ) ~ 1011 h-2 from the CO luminosity and a lower limit to r of 0.8 h-1 kpc. For such a compact emitting region, the observed gravitational lensing decreases the mass estimate by the magnification factor m ;:::: 5, while for a more extended region the lensing becomes negligible but the r- 2 factor depresses the column density. As a result we get in any case NH ~ 1024 cm-2. Assuming Galactic column density to extinction ratio this implies Te ~ 800. Note that a change in the metallicity of the gas with respect to the solar abundances would affect the CO-to-H2 conversion factor, but would also affect by a similar factor the hydrogen column density entering into the X-ray transmission.

In conclusion the upper limit to the 0.5 - 2 keY flux obtained with ROSAT (Lawrence et al. 1994) is marginally compatible with column density inferred from IR and CO observations and would imply detection with a 40 Ksec exposure by ASCA. On the other hand non-dectection by ASCA at flux limit of F(2 - 10 ke V) '" 10-13 erg s-l would imply a column density NH ;:::: 4 X 1024 cm-2, quite large but already claimed for Seyfert 2 galaxies, blocking the hard X-ray flux. Such absorber can NOT be explained by gas associated with the putative nuclear dust torus, NOR by the CO emitting gas. To preserve the hidden quasar hypothesis we have to admit that either X-ray absorption may originate in a gas-rich region inner to the dust sublimation radius, or the fraction of heavy elements condensed into dust grains is much smaller in the torus than in the ISM of our Galaxy.


References

Barcons, X., Franceschini, A., G., D., Danese, L., & Takamitsu, M. 1995, ApJ, in Press Eisenhardt, P. R., Armus, 1., Hogg, et al. 1996, ApJ, In Press Granato, G. L., & Danese, L. 1994, MNRAS, 235 Granato, G. L., Danese, L., & Franceschini, A. 1995, in preparation Iwasawa, K, Koyama, K., Awaki, H., et al. 1993, ApJ, 155 Kralik, J. H., Madau, P., & Zycki, P. T. 1994, ApJ, L27 Lawrence, A., Rigopoulou, D., Rowan-Robinson, M. et al. 1994, MNRAS, L41 Solomon, P. M., Downes, D., & Radford, S. J. E. 1992, ApJ, 398, L29

GAS AND DUST IN HIGH REDSHIFT RADIO GALAXIES

P. J. MCCARTHY Carnegie Observatories 813 Santa Barbara St. Pasadena, CA 91101 USA

1. Introduction

For roughly a decade radio galaxies provided us with our only view of galaxies at redshifts beyond one. In recent years two significant changes have occurred. Firstly, there are now a significant number of radio quiet galaxies known at redshifts between 1 and 4 and these provide more useful probes of galaxy evolution and cosmology. Secondly, we have learned that radio galaxies are not simple systems and that non-stellar light may contribute significantly to their continua in the rest-frame UV. Radio galaxies are still of interest in their own right and they are quite relevant to the topic of this workshop. Radio galaxies offer us our best examples of large scale line-emitting gas at interesting redshifts. The may also ultimately provide us with a sensitive probe of dust in the early universe.

2. The Ionized Gas

2.1. SPATIAL DISTRIBUTION

There are now emission-line images of a large number of radio galaxies spanning a wide range of redshifts. A significant fraction (rv 75%) of high power sources show resolved narrow emission lines and the visible extents can reach well beyond 100 kpc (e.g. McCarthy et al. 1995; van Ojik et al. 1996). The morphologies are complex and varied but share one common trait, the largest extent ofthe visible line emission is closely aligned with the radio source axis (McCarthy et al. 1987). In Fig. 1 I show an example, a Lya image of a z = 2.4 radio galaxy. This object has high surface brightness Lya emission with an extent of 12" (rv 120 kpc). By combining measurements from the Ha, [0 III]5007, and [0 IJ]3727 images of 3CR galaxies from Baum et al. (1988) and McCarthy et al. (1996) with those from Lya images of

349


350 P. J. MCCARTHY

Figure 1. Contour plot of the Lya emission associated with the galaxy MRC 2104-242 at z = 2.42. The area shown is 20" x 20". There is a bright star to the NW.

various radio galaxies from the literature (e.g. McCarthy et ai. 1993; van Ojik et ai. 1996; Dickinson et ai. unpublished) I have made a compilation of the maximum extent of the emission lines at a fixed rest-frame surface brightness level. This involves some normalization for the relative strengths of the various emission lines used. The result is shown in Fig. 2. One can see from Fig. 2 that there are large (> 100 kpc) emission line nebulae present at all redshifts beyond z fV 0.3 and that the maximum size at a fixed isophot (rest-frame) does not change dramatically with epoch.

There have been suggestions in the literature that the z rv 2 radio galaxies have smooth low surface brightness halos that extend to very large radii. While most of these claims are likely to be correct, it is a difficult measurement to make. Imaging observations require very accurate flat-field corrections and sky subtraction to detect the low surface brightness emission at large radii. The spectroscopic measurements (e.g. van Ojik et al. 1996) are' more robust, but they contain less spatial information. In Fig. 3 I plot a growth curve for the Lya emission associated with B2 0902+34 (Lilly 1988) derived from an image obtained by Dickinson et al. at Lick Observatory. The Lya flux is still increasing at a radius of 15", corresponding to a diameter of some 300 kpc.

HIGH REDSHIFT RADIO GALAXIES 351

2.2. IONIZATION AND PHYSICAL CONDITIONS

The physical conditions in the emission line regions at moderate to large redshifts are poorly constrained. Since the standard diagnostic line ratios are redshifted beyond the reach of CCDs and spectrographs, the density and temperature of the line-emitting gas.must be measured indirectly. The usual approach is to make some assumptions regarding the shape and intensity of the ionizing spectrum and compute a grid of photoionization models and choose those models which best represent the observed line ratios. The implied ionization parameter and the distance from the ionizing source then gives one an estimate of the density of the line-emitting gas. Typical numbers from this type of analysis are: ne '" 10 cm -3, Iv = 10-5 , and M(H II)", 108 M0 (e.g. McCarthy et at. 1990, Heckman et al. 1991). These analysis rest on the belief that the gas is ionized by a central source with the a roughly power-law spectrum. The case for nuclear ionization in the extended emission lines is made in Robinson et al. (1987) and elsewhere. Local photoionization and auto-ionizing shocks are not ruled out and may be the preferred ionization mechanisms in a few objects (e.g. Tadhunter et al. 1987).

The high red shift objects have a remarkable homogeneity in their UV line spectra (McCarthy 1993; R6ttgering et al. 1995), suggesting that either we do not fully understand the processes that produce the UV emission lines in these objects or that the physical conditions in these objects are remarkably uniform. In Fig. 4 I show the rest-frame UV spectrum from a composite of some'" 80 radio galaxy spectra. The strong UV emission lines are marked. The hallmark of the high red shift radio galaxies is their strong Lya emission, although there are now known to be objects with little or no Lya (see Sect. 3.2 below).

2.3. KINEMATICS

Long-slit spectra of the extended emission lines provide information on the kinematics of the ionized material. Early spectra of the 3CR galaxies by Spinrad and Djorgovski (1984) revealed velocity fields with amplitudes of several hundred km S-1. Baum, Spinrad, and I (McCarthy, Baum and Spinrad, 1996 in press) have recently assembled a large set of velocity field measurements for the 3CR and 1 Jy radio galaxies and Baum and I have compiled all of the velocity measurements at low and high redshifts that we could find in either the literature or our own data sets.

In Fig. 5 I show plots of the velocity fields of two objects that are typical of the large emission line regions. While the amplitudes are large, one cannot easily identify the motions as rotation or smooth infall/outflow in many cases, in contrast to the lower redshift objects which could usually be

352 P. J. MCCARTHY

1000 ...........

CJ • O-t 300 .. ~ • "--" • -+o-J 100

. . .. , ~ • • t ~ .. ,. ... Q) • . . " ...... -+o-J

>< 30 • • • • w • • •• • I· S

• •• • • 10 • ;::l • S

, ...... • >< 3 • 2" cd ~

1 .0 1 .03 .1 .3 1 3

Redshift

Figure 2. The linear sizes of the emission line regions for 3CR and other radio galaxies for Ho = 50 km S-1 Mpc-1 , qo = 0 at a fixed rest frame isophote. The solid line is the linear size of the worst case resolution limit.

classified as dominated by either either rotation or turbulence (Tadhunter et at. 1989; Baum et at. 1992). In Fig. 6 I plot the maximum amplitude seen in each object against redshift. There is a clear increase in the velocities at higher redshifts. The meaning of this trend is still unclear, but I speculate that the increase in velocities for z '" 0.5 is sufficiently close to the redshift at which radio galaxies first (or last, really) appear in rich clusters (Hill and Lilly 1991) to suggest that we are seeing signs of dynamical interactions between pairs or groups of galaxies. The amplitudes of the velocity fields at high redshift are comparable to the velocity dispersions of the richest clusters today.

3. Is There Dust in Distant Radio Galaxies?

3.1. THREE ARGUMENTS AGAINST DUST ...

Nearly all analyses of the colors and luminosities of radio galaxies have explicitly ignored the possible effects of dust (e.g. Lilly and Longair 1984;

CIl

S 0

"'-.. 2.5 0 Q) CIl

"'-.. 2 bD ~ Q)

1.0 1.5 -' I 0 ~

1 "'-..

>< .5 ;::J .--4

~

t$ a >. ~

HIGH REDSHIFT RADIO GALAXIES

•

0

• •

• • •

82 0902+34 Ly a

5 10 15 20 Radius (acrseconds)

353

•

Figure 9. The growth curve (F(r < R» for B20902+34. The image and the growth curve were obtained by Mark Dickinson using the Lick 3 m and a narrow band filter. The curve show that the Lya emission can be traced to a 30" diameter.

Dunlop et ai. 1989). This was done in part to avoid adding yet another poorly constrained parameter to the already under-constrained models of the spectral evolution of radio galaxies. There are a number of arguments made in defense of the dust-free approach and three that have been widely cited are as follows: 1) The strong Lya emission that is seen in radio galaxies at z > 1.6 would not escape from a galaxy with even a relatively low abundance of dust since resonant scattering by H I will make the path length for Lya photons many times the geometric path length and hence will lead to absorption by dust. 2) The UV continua of radio galaxies are essentially flat in fll units, roughly as blue as an extreme starburst galaxy. This is taken as evidence of negligible extinction at 1500A, and a fairly strong constraint on the dust content. 3) The small scatter in the K - z relation (Lilly 1989) is also cited as an argument against significant extinction and hence low dust content in radio galaxies.

354 P. J. MCCARTHY

8 .---, .---, ...... .---, .........

6 ...... > ......... :::: ...... ::::<:Q. .-...... cu :::: cu ...... ....... QO ...... cu .-~

Z a Z a ~~

a '--' '--' '--' '--' '--'

4

2

0 ::::. 2500 3000 3500 4000 4500 5000 5500 ~

4

......... 3 ~ > .- ,.--, .-

> .- ....... ......... cu >. cu ...... .......

Z .- ...... .-~ u ::r:: u u '--'

2

1

0 1000 1500 2000 2500

Rest A

Figure 4. Composite radio galaxy spectrum obtained by averaging the spectra of 65 radio galaxies with 0.3 < z < 3.5. The strong emission lines are labeled.

3.2. AND WHY THEY ARE WRONG.

I will discuss each of these arguments in turn and argue that in the light of new data and a better understanding of the physical processes that are likely to be operating in these systems, these arguments are not as compelling as once thought.

Strong Lya: Resonant scattering by a neutral or partially ionized halo is the mechanism that make Lya such a powerful tool for detecting even


400

-200

-5 0 5 arcseconds

Figure 5. The velocity field for the Lya emission associated with MRC 2104-242. The slit was oriented along the two bright peaks of Lya emission shown in Fig. 1.

,-....

~3000

] 1000 '-...-./

300 • 100

30 .03

• • •• 1-) .. . -. . • • • •• •

.1 .3 1 Redshift

• •

3

Figure 6. The maximum amplitude of the measured velocity field for 3CR and MRC radio galaxies as a function of redshift. There is a sudden increase in the peak velocities at z '" 0.6.

356 P. J. MCCARTHY

minute amounts of dust. Without the benefits of resonant scattering Lya is no different from other UV emission lines (e.g. Hell 1640), except that it is at a somewhat shorter wavelength. Resonant scattering is thought to contribute significantly to the quenching of Lya emission in present day starburst galaxies but this result cannot necessarily be directly applied to the radio galaxies. The issue here is the width of the Lya emission line at is enters the neutral or partially ionized medium. If the line width is larger than a few hundred kms-1 , the most of the Lya photons are in the wings of the absorption profiles of the neutral atoms and so do not behave like resonant photons. The typical z '" 2 radio galaxies has FWHM(Lya) '" 1000 km s-1, making resonant scattering less important than often believed.

In recent years a number of objects have been found in which Lya is being selectively destroyed. Two of these are powerful radio galaxies (MG1019+0535, Dey et al. 1994; TX0211-122 van Ojik et al. 1994) and one is the famous IRAS source F10214+4724 (Rowan-Robinson et al. 1991). While IRAS F10214+4724is not a powerful radio galaxy, its UV spectrum is so similar to the radio galaxies that I will include it in much of the discussion that follows. Each of these objects has strong UV emission from C IV1549, He 111640, C III] 1909, but Lya is either absent or suppressed by a factor of 5 - 10 from the typical radio galaxy line ratios (e.g. McCarthy 1993). While the dust masses needed to suppress Lya are not enormous, the large far-IR luminosity of IRAS F10214+4724 implies that there is quite a bit of extinction in the region where the FUV continuum is being produced. The radio galaxies may be similar in this respect, but since they apparently do not benefit (or suffer) from gravitational amplification to the same degree that IRAS F10214+4724 does (Elston et al. 1994, Eisenhardt et al. 1995) we will have to await ISO or SIRTF before we can tell.

There is yet another argument as to why Lya tell us little about the presence or lack of dust. Fosbury and co-workers have shown that when one considers dust/gas clouds that are illuminated by an external source of UV photons, be they line or continuum photons, the dust will efficiently reflect the incident light rather than absorb it. The alignment of the UV line emitting regions with the radio source axis in the vast majority of distant radio galaxies, and their high degree of ionization, argue strongly for a central source of ionization. If this is the case then the externally illuminated cloud geometry is the correct one to consider. The high effective albedo of the dust to Lya photons can lead to enormously strong observed Lya even in the presence of normal gas to dust ratios (e.g. Cimatti et al. 1993). The colors are too blue for there to be significant reddening: Below rest-frame wavelengths of about", 2500A the typical radio galaxy

HIGH RED SHIFT RADIO GALAXIES 357

has a spectrum that is roughly flat in ill units (Lilly 1989; McCarthy 1993). While this statement is true, it has unfortunately been taken to imply that all radio galaxies have flat spectra in the rest-frame UV, and therefore cannot be significantly reddened. This is certainly not the case, there is a wide variety of spectral slopes seen in the continua of radio galaxies at high z, and some are quite red. Dickinson and Dey (1996) show that 3C 324 (z = 1.21) has a spectral slope of roughly v-2 in the UV. Thus the colors alone do not rule out significant reddening in some of the galaxies.

The detection of significant linear polarization in the UV continua of several radio galaxies (e.g. di Serego Aligheri et al. 1989, Scarrott et at. 1990, Jannuzi and Elston 1991; Cimatti et al. 1993) adds a new dimension to the question of dust. As discussed above in the case of Lya, externally illuminated dust can act as a very efficient reflector of UV continuum. Rather than being reddened, the reflected spectrum may be significantly bluer than the incident spectrum, depending on the distribution of grain sizes (e.g. Tadhunter et al. 1989). Thus the shape of the rest-frame UV continuum provides little or no constraint on the lack of dust and may eventually become a useful tool for inferring the presence of grains.

The Scatter in the J( - z relation is too small to allow significant extinction: The J( band Hubble diagram is one of the great mysteries of radio galaxy research. Why galaxies selected at meter wavelengths should have a dispersion in Mv(rest) of", 30% (Lilly 1989) is unclear. The very small scatter in J( at any z less than '" 1.5 has been cited as further evidence against significant extinction. This argument is not as compelling as it seems on the surface. At z = 1 the J( Hubble diagram probes rest-frame wavelengths of '" 1 /-lm. If ones takes the position that all of the scatter in J( at z = 1 is due to extinction this implies A1ILm < 0.3 and that Av < 1, hardly a strong constraint. The J( Hubble diagram out to z = 1 simply does not reach wavelengths that are short enough to provide a sensitive probe of extinction. At higher redshifts the scatter in the observed J( - z relation becomes, in principle, a more powerful constraint. In Fig. 7 I show the current J( Hubble diagram based on galaxies from the 3CRR (see McCarthy 1993 for references) and MRC/1 Jy (McCarthy et at. 1990; McCarthy et al. in prep) surveys plus a few measurements of high redshift objects taken from the literature (e.g. 0902+34, Eisenhardt and Dickinson 1992; 4C 41.17, Chambers et al. 1990). The scatter for z > 2 is now significantly larger than it is for z < 1.5. Some of this increased spread may arise from the shorter wavelengths probed at these redshifts. The increased dispersion, however, weakens the test to the point that the J( Hubble diagram probably does not provide any more stringent of a constraint at z = 2 - 3 than it does at z = 1.

The observed R - J( colors are potentially sensitive to reddening as they

358

20

18

--Cl) ~ 16 ;j ~ ~ Cl)

0.-<I:! 14 co ~

~

12

10

.01

P. J. MCCARTHY

• : J • .......

. 1J1f . .., . V· .. ... , .. .

•• 1.· • • • . :.-. .,e r.

• • • •• • ,

.03 .1 .3 Z

1 3

Figure 7. The K Hubble diagram for the 3CRR galaxies (z < 2) and the MRC/l Jy radio galaxies (z > 1) plus a few galaxies from the literature. There is a significant increase in the observed scatter above z '" 2.

probe a large baseline of rest-frame wavelengths. The very large dispersion in color observed in radio galaxy samples (e.g. Lebofsky and Eisenhardt 1986; Lilly and Longair 1984; Dunlop et al. 1989) complicates the test, forcing one to work only with the envelope of the reddest galaxies.

Dunlop et al. (1990) have measured the R - J( colors of galaxies from the Parkes Selected Regions. They find a population of galaxies that is redder than expected from either passive evolution or no evolution models


normalized to the colors or present day gE galaxies. This effect is now seen in a number of other surveys (e.g. MRCj1Jy, McCarthy et al. in prep) and appears to be a fairly robust result. While it is possible that these excessively red colors are due to something missing in the spectral evolution models, the fact that they are seen at a very wide range of redshifts leads me to speculate that a modest amount of reddening at all redshifts is a simpler explanation. The effect is not large when comparing with a nonevolving spectral energy distribution (a few galaxies are more than 0.2 mag redder), but the effect it quite large compared to the reddest evolving models. Dunlop et al. find galaxies that are '" 1 magnitude redder than their reddest models at z '" 1 and more than 2 magnitudes too red at z = 2.5. For a fixed Av one expects that the observed R - K colors will be more effected at high redshifts as R moves down into the UV portion of the extinction curve.

4. Direct Evidence for Dust

4.1. IMAGING

Images taken from the ground have rarely shown direct morphological evidence for dust in all but the closest examples of radio galaxies (e.g. Cen A). The Hubble Space Telescope has shown itself to be a powerful tool for detecting dust features in galaxies out to fairly interesting redshifts. Baum et al. (these proceedings) show evidence for dust lanes in a large fraction of 3CR galaxies with redshifts '" 0.1. De Koff et al. (1996) have imaged a nearly complete subset of the 3CR galaxies with 0.1 < z < 0.5 with HST in the snap-shot mode. They find clear signatures of dust lanes in roughly 45% of these galaxies, showing that dust is quite prevalent even to modest redshifts. The deep HST images of galaxies at z > 1 are more difficult to interpret by themselves since they rarely have smooth enough profiles to clearly silhouetted dust lanes. By comparing the deep HST images of 3C324 (z = 1.206) with ground based images at K, Dickinson et al. (1996) find that the central regions of the galaxy are undetected in the HST images which sample rest-wavelengths of rv 3000A. The color of the central one arcsecond of 3C 324 is redder than that of an unevolved present day giant elliptical redshifted to z = 1.2. This suggests that there is significant extinction in the central few kpc in this galaxy. Shallow HST images of other galaxies at z '" 1 also show little or no flux at the position of the nucleus (3C 124, 3C 266 McCarthy et at. 1996), suggesting that this result is not unique to 3C 324.

360 P. J. MCCARTHY

30

3C 22 Ha

[NII]6548 [NII]6584

20

10

o 1.24 1.26 1.28 1.3

A(J.-l)

Figure 8. A J band spectrum of 3C 22 (z = 0.937). The HQ' and [N II] emission lines are marked.

4.2. SPECTROSCOPY

The most sensitive measurements of reddening in most astrophysical systems come from spectroscopy. Comparing the strengths of lines arising from the same ions and involving a common ground state with calculated line ratios provides a sensitive measure of the reddening. The recent gains in the sensitivity of near IR spectrometers opens up the possibility of making such measurements for radio galaxies and other high redshift objects. The


Lya to Ha ratio have been measured in a number of radio galaxies with z> 2 (McCarthy et ale 1992; Eales et al. 1993). In Fig. 8 I show a K-band spectrum of 3CR 22 obtained with the Cryogenic Infrared Spectrometer on the KPNO 4 m in collaboration with Richard Elston and Peter Eisenhardt.

There is a wide range in the values of Lya/Ha for the objects observed to date. Some are very close to the case B value of 8.25 (Binette et al. 1991) others are closer to unity. Taken at face value these imply a significant amount of reddening in some objects. McCarthy et ale (1992) derive Av '" 0.3 for two radio galaxies at z = 2.43, while the smaller Lya/Ha values measured by Eales et al. (1993) for other objects imply larger values for Av. The interpretation of the observed Lya/Ha in terms of reddening by a simple screen of dust is almost certainly hopelessly naive. The scattering processes discussed above and the inevitably complex geometry of the line emitting regions makes a more realistic interpretation of the Lya/Ha ratios difficult if not impossible. A safer, but more difficult, approach would be to measure the Ha/H,B ratio at z = 2.5 in the Hand K windows. This has been done for IRAS F10214+4724 by Elston et al. (1993) who find Ha/H,B > 20 consistent with the large amount of reddening expected in this object. With the large telescopes that are now, or will soon become, available, this kind of measurement should be possible for a large number of radio galaxies.

5. Conclusions

For years we have tried to avoid dealing with the issue of dust in distant radio galaxies. Until recently there was no real motivation for doing so. The data obtained over the past few years and that being acquired now may force to accept dust as an important component of high redshift objects. While this will undoubtedly make life more complicated, we can take heart from the prospect that in the near future ISO and in the not so near future SIRTF may provide us with unambiguous data regarding the amount of dust present in high redshift radio galaxies, its spatial disposition, and its effect on the observed properties of the distant universe.

References

Baum, S. A., Heckman, T., Bridle, A., van Breugel, W., and Miley, G. 1988, Ap. J. Suppl., 68, 643

Baum, S. A., Heckman, T., and van Breugel, W. 1992, Ap. J., 389, 208 Binette, 1., Magris, G., and Bruzual, G. 1991, in Relationship Between Starbursts and

AGNs. ed. A. V. Filippenko, (ASP, san Francisco) Chambers, K., Miley, G., and van Breugel, W., 1990, Ap. J., 363, 21 Cimatti, A., di Serego Alighieri, S., Fosbury, R., Salvati, M. and Taylor, D. 1993, MNRAS,

264,421 Dey, A., Spinrad, H., and Dickinson M., 1995, Ap. J., 440, 515.

362 P. J. MCCARTHY

Dickinson, M., Spinrad, H., and McCarthy, P. unpublished Dickinson, M. and Dey, A. 1996 in prep. di Serego Alighieri, S., Fosbury, R., Quinn, P., Tadhunter, C. 1989, Nature, 341, 307 de Koff, S., et al. 1996, Ap. J., in press Dunlop, J., Guiderdoni, B., Rocca-Volmerange, B., Peacock, J., and Longair, M. 1989,

MNRAS, 240, 257 Eales, S. A., and Rawlings, S. 1993, Ap. J., 411, 67 Eisenhardt, P. and Dickinson, M., 1992 Ap. J., 399, L47 Eisenhardt, P. et al. 1995 Ap. J., 461, 72 Elston, R. et al. 1994 A. J., 107, 910. Heckman, T., Lehnert, M., Miley, G., van Breugel, W. 1991, AP. J., 370, 78 Hill, G. and Lilly, S. J., 1991, Ap. J., 367, 1 Jannuzi, B., and Elston, R. 1991, Ap. J., 366, L69 Lebofski, M. and Eisenhardt, P. 1986, Ap. J., 300, 151 Lilly, S. J., and Longair, M. 1984, MNRAS, 211, 833 Lilly, S. J. 1988, Ap. J., 333, 161 Lilly, S. J. 1989, Ap. J., 340, 77 McCarthy, P. 1993, Annual Reviews of Astronomy and Astrophysics, 31, 639 McCarthy, P. J., Baum, S. A., and Spinrad, H. 1996, Ap. J. Suppl., in press. McCarthy, P., Elston, R., and Eisenhardt, P. 1992, Ap. J., 387, L29 McCarthy, P., Miley, G. et al. 1996, in prep. McCarthy, P., Spinrad, H., van Breugel, W., Liebert, J., Dickinson, M., Djorgovski, S.,

and Eisenhardt, P. 1990, Ap. J., 365, 487 McCarthy, P. J., Spinrad, H., and van Breugel, W. J. M., 1995 Ap. J., 99,27 McCarthy, P. J., van Breugel, W. J. M., Spinrad, H., and Djorgovski, S. 1987, Ap. J.,

321, L29 Robinson, A., Binette, 1., Fosbury, R., and Tadhunter, C. 1987, MNRAS, 227, 97 ROttgering, H., van Ojik, R. and Miley, G. K. (1996), in press. Rowan-Robinson et al. 1991, Nature, 351, 719 Scarrott, S. M., Rolph, C., and Tadhunter, C. 1990, MNRAS, 240, 5p Spinrad, H., and Djorgovski, S. 1984, Ap. J., 285, L49 Tadhunter, C. N., Fosbury, R. Binette, 1., Danziger, I., and Robinson, A. 1987, Nature,

325, 504 Tadhunter, C. N., Fosbury, R. and Quinn, P. 1989, MNRAS, 240, 225 van Ojik, R., Miley, G. K., Rottgering, H. 1996 A. and A., in press van Ojik, R., Rottgering, H., Miley, G., Bremer, M., Macchetto, F., Chambers, K. 1994

A. and A., 289, 54.

KINEMATICS AND H I ABSORPTION IN LYMAN ALPHA HALOS AROUND Z>2 RADIO GALAXIES

HUUB ROTTGERING

Leiden Observatory

Abstract. We discuss recent spectroscopic observations of LyQ halos around z > 2 radio galaxies. A large rotating LyQ disc of size 135 kpc is found around the radio galaxy 1243+036 (z = 3.6) and could well be associated with the accretion of gas during the formation of the galaxy. In a sample of 18 LyQ halos we find that (1) large extended regions (~ 20 kpc) of high column density neutral gas are widespread and (2) there is a strong link between the properties of the radio sources and those of the LyQ halos. From this we argue that the small radio sources reside in relatively dense regions in the early universe.

1. Introduction

The LyQ emission of distant (z > 2) radio galaxies is often spectacular (e.g. McCarthy 1993). It can be as luminous as 1044 erg S-1 and extent up to 100 kpc (f'V 10" at z = 2.5). These halos could well trace the reservoir of gas from which the galaxies are forming.

During the last decade the number of known radio galaxies with measured redshifts well over 2 has grown dramatically. During a successful ESO Key Programme we have found more than 30 of the f'V 60 galaxies known at z > 2 (e.g. Rottgering et ai. 1995a; van Ojik 1995) We have used this sample of distant galaxies to study the properties of the LyQ halos in detail. This study showed a wealth of structures associated with the LyQ halos, including large extended regions (> 20 kpc) of high column density neutral gas, filamentary structured LyQ halos with sizes up to 100 kpc, rotation over the scale of the halos, and gas with large velocity dispersion at the location of the radio jet.

363

M. N. Bremer et al. (eds.J. Cold Gas at High RedshiJt. 363-366. © 1996 Kluwer Academic Publishers.

HUUB ROTTGERING

Here we will briefly present results from this study. ''''e will first discuss the objects 0943-242 (z = 2.9; Rottgering et 01. 199.5), since this object is a good example of absorption by extended regions of neutral gas and L24:{+036 (z = :Uj; van Ojik et al. 1995), since this is a good example of a complex velocity field within a Lya halo. Secondly, we will report 011

spectroscopic observations of a sample of IS Lya halos. We will discuss the results from these observations in term of differences in the environment of the distant radio sources.

2. 0943-242 (z = 2.9)

Although the study of quasar absorption lines has been an iJnportant technique to study conditions in the early universe, the most fnnclamentallimitation of such studies is that quasars are unresolved and that therefore, in general, no information can be obtained about the spatia.! sca.!e of the absorbers.

High-resolution spectra (1.5 A) of the Lyn region of the z = 2.9 radio ga.la.xy 094:{-242 reveal a complex emission line profile which is dominated by a black absorption trough centred 250 km 5- 1 hlueward of the emission peak (see Rottgering et a1. 1995 and Rottgering 1994). We interpret this trough as H I absorption with a column density of 1 x 1019 CIIl-'2. This absorption covers the entire Lya emission region which has a spatial scale of 1.7/1. The linear size of the absorber is thus at least 13 kpc, making this the first direct nwasurement of the spatial scale of all absorber with a column density of rv 1019 cm- 2 .

3. 1243+036 (z = :~.6)

Deep narrow ba.nd ima.ging and high resolution spectroscopy of the radio galaxy 124:~+0:{6 (z = 3.6) show an extended Lyo halo with complex kinematics. In Fig. la shows the Lyo distribution, observed with a resolution of 0.6/1 (£SO NTT), together with a 0.23/1 resolution VLA ma.p at I'U GHz. The ionized gas halo extends for rv 100 kpc and is highly clumped. The inner region is dominated by bright centra.! emission which has the shape of a cone. Such a cone-shape can be explained if this region of the Lyo gas is photoionised by a beam of photons from an obscured nucleus (e.g. Antonucci 1993). The radio jet shows a strong bend at the location of the bright radio knot at 12h 45 fl1 :{S.4:P; 03° 23' 19 . .5" (.J2000). The location of this bend coincides with a region of enhanced Lyo emission suggesting a direct interaction between the radio jet and the emission line gas.

In Figure Ib, we show a two-dimensional representation of the 2.8 A resolution spectrum of LyO' taken through a slit oriented along the main axis of the radio emission. This spectrum, together with the image of the

LYa HALOS AROUND Z > 2 RADIO GALAXIES 365

arc sees

032325 CV~ @ Q ;;

0

24 " 0

z 0° a ~ i= 23 c( 0 z ::J

22 Q, 0 w

~ Q

21 II > ~ = <r-., 20 0 =

:: '" ~&

19

18 {£

~

17 0 0

0

38.4 38.3 RIGHT ASCENSION

Figure 1. (a) A contour plot of the Lya halo of the radIO galaxy 1243+0.'16 at z = 3.6, with a greyscale plot of the 8.3-GHz VLA map supenmposed. (b) a two-dimensIOnal representatIOn of the 2.8 A resolutIOn spectrum of Lya taken through a slit onented along the mam aXIs of the radIO emiSSion

LyO' emission, shows that the LyO' gas has three distinct components: (i) gas with a high velocity dispersion (1550 km s-1 FWHM) located inside the radio structure, (ii) enhanced LyO' emission blue shifted by 1100 km s-l at the location of the strong bend in the radio jet and (iii) Lyo emission extending out well beyond the radio lobes. This emission has a low velocity dispersion (250 km s-l FWHM) and a velocity gradient of 450 km s-1 over the extent of the emission, indicative of large scale rotation.

Various models have been proposed to explain the high velocity dispersion in the extended emission regions around distant radio galaxies. Since the high velocity dispersion gas is located in a region confined by the radio source and the low velocity dispersion gas is located outside the radio source structure, the high velocity dispersion is naturally explained as the result of hydrodynamical interactions of the radio jet with the gas. Such an interaction is directly seen at the bend of the radio jet, where the gas is enhanced and accelerated by the jet-cloud encounter.

The observed velocity gradient of the outer halo of 1243+036 might well be due to a large scale rotation of a gas disk. A gravitational origin of the rotation of such a large disk implies a mass of", 1012 sin-2 U) MG), where 'i is the inclination angle of the disk with respect to the plane of the sky. We advocate a scenario for the formation of this object in which the outer halo is associated with the accretion of gas during the formation of the galaxy.

366 HUUB ROTTGERING

Numerical simulations by Evrard et al. 1994 of the formation of galaxies in hierarchical clustering scenarios indicate that rotating disks with radii of several tens of kiloparsecs should be common around forming galaxies at high redshift.

4. Spectroscopy of Lyo: halos

We have analysed the high resolution spectroscopy from a sample of 18 distant radio galaxies (van Ojik, 1995; van Ojik et al. 1995b). An immediate result is that H I absorption features are widespread in the Lyo: profiles. 11 radio galaxies of the sample of 18 have strong (> 1018 cm -2) H I absorption. Since in most cases the Lyo: emission is absorbed over the entire spatial extent (up to 50 kpc), the absorbers must have a covering fraction close to unity. Given the column densities and spatial scales of the absorbing clouds, the typical H I mass of these clouds is '" 108 M0 . We find clear indications that the properties of the Lyo: halos are strongly linked to the size of the radio source. The smaller radio sources generally show strong absorption, have relatively high velocity dispersions in the Lyo: gas and are associated with the small and relatively distorted Lyo: halos.

The observed differences among the radio galaxies can be understood if the smallest radio sources are in the densest environments. In such an environment the large amount of H I gas will absorb part of the Lya. Due to the interaction of the radio jet with this dense medium, the velocity dispersons in the the Lya is high and the radio sources are relatively small and distorted.

Acknowledgements. I would like to thank my collaborators, Malcolm Bremer, Chris Carilli, Dick Hunstead, George Miley and Rob van Ojik.

References

Antonucci R., 1993, ARA&A, 31, 473 Evrard A. E., Summers F. J., Davis M., 1994, ApJ, 422, 11 McCarthy P. J., 1993, ARA&A, 31, 639 Rottgering H., HUllstead R., Miley G. K., van Ojik R., Wleringa M. H., 1995a, MNRAS,

277, 389 Riitt.gering H., van Ojik R., Miley G., Chambers K., 1995b, Spectroscopy of Ultra-Steep

Spectrum Radio Sources: A sample of z > 2 Radio Galaxies, submit.t.ed Rot.t.gering H. J. A., 1995, in Hippelein H., Meisenheimer K., eds, Galaxies in t.he Young

Universe. Springer-Verlag, in press van Ojik R., 1995, Ph.D. thesis, University of Leiden van Ojik R., Riittgering H., Carilli C., Miley G., Bremer M., 1995a, A radio galaxy at.

z = 3.6 in a giant. rotating Lyman 0' halo, A&A: in press van Ojik R., Riitt.gering H. J. A., Miley G. K., Hunstead R., 1995b, TIle Gaseous

Environment of Radio Galaxies in the Early Universe: Kinemat.ics of t.he Lyman 0'

Emission and Spatially Resolved HI Absorption, A&A: submit.t.ed

THE LYMAN ALPHA VELOCITY FIELD OF THE RADIOGALAXY 4C 41.17 (Z=3.8) WITH TIGER/CFHT

B. ROCCA-VOLMERANGE Institut d'Astrophysique de Paris 98bis Bd Arago, F-75014 PARIS

Abstract. The velocity field of the ionised hydrogen in the radiogalaxy 4C 41.17 at z = 3.8 is mapped in the Lyo: 1215A emission line. The threedimensional spectroscopy was carried out with the integral field spectrograph TIGER at the 3.60m CFHT in a short exposure time (~ 2.0h). The narrow intense peak emission is confirmed. As new results, the values of the radial velocity relative to systemic and projected along the line of sight (Okms-1 ~ Vr ~ -115kms-1 ) are mostly negative. The detection of low surface brightness ionised clouds surrounding the main body of the galaxy confirm the anisotropy of the ambient medium. Individual spectra show signatures of star formation. A comparison with previous observations of velocities in 4C 41.17 (Chambers et al., 1990, Hippelein and Meisenheimer, 1993) is presented. The radiogalaxy 3C 435A, observed with TIGER at z = 0.471 (Rocca-Volmerange et al., 1994) and the present observations of 4C 41.17 are favoring an expansion model. The performance of TIGER for such faint objects is confirmed, with a sensivity comparable to long slit spectrographs with spectra better identified.

1. Introduction

The possibility of mapping the gas kinematics of a radiogalaxy and its environment at a red shift z = 3.8 is exceptional because the look-back time is about 80-90% of the age of the universe. At such large distances, the velocity field of the ionised gas gives constraints on the physical processes of galaxy formation as well as on the temperature and the density of the intergalactic medium. By chance, the Lya 1215A emission line, red shifted into the visible, is so intense for distant radiogalaxies that the detection of their optical counterparts is possible (see the first sample from Djorgov-

367


368 B. ROCCA-VOLMERANGE

ski et al., 1984). Many distant Lya galaxies were subsequently discovered, in particular 4C 41.17 at z = 3.8 (Chambers et al., 1990) and the most distant ones at z = 4.25 (Spinrad et ai., 1995) and at z = 4.4, recently discovered at ESO (D'Odorico et al., 1996, in preparation). In these galaxies, the ionised gas has been considered for several years to be a significant indicator of star formation, supposed to be triggered by the interaction of the radiojet with the intergalactic medium or by the over-pressurized cocoon of the radiosources (Begelman and Cioffi 1989, Rees 1989, de Young 1989) or by the interaction of the relativistic electrons with photons of the cosmic background (Daly, 1992). These models tentatively explained the observed alignment of the radio and ultraviolet axes (McCarthy et ai., 1987, Chambers et ai., 1987). The star formation process is evidently efficient, and confirmed by the HST observations of the continuum (Miley et ai. 1992, van Breugel 1995) and possibly by th~ presence of dust identified by the submillimetre emission (Dunlop et al., 1994). But such evidence is not a proof that photoionisation from massive stars are the unique sources of the Lya emission. Other processes (shocks, non-thermal component) could also be the origin of the ionisation of gas. Only three-dimension spectrophotometry allows a determination of the respective distributions of stars and gas and so to identify the possible relation between these, as well as the nature and intensity of each process. The velocity field from the Lya 1215A emission line and the corresponding line widths were derived for the radiogalaxy 4C41.17 with the integral field spectrograph TIGER at the CFHT 3.60m telescope (Adam et al., 1995, Rocca-Volmerange et al., 1995, in preparation). These results are compared to the previous estimates of velocities of 4C 41.17 from long slit (Chambers et ai., 1990) and Fabry-Perot (Hippelein and Meisenheimer, 1993) measurements. Finally, similarities with the radiogalaxy 3C435A (z = 0.471), also observed with TIGER (RoccaVolmerange et ai., 1994), favor expansion models of radiogalaxies and give possible signatures of evolution of the adopted model.

2. The observations

The integral field spectrograph TIGER (Courtes et al., 1987, Bacon et al., 1995) is installed at the 3.60 m CFHT. In spectroscopic mode, a 400 microlens array gives a set of micropupils dispersed by a grism, and a series of spectra, well calibrated on standards, identifying the total field of view of 11/1 X 11/1. The software package of spectrum extraction was developed in the MIDAS environment by Rousset, 1992. The wavelength calibration is checked on the 01 night sky line, yielding a ±1 A measured accuracy. The TIGER observations were carried out in November 1994 with a spatial sampling 0.61/1, the size of a microlens on the sky. The wavelength range is

LYMAN ALPHA VELOCITY FIELD OF 4C 41.17 369

Figure 1. The Lyll' 121SA image of the radiogalaxy 4C41.17 (z = 3.8) and its environment. Scale is in arcsec. The image has been reconstructed from the LYll'121SA emission lines of the individual spectra observed with TIGER.

5000-7000 A. and the CCD is a Loral3 with pixels of 15 J-lm, improved in the blue with an effective resolution 17.3A.. The seeing was excellent, ~ 0.5", during the relatively short two exposures of 3600 sec + 3130 sec. The data processing was carried out by S. Gerard and G. Adam at the Observatoire de Lyon with the recently updated version of the TIGER software.

A so-called image of the field is rebuilt from individual spectra, showing the main characteristic features (Fig. 1). The galaxy (about 6" X 2.5") is identified with two main components separated by a dark zone. One component corresponds to the brightest isophotes surrounding an intense narrow peak( 4.8 X 1043 erg s-1). The luminosity, integrated inside the galaxy isovelocity contours without the central peak, is 3.8 X 1044 erg S-1, in agreement with the results from Chambers et at., 1990 even if these last authors found a galaxy size of 10" X 15".


Figure 2. Map of the radial velocity field relative to systemic value (white peak at z = 3.8) along the line of sight. The map is superimposed on the LyO' 121SA image of Fig. 1. Velocities are in km S-1 .

The surface brightness 6.5 X 1043 erg s-1 arcsec-2 in the peak decreases by a factor 3 in the rest of the galaxy. A slight curvature of these isophotes is visible towards the North-East, likely following a curvature of the radiojet. The other component visible on the other side of the dark zone, apparently aligned with the radio axis (Carilli et ai., 1994) is either intrinsically fainter or partly absorbed by a foreground cloud. The indidual spectra have been summed, and the integrated spectrum shows a depression of the Lyman continuum, attributed to the Lya forest, as for quasars. The slope of the continuum (1250A-1450A) is typical of massive stars. The external boundaries of the galaxy are limitated by a thick envelop of diffuse ionised gas, surrounding the main part of the galaxy. Many gaseous extensions are visible on the image, the most important of which is in the south and low surface brightness clouds are dearly identified implying a typically inhomogeneous medium, which confirms the dumpiness of the gaseous halo (Chambers et at., 1990).

LYMAN ALPHA VELOCITY FIELD OF 4C41.17 371

3. The velocity field of the radiogalaxy 4C 41.17

The radial velocity was determined for each spectrum of sufficient signal to noise ratio with an accuracy of ± 1A. We calculate the radial velocity relative to systemic (fixed on the bright peak, at z = 3.800)) along the line of sight. A map of these relative values is presented in Fig. 2. Surprisingly, most values are negative inside the error bars with Okms-1 ~ Vr ~ -113kms-t, relative to the peak. More regular isophotes between -30 and -50 km S-1 appear between the two components. Roughly these aligned isophotes correspond to the sharply cut off Lya emission profile observed in other radiogalaxies (van Ojik et al., 1995). Their interpretation by an optically thick cloud, a disklike or other morphology of the radiogalaxy is not yet confirmed and needs more extensive data. In contrast with these surprisingly low velocities, the line widths are large. They would correspond either to high velocity dispersions (up to :=1700 km s-1) in the rest frame of the radiogalaxy or to the two expanding sides of an optically thin ionised lobe. Processing of the individual spectra will allow to distinguish the two possibilities. The velocities of the gas encompass a range of about 120 km s-1, much lower than the estimates of 2000 km S-1 published by Chambers et ai., 1990 and of 500 km S-1 from Hippelein and Meisenheimer (1993). Differences could come from the instruments since the spatial resolution of a long slit is more uncertain than of the microlens array and the separation of orders in the Fabry-Perot is a cause of uncertainties. However in these two papers the separate components (lobes in expansion, the galaxy itself, a peak and a disk) were already identified. Another result concerns the continuum of each spectrum, the slope of which locally identifies the star formation process with typical massive stars. A comparison with the HST data will be most interesting.

4. Comparison with the radiogalaxy 3C 435A (z=O.471) and preliminary interpretation

The galaxy 3C 435A (R = 19), a projected pair with 3C 435B (McCarthy et al., 1990) was observed with Tiger in July 1992 with an average seeing 0.7" in the blue(5000-7000A) and in the red (6500-8500A), allowing the identification of the [011](3727 A) and [0111](5007 A) lines. The astrometric accuracy was ±0.4". The spatial sampling in the adopted spectroscopy mode is 0.61" with a field 11" X 11". The spectral sampling 8A induces an effective spectral resolution of 20.8 A. Exposure times were 2 X 3600 sec in the blue and (2 X 3600 + 2463 + 1808) sec in the red. From morphology and nebular emission maps (Rocca-Volmerange et ai., 1994), the main result is the identification of the stellar and nebular components. The ionised gas, traced by the [0 II] and [0 III] isophotes follow the curvature of the ra-


dio (1.4GHz) isophotes. The nebular components identify the plasma lobes. Stellar contin~a of presumed galaxy components are significantly des placed from the nebular zones. One component is a typical elliptical galaxy with V = 22.3 fitted with our spectral atlas (Rocca-Volmerange and Guiderdoni, 1988) at an age of 8 Gyrs. The central component, also elliptical with R = 21.5, has an age of 11 Gyrs and a blue fainter component has an age of ~ 5 Gyrs, all at z ~= 0.37. These large ages for the radiogalaxy and one companion induce a hierarchical system which are constraining for models of galaxy formation in groups or clusters. Compared to a number of models, the observations of the two radiogalaxies 4C 41.17 (z = 3.8) and 3C435A (z = 0.471) fit the model of the Lya emission with lobes of overpressured gas expanding towards the observer (Begelman and Cioffi, 1989, Nath, 1995). A more precise comparison needs to analyse the multispectral radio counterparts with available astrometry and numerical fits with hydrodynamic simulations, which are in progress.

References

Adam, G.,Gerard, S., Rocca-Volmerange, B., Bacon, R., to be subm'jtted Bacon, R., Adam, G., Baranne, A., Courtes, G., Dubet, D., Dubois, J.P., Emsellem,

E., Ferruit, P., Georgelin, Y., Monnet, G.,pecontal, E., Rousset, A.,Sayede,F, 1995, preprint

Begelman, M.C., Cioffi, D.F., 1989, Astrophys. J., 345, L21 Carilli, C. L., Owen, F.N., Harris, D.E., 1994, Astron. J .. , 107,480 Chambers, K., Miley, G., van Breugel, W., 1987, Nature, 329, 604 Chambers, K., Miley, G., van Breugel, W., 1990, Astrophys. J., 363, 21 Courtes, G., G eorgelin, Y.,Bacon, R., Monnet, G., Boulesteix, J., Santa Cruz Summer

Workshop, July 1987 Daly, R., 1992,Astrophys. J., 386, L9 Djorgovski, G., Spinrad, H., Marr, J., 1984, in New aspects of Galaxy Photometry, ed.

J.L. Nieto Dunlop, J., Hughes,D., Rawlings, S., Eales, S., Ward, M. 1994, Nature, 370, 347 Hippelein H., Meisenheimer,K., 1993, Nature, 362, 224 McCarthy P.J., van Breugel W., Spinrad H., Djorgovski, S., 1987, Astrophys.J., 321 ,

L29 Miley, G.,Chambers, K., van Breugel, W., Macchetto, F., 1992, Astrophys. J., 401, L69 Nath, B., 1995, Mon. Not. R. astro. Soc., 274, 208 Rees, M., 1989, Mon. Not. R. astro. Soc, 259, 265 Rocca-Volmerange, B., Adam, G., Ferruit, P., Bacon, R., 1994, Astron. Astrophys., 292,

20 Rocca-Volmerange, B., Guiderdoni, B., 1988, Astron. Astrophys.Sup.Series, 75,93 Rousset, A., 1992, These d'Universite J. Monnet, St Etienne van Breugel et al., 1995, in

Proceedings of the extragalactic radiosources, Bologna, in press van Ojik, R., Rottgering H.J.A., Carilli,C.L., Miley, G., Bremer, M., Macchetto, F., 1995,

Astron. Astrophys., preprint

THE EFFECT OF A SURROUNDING COOLING FLOW ON A POWERFUL RADIO SOURCE

M.N. BREMER

Sterrewacht Leiden, Postbus 951,"], 2.JOORA Leiden, NL.

AND

A.C.FABIAN AND C.S. CRAWFORD

Institute of Astronomy, Madingley Road, Cambridge CB,"] OHA, UK.

Abstract. We review the evidence for powerful, distant radio galaxies and radio-loud quasars being found at the centres of clusters of galaxies undergoing strong cooling flows. We describe the situation in the cluster and host galaxy of a radio source before and after the initiation of emission from the central engine, based on an extension of the situation found in strong cooling flows at low redshift. We show that many of the commonly observed features of distant radio sources can be explained by the interaction between the radio source and the surrounding multi-phase intracluster medium in this scenario. Finally we note that this scenario is compatible with interaction-based scenarios for triggering radio sources.

1. Introduction

There is now considerable multi-waveband evidence that powerful (FRII), high redshift, radio galaxies and quasars are found at the centres of clusters of galaxies which may be undergoing strong cooling flows.

Optical number counts of faint galaxies around radio-loud quasars and radio galaxies show an excess of objects relative to the field for objects at z> 0.5 (e.g. Vee & Green 1987, Hill & Lilly 1991). Recent work based on IR data (e.g. Dickinson & Eisenhardt 1994) and ongoing HST work show that this continues to z > 1. The optical and IR luminosities of radio-loud quasar host galaxies (Romanishin & Hintzen 1989) and of powerful radio

373


374 M.N. BREMER ET AL.

galaxies (Lilly & Longair 1984) are comparable or more than those of low red shift giant elliptical and cD galaxies, found at the centres of clusters.

The two most luminous 3C sources at low redshift, 3C295 and 3C405 are in clusters that are clearly detected in X-rays. Although at comparatively low redshift, these sources have similar radio luminosities to 3C sources at z = 1, suggesting that the highest power sources at any redshift are in clusters of galaxies. ROSAT has detected several more distant galaxies consistent with emission from hot gas (Intracluster Medium) surrounding the radio galaxies. The soft X-ray spectra of distant radio-loud quasars often show absorption by high column densities of cool and cold gas (see the contribution by Elvis). Similar absorption is often seen in the X-ray spectra of low red shift clusters with cooling flows (see the contribution by Johnstone), where it is thought to be caused by gas that has cooled out of the hot phase. Such absorption is not seen in the spectra of low redshift, lower power radio-loud quasars.

The pressures of the surrounding environment of powerful extended radio sources can be estimated from both the minimum pressure of the extended radio structure and from conditions in extended emission-line gas. The pressures so determined are as high or higher than those found in low redshift clusters with the strongest cooling flows (Bremer et at. 1992 and references therein). The scale size ofthe radio emission (and therefore of the radio working surface) and the emission-line gas (100 kpc) along with the ubiquity (and therefore longevity) of extended emission-line regions implies that the surrounding (hot) medium has a 100 per cent filling factor and a high pressure over scales of > 100 kpc. Again this implies that the sources are surrounded by a hot, dense ICM. The high pressure of the ICM means that a strong cooling flow must be occurring (by analogy with low redshift clusters). The high rotation measures seen in the extended radio emission of distant radio sources (e.g. Garrington et at. 1988, Carilli, Owen & Harris 1994) are easily explained by the presence of a surrounding dense ICM. At low red shifts such high rotation measures are only seen in sources at the centres of cooling flow clusters (e.g. Ge & Owen 1993).

2. The initiation of a radio source in a cooling flow

Given the above evidence, we now discuss a scenario in which a powerful radio source starts up in a galaxy at the centre of a strong cooling flow with conditions like those implied by the previously discussed observations.

A (forming) giant elliptical galaxy lies at the centre of a cluster of galaxies. Within the galaxy, there is a central massive black hole, currently inactive. Surrounding the galaxy is a hot intracluster medium at high pressure (> 107 cm-3 K within 10 kpc of the central galaxy). Consequently a strong

POWERFUL RADIO SOURCES 375

(perhaps 500-1000 M0 yr- I ), well established, cooling flow has been occurring. Within the cooling region, cold, largely molecular (and possibly dusty, e.g. Allen 1995) clouds are embedded in the hot medium. The clouds are distributed in a centrally condensed manner, the mean density of cooled gas follows a distribution as least as steep as r- 2 , where r is the distance from the centre of the cluster. Close to and perhaps within the galaxy (at a distance of about 1 kpc), the covering fraction of clouds can be above unity, the clouds having a much higher filling factor than further away from the galaxy. Perhaps 109 M0 of cool and cold clouds can lie within 1 kpc of the centre of the galaxy, depending on how long the cooling flow has been established and how efficient star-formation is in these clouds. The cold clouds control the dynamics of the central region, as in this region more mass resides in cold clouds than in the hot phase. The velocity spread of these clouds reflects the gravitational potential of the cluster. Typically the full-width half-maximum of the distribution of velocities of these clouds is several hundred kms- I . Any turbulence may cause line-emission from these clouds and those out to 10 kpc from the centre of the cluster. Up to this point, the central galaxy and cluster appears as an extreme example of a low redshift cluster with a cooling flow. This situation is portrayed in Fig. 1.

We now assume the radio source starts up (caused for example, by an interaction between the host galaxy and another galaxy). We assume the nucleus starts to emit ionizing radiation at the same time as it starts to emit in the radio. In the X-ray, the source will appear similar to Cygnus A (Carilli, Perley & Harris 1994). Figure 2 characterises what happens to the surrounding medium once the source starts up. The material outside the radio plasma and ionizing beam appears little different from the earlier situation, as they experience little increased ionizing photon flux or turbulence induced by the expanding radio plasma.

The clouds within the ionizing beam and along the radio axis change considerably. Providing the radio plasma escapes the dense clumpy inner region of the ICM within the galaxy, ionizing radiation escapes along the path cleared by the radio jet. The cold clouds in the beam see upwards of 1000 times the photo-ionizing flux than from X-ray emitting hot phase alone. They develop ionized skins, emitting UV /optical/IR emission lines, increasing in total volume as they do so (as their pressure remains approximately constant). The higher column density douds retain neutral and molecular cores. Thus the covering fraction of ionized material increases within the beam, though the covering fraction of neutral and molecular material may not decrease by very much (this depends entirely on the geometry of the clouds).


e VI G

., G , , 0 ,.

CD C G CD ., (l) 0

e 0 CD 0 CD • e • °0 0 0 e • CD

0 Qe • G

Q

0 f& (I) e 0 °e Q

(J ~ 0°

CD 0 0 0 0 ., f!) G. (') e

0 Cl) e e

0 e CD

Q) 0

Figure 1. Diagram ofthe situation in an extreme cooling flow before an AGN has started up. The central cluster galaxy contains a quiescent massive black hole. Surrounding the galaxy is the hot phase of the 10M, embedded in which are cold, mainly molecular clouds. The turbulence at the center of the cluster within 10-15 kpc is sufficient for some clouds in this region to have an ionized skin due to cloud-cloud collisions and the action of the ambient X-ray emitting medium on the turbulent mixing layers at the surface of the clouds. What is important to note about this scenario is that the cold phase is spheroidally distributed around the radio source.

3. Implications and predictions of this scenarIO

We now show that the above scenario can naturally explain many of the observed properties of distant, powerful radio sources. We also make predictions of properties yet to be measured. For all of this we assume orientation dependent unification (e.g. see Baker, these proceedings).

Cold clouds exposed to the ionizing radiation of the central source develop ionized skins and are observed as the extended emission-line regions seen around radio galaxies and quasars. The emission will be aligned with the radio axis assuming the radio and optical axes are the same. The same clouds can also provide extended optical and UV continuum emission, either by the mechanism discussed by Dickson et ai. (1995), or by scattering of nuclear continuum if the clouds contain dust.

Because the highest column density clouds remain optically thick even when photo-ionized by the nucleus, they are a reservoir of dust and unionized gas. As the radio jet and surrounding plasma passes through a region containing these clouds, the increased turbulence suffered by the clouds will cause some of them to break apart into smaller clouds. Locally, this will expose more gas to the ionizing beam of the nucleus, increasing the surface

•

POWERFUL RADIO SOURCES

". 0 NeutmlJMo!ecullll' Clouds

~adiO Galaxy

Ionized

377

Figure 2. Diagram of the situation in an extreme cooling flow once the central massive black hole in the central cluster galaxy has become active. For the hot phase and the cold clouds outside of the ionization cone and radio emitting region, there is little change from the situation before the AGN begins to emit. Within the ionization cone, the cold clouds develop an ionized skin swelling up to many times their original volume, photoionized by the quasar nucleus. Within the radio emitting region, the velocity distribution of the clouds is increased by interaction with the radio plasma. Those clouds strongly interacting with the radio plasma can be ripped apart increasing the amount of cool gas that can be photoionized by the quasar in that region. The hot phase is heated by interaction with the radio plasma, and is also excluded from the backflow region, as the radio plasma expands into the hot phase until it reaches pressure equilibrium.

brightness in emission-lines and possibly continuum at these points. Thus there should be clear linear and knotty structures in the extended emission that trace the path of the radio jet through the surrounding gas. These regions of increased surface brightness should trace even sharp bends in the radio jet, providing these regions remain in the ionizing beam of the source (such structures are seen in HST images of distant radio galaxies, e.g. see Best, these proceedings).

Lines-of-sight towards the nucleus, passing through the unionized cores of clouds will cause associated absorption in the spectra of both quasars and radio galaxies. As the clouds along the line-of-sight to a quasar will be photo-ionized by the nucleus, and clouds along lines-of-sight to radio galaxies see little or no extra ionizing radiation, there will be more and stronger (lower ionization) absorption towards radio galaxies (see Rottgering, these


proceedings). Moreover, optically fainter radio-loud quasars should have more associated absorption than the most luminous ones, so, for example gravitationally lensed quasars in flux limited samples should show more associated absorption (and at lower ionization states) than unlensed quasars. The associated absorption in radio galaxies will have a lower velocity velocity dispersion than the emission (emission-line clouds will be accelerated by turbulent interactions with the radio plasma).

The scenario predicts clear correlations between the size of a radio source and its other properties. For a given jet power and source age, the following should correlate with the density of the surroundings: the denser the medium, the smaller the size of the radio source, the stronger the depolarization and the steeper the radio spectrum of the extended radio emission. As increasing density leads to increasing mass deposition rate and so we expect the smaller sources to have the stronger aligned UV /optical emission in general. Similarly, line-absorption should be stronger in the smaller radio sources.

Finally we note that this scenario is not in conflict with the usual paradigm for the initiation of an AGN, that of a galaxy-galaxy interaction delivering material to the central black hole. Indeed galaxy-galaxy interactions are a natural consequence of a cluster based scenario. In this case, the interaction need only provide the gravitational trigger for the initiation, the presence of the gas could be due to the cooling flow alone. Interactions with gas poor galaxies could be equally effective in starting up a powerful radio source. Unlike the interaction scenario, the cluster scenario predicts in a straightforward manner many of the large-scale extended properties (both in the optical and radio) of powerful distant radio sources.

References

Allen S., 1995. MNRAS 276, 947. Bremer M.N., Crawford C.S., Fabian A.C. & Johnstone R.M. 1992. MNRAS, 254,614. Carilli C.L., Owen F.N. & Harris D.E., 1994. AJ, 107, 480. Carilli C.L., Perley R.A. & Harris D.E., 1994. MNRAS, 270, 173. Dickinson & Eisenhardt 1994, NOAO Newsletter, 37, 1. Dickson R., Tadhunter C., Shaw M., Clark N. & Morganti R., 1995. MNRAS 273, L29. Ge & Owen 1993, AJ, 105, 778 Garrington, S.T., Leahy, J.P., Conway, R.G. & Laing, R.A., 1988. Nat, 331, 147. Hill, G.J. & Lilly, S.J., 1991. ApJ, 367, 1. Lilly S.J. & Longair M., 1984, MNRAS, 211,833 Romanishin W., & Hintzen P., 1989, ApJ, 341, 41 Vee, H.K.C. & Green, R.F., 1987. ApJ, 319, 28.

AN HST LOOK AT DUST IN 3CR GALAXIES

S. A. BAUM, S. DE KOFF, W. SPARKS, J. BIRETTA, D.GOLOMBEK,D.MACCHETTO STScI, Baltimore MD

G. MILEY Sterrewacht, Leiden

AND

P. MCCARTHY Carnegie Institute, Pasadena

1. Introduction

We have obtained HST snapshots observations using the WFPC2 through the F702W broad band red optical filter of nearly the complete 3CR sample of radio galaxies and quasars. Here we present results on the dust in the 3CR radio galaxies in the redshift range 0.0 < z < 0.5. The images have ,...., 0.1" resolution, corresponding to (6, 55, 170,500) parsecs at redshifts of z = (0.003,0.03,0.1,0.5) respectively.

2. Dust Content and Morphology

We find dust (lanes, patches, wisps) out to a redshift of,...., 0.48. Roughly 40% of 3CR galaxies from 0.0 < z < 0.5 show obvious signs of dust. There is no sign of a change in this fraction between low (0.0 < z < 0.1) and intermediate redshifts (0.1 < z < 0.5). Examples of the dust seen in images of low redshift 3CR galaxies are shown in Fig. 1. At low z, the distribution of dust is frequently disklike (e.g., 3C83.1, 3C270, 3C326), but can be unsettled and filamentary (e.g., 3C 84, 3C 293, 3C 305). At the highest resolution (e.g., 3C 272.1) the disks appear to separate into series of parallel strands.

Examples of the dust seen in images of intermediate redshift 3CR galaxies are shown in Fig. 2. Structure in the dust is more difficult to detect at z > 0.1, but the effects of dust obscuration are seen (e.g., 3C 52, 3C 223.1,

379

M. N. Bremer et al. (eds.). Cold Gas at High RedshiJt. 379-383. © 1996 Kluwer Academic Publishers.

380 S. A. BAUM, S. DE KOFF, W. SPARKS, J. BIRETTA, ET AL.

ac 806 3C 293

-4 -4

-2 -2

0 0

2 2

4 4

4 2 0 -2 -4 4 2 0 -2 -4

3CS:U 3C449

4 .. 2 2

0 0

-2 -2

-4 -4 4 2 0 -8 -4 4 2 0 -2 -4

8C 828 3C 2711.1

., 10

2

0 0

-8

-10 -4

10 0 -10

Figure 1. Greyscales of HST snaps of selected low redshift 3C galaxies.

DUST IN 3CR GALAXIES 381

SC 52 BC 2M

2 1

0 0

,', ~<

-1

-2

-1 0 1

30327 BC 433

Figure 2. Greyscales of HST snaps of selected intermediate redshift 3C galaxies.

3C 284, 3C 327, 3C 433). It is sometimes difficult to determine what is dust obscuration and what is emission dumpiness (see below).

382 S. A. BAUM, S. DE KOFF, W. SPARKS, J. BIRETTA, ET AL.

Dust Orientations, z<0.1 shaded 10r-'-~-'-'-'--'-'-'-~-'-'-'r-'-'-~-'-'-'

8

6

4

2

20 40 60 80

IpMradio jet) - PA(dust lane)1

Figure 3. Histogram of Position Angle Differences between radio axis and dust axis, low z sources shaded.

3. Dust Orientation

We compare the dust "disk" orientation (taken to be the angle of the dust closest to the nucleus) and the radio source axis (see Fig. 3). We find a strong tendency for the dust major axis to be perpendicular to the radio source axis (26 sources with .6. > 45°), 6 with .6. < 45°. We assume that dust obscuration dominates the appearance of the intermediate redshift sources; we may be "helping" our result if we are mistaking aligned clumpy emission for dust obscuration. For the z < 0.1 sources, the result holds, although the effect is weakened somewhat (12 sources with .6. > 45°, 4 with .6. < 45°).

This confirms the earlier result (Kotanyi and Ekers, 1979), giving credence to the idea that gas from kpc scales ultimately feeds the central engine and affects its collimation axis. Note, however, the suprising dispersion in the result and the sources (e.g., 3C 305, 3C 449) whose jets issue directly into the dust disk.

4. Dust, Clumpiness, and the Alignment Effect

Roughly 15% of the sources in the redshift range 0.1 < z < 0.5 show "clumpy" emission, with two or more distinct components, rather than a smooth "elliptical" profile. In many of our "clumpy" sources it appears as

DUST IN 3CR GALAXIES 383

if the apparent clumpiness owes to the presence of a dust "lane". It is not possible for us to unambiguously determine what owes to true emission clumpiness and what owes to dust obscuration for many of these sources, particularly towards the upper end of the redshift distribution. However, in some of these sources we clearly do see dust obscuration which influences strongly the apparent morphology of the optical galaxy. This indicates that dust is an important factor in determining the morphologies we see when we look at radio galaxies at z "-' 0.5. This in turn raises the interesting possibility that the alignment effect seen in powerful/high redshift radio galaxies (McCarthy et al. 1987, Chambers et al. 1987) may be in part due to the presence of dust obscuring regions of the source. If that dust is preferentially oriented perpendicular to the radio source axis it will tend to accentuate the apparent alignment of the optical continuum in sources where the true continuum is slightly aligned along the radio axis. In addition, the dust may serve to scatter light from the nucleus.

References

Chambers, K. C., Miley, G. K., van Breugel, W. (1987), Nature 329,604. Kotanyi, C. G., Ekers, R. D. (1979), Astronomy and Astrophysics, 73, 1 McCarthy, P. J., van Breugel, W., Spinrad, H., Djorgovski, S. (1987), ApJ 321, 29.

DETAILED STUDIES OF THE LYMAN ALPHA KINEMATICS IN 2104-242

A.M.KOEKEMOER Institut d'Astrophysique de Paris, 98bis Ed. Arago, Paris 75014, France

W. J. M. VAN BREUGEL Institute for Geophysics and Planetary Physics, Lawrence Livermore National Laboratories, Livermore, CA, U.S.A.

P. J. MCCARTHY Observatories of the Carnegie Institution of Washington, Pasadena, CA, U.S.A.

AND

J. BLAND-HAWTHORN Anglo-Australian Observatory, Epping, N.S. W. Australia

1. Introduction

We present Fabry-Perot observations of the extended Ly 0: emission around the radio galaxy 2104-242. Constraints are derived on the physical properties of the gas, including the volume filling factor, total kinetic energy, and dynamical and radiative timescales. These constraints are investigated in the context of dynamical models for the gas, such as accretion from a merger event and energy input from the radio jets. Various mechanisms for energy input to the gas are also considered, in particular photoionization by the active nucleus and shocks from turbulence related to interactions with the radio plasma.

2. Lyo: observations of 2104-242

The radio source 2104-242 is optically identified with a galaxy at z = 2.491 (McCarthy et al. 1990a) and contains a very extended Ly 0: region along the radio axis ("" 12", i.e. 135 kpc for Ho = 50 km seC1 Mpcl, qo = 0). In

385


386 A. M. KOEKEMOER ET AL.

oN 11911

!

!IS

I II)

I IS

~

~ :lIl

til

31)

n 1M f)4.S !1M!

GNfmull .......... 1'''* .<m\OW 1Imt" , !Ai_: 1.~.(

t:!M, 26.00. ft. ... l00,,1l, lOIIJI, .... l!IIIJJI)

Figure 1. Ly Q! (greyscale) and 3 cm radio continuum (contours) of 2104-242. The radio core is identified with the K-band continuum core. The bright object to the N-NW is a foreground star.

Fig. 1 we present an overlay of the Lya emission (McCarthy et al. 1990a) and a 3 cm VLA image; the registration assumes a co-incidence between the radio and K-band continuum cores.

We observed the Lya emission in 2104-242 during October 1994 using the TAURUS 2 Fabry-Perot on the Anglo-Australian Telescope. Thirtythree frames were obtained at 0.88 A intervals, each frame having a spectral resolution", 1.2 A and exposure time of 300 s. Standard calibration procedures were applied and field stars were used to register the frames, which were also convolved to a uniform seeing FWHM of 1.5". The spatial Lya distribution agrees well with that presented by McCarthy et al. (1990a).

In Fig. 2 we present the kinematic data ofthe entire emission-line region, as a function of position along the radio axis. Note that the rest-frame Lya equivalent width is large, W). = 445 A (McCarthy et al. 1990a). It is evident that the velocity distribution is extremely broad, with the emission covering'" 1000 - 1500 km s-1 for both the north and the south lobes; the velocity difference between their centroids is about ±400 km s-1. Each lobe

LYMAN ALPHA KINEMATICS IN 2104-242 387

@

V t:: I:l >:Ii +' ,,:2 '"0

] "D it -5 ...

Figure 2. Ly (l' velocity distribution as a function of radial distance along the radio axis (the greyscale represents flux per unit wavelength). Note the multiple line-of-sight velocity components.

contains multiple velocity peaks, and kinematic data at a various position angles indicates that there is no well-ordered rotation.

3. Physical Properties of the gas

The observed Ly (X extent corresponds to a volume rv 1.6 X 105 kpc3 (assuming a line-of-sight depth comparable to the projected size). Its total luminosity is LLyc; = 9.8 X 1044 erg s-l, which we use as a basis for constraining the geometry and energetics of the gas, together with the electron density ne and filling factor f. Two general physical scenarios are considered for the spatial structure of the line-emitting gas:

Line-emitting clouds distributed uniformly throughout the volume, as is the case for gas that is largely fragmented. Such fragmentation can be associated with violent mergers, radio jet / gas interactions, or cooling flows. In this scenario we find n~ f = 0.05. In comparison, Heckman et al. (1989) showed that for cooling flow galaxies, n~ f lies in the range 0.012 to 2.3 with a mean value of 0.51, and f is in the range rv 1 X 10-7


to 4 X 10-5 with a mean of 1 X 10-5 (similar to radio galaxies, e.g. van Breugel et aI. 1985). Hence we adopt a fiducial value f = 10-5 .

This corresponds to a density ne = 70 cm-3 for individual gas clouds, consistent with the general observation that extended emission-line gas regions are in the low-density limit. The corresponding mass of ionized hydrogen is 2.8 X 109 Mev. Ly 0' emission from a relatively thin shell corresponding to the outer extent ofthe observed emission (e.g. Begelman and Cioffi 1989, Meisenheimer and Hippelein 1992). This scenario might apply if the radio axis is confined by substantial amounts of gas, so that the line emission traces the interface between the radio plasma and the surrounding gas. If gas density is to remain ~ 100 cm-3 then the filling factor must increase, e.g. f ;::: 10-3 for a shell of thickness 1 kpc.

The kinetic energy of the ionized gas is EK = 1.3 X 1058 erg, obtained by equating the line-of-sight velocity dispersion with an isotropic velocity distribution, and assuming uniform density clouds.

4. Origin and dynamics of the gas

Two important questions about the Ly 0' gas are the mechanisms responsible for its extended morphology and current kinematics. The gas could be:

remnant material from an interaction with another galaxy, in which case its current kinematics is either the result of the encounter geometry or interactions with the radio plasma. deposited by an outflow, which should exhibit an increased velocity dispersion towards the centre. The lack of such a trend in the observations makes this scenario difficult to apply. "cooling filaments" from material around the galaxy; however this is unlikely since the velocity dispersion is an order of magnitude above that of typical cooling flows (100 - 200 km s-l, Heckman et ai. 1989).

If the gas motion is purely gravitational, its velocity amplitude (±400 km s-l at its maximum extent of ±70 kpc) implies a dynamical mass and timescale M dyn '" 2.6 X 1012 Mev, Tdyn '" 1 X 109 yr. The observed velocity width corresponds to an absolute magnitude MB = -22.5 (Bender et ai. 1989), and the typical MIL ratio'" 10 suggests a galaxy mass rv 1.6 X 1012 Mev, in good agreement with the dynamical mass estimate. However, Tdyn is two orders of magnitude larger than the typical radio source lifetime (106 - 107 yr, Alexander & Leahy 1987). Together with the lack of regular rotation, this implies that the gas is dynamically young, or disrupted by interactions with the radio plasma.

LYMAN ALPHA KINEMATICS IN 2104-242 389

5. Energy Input to the Gas

The observed Ly 0: luminosity requires a continuous energy input to the gas. This is seen by considering the available kinetic energy of the gas; if this is converted into Ly 0: radiation through cloud-cloud collisions with an efficiency "7, then the radiative lifetime is Trad = 4.1 X lOS "7 yr. Typically "7 ;S 0.1 (McCarthy et al. 1990b), therefore Trad is at least one to two orders of magnitude less than the source lifetime. Thus a single event (such as a merger) cannot produce the necessary energy input; a continuous energy supply is required. Two likely energy sources are photons from the active nucleus, or the radio plasma itself.

Assuming that the gas is in ionization balance, the total number of ionizing photons required is Nphot = 6.0 X lOsS s-l. If the AGN is responsible for its ionization then the solid angle subtended by the gas implies a total isotropic flux of 1.8 X lOS7 s-l. For a typical AGN power-law spectrum Fv ex: v-1.S, log U = -2.5 (e.g. Ferland and Osterbrock 1986), the observed nuclear flux density at 1200A is about two orders of magnitude lower than expected, consistent with substantial obscuration as is generally inferred for such sources.

If the energy is supplied instead by direct interactions with the radio plasma, then the gas kinetic energy can be accounted for by ;S 5% of the energy deposited into the radio lobes, if the radio luminosity of Lradio = 1.1 X 1046 erg s-l remains approximately constant over the source lifetime (Tradio ~ 106 yr). We consider a scenario where the energy is transferred through shocks resulting from cloud-cloud collisions (e.g. Sutherland et al. 1993); the radiation from the shocks photoionizes the surrounding gas. Considering shock velocities representative of the velocity differences between multiple components, i.e. rv 400 - 500 km s-l, the column density required for gas to become optically thick to ionizing radiation is 100 rv 1021 cm-2. In this case the total shock surface area and expected Ly 0: luminosity is :

Ash 3.5 X 102 n70 Is ( V S 3) kpc2 1.6 X 10 kpc

LLya,sh 1.7 X 104s n7o (400~~s-1 r (3.5 X ~~~kPc2) ergs-1

(where ne = 70 n70 cm -3, I = lO-s 15). This expected luminosity agrees well with that observed, the primary uncertainties being due to ne and I.

6. Conclusions

It has been shown that the observed properties of the Ly 0: emission-line region in the radio galaxy 2104-242 suggest substantial interaction between


the radio plasma and the emission-line gas. The energy input to the gas can be explained either in terms of nuclear ionization (the primary uncertainties being the energy distribution and degree of obscuration of the AGN) or through shocks from direct interactions between the radio plasma and the gas, in which case the predominant free parameters are the filling factor and density of the gas. However, the shock model can potentially account for both the kinematics and morphology of the gas through a single physical mechanism.

Future work should be aimed at measuring rest-frame optical emissionline ratios to provide further constraints on the extent to which nuclear photoiohization or shock ionization operate in the gas. Work is also currently in progress on detailed hydro dynamical models of interactions between radio plasma and gas, and this will allow more detailed constraints to be placed on the formation of shocks from turbulence, together with the relationship of turbulence to star-formation and the more general question of the effect of the environment of radio galaxies on their evolution.

References

Alexander, P. A. and Leahy, J. P.: 1987, M.N.R.A.S. 225, 1 Begelman, M. C. and Cioffi, D. F.: 1989, Ap.J. 322,585 Bender, R., Surma, P., Dobereiner, S., Mollenhoff, C. and Madejsky, R.: 1989, As

tron.&Ap. 217,35 Ferland, G. J. and Osterbrock, D. E.: 1986, Ap.J. 300, 658 Heckman, T. M., Baum, S. A., van Breugel, W. J. M. and McCarthy, P. J.: 1989, Ap.J.

338,48 Koekemoer, A. M., Bicknell, G. V., Dopita, M. A. and Ekers, R. D.: 1995, In preparation McCarthy, P. J.: 1993, Ann.Rev.Astron.&Ap. 31, 639 McCarthy, P. J., Kapahi, V. K., van Breugel, W. and Subrahmanya, C. R.: 1990a, A.J.

100, 1014 McCarthy, P. J., Spinrad, H., van Breugel, W., Liebert, J., Dickinson, M., Djorgovski, S.

and Eisenhardt, P.: 1990b, Ap.J. 365, 487 Meisenheimer, K. and Hippelein, H.: 1992, Astron.&Ap. 264, 455 Sutherland, R. S., Bicknell, G. V. and Dopita, M. A.: 1993, Ap.J. 414,510 van Breugel, W., Miley, G., Heckman, T., Butcher, H. and Bridle, A.: 1985, Ap.J. 290,

496

ORIENTATION EFFECTS IN QUASAR SPECTRA: DUST AND OBSCURATION

JOANNE C. BAKER Mullard Radio Astronomy Observatory, Cavendish Laboratory, Madingley Rd, Cambridge CB22HS, UK.

AND

RICHARD W. HUNSTEAD Department of Astrophysics, School of Physics, University of Sydney, NSW 2006, Australia.

Abstract. We have created a set of composite optical spectra for radioloud quasars primarily to illustrate changes in their optical properties with viewing direction. By co-adding spectra drawn from the 408 MHz Molonglo Quasar Sample in four sets according to the ratio of radio core-to-Iobe flux density, R, we find many continuous spectral trends. Together, these indicate strongly that dust reddening is prevalent at large viewing angles to the radio jet axis. Compact steep-spectrum quasars have been combined separately, their average spectra showing many distinguishing features.

1. Introduction

The suggestion that the diversity of Active Galactic Nuclei (AGN) may be increased artificially by the effects of viewing angle is gaining wide acceptance. The so-called "unified schemes" (see review by Antonucci 1993) have been remarkably successful in amalgamating sub-classes of AGN with orientation as the major discriminator. Radio-loud quasars have been described particularly well by this picture; core-dominated quasars can be portrayed simply as foreshortened lobe-dominated quasars with relativistically boosted cores (e.g. Orr & Browne 1982; Kapahi & Saikia 1982).

At optical wavelengths, however, the effects of orientation are understood poorly. It has been noticed that core-dominated quasars appear to have both brighter continua and smaller narrow-line-to-continuum ratios

391

M. N. Bremer etal. (eds.). Cold Gas at High Redshift. 391-395. © 1996 Kluwer Academic Publishers.

392 JOANNE C. BAKER AND RICHARD W. HUNSTEAD

than their lobe-dominated counterparts (e.g. Browne & Wright 1985). This has been interpreted as evidence for enhanced optical continuum emission accompanying the radio core (Browne & Murphy 1987; Jackson & Browne 1989). Orientation dependence has also been claimed for a number of broadline properties, including their shapes and strengths (e.g. Wills & Browne 1986; Jackson & Browne 1991).

Despite the success of the orientation-based unified models for quasars, they are unable to accommodate all classes, notably compact, steep spectrum (CSS) quasars (see review by Fanti & Fanti 1994) as well as the radio-quiet majority. The radio plasma in CSS quasars is thought be confined within the host galaxy, either because they are young or 'frustrated' by an unusually dense, clumpy interstellar medium (ISM). However, CSS quasars show remarkably few distinguishing features in the optical (Gelderman & Whittle 1994) and infrared (O'Dea et al. 1994).

2. The 408 MHz Molonglo Quasar Sample

To minimise orientation dependent selection biases we have defined the Molonglo Quasar Sample (MQS), selected initially at 408 MHz (Kapahi et al. in prep.). Sources with S408 > 0.95 Jy were drawn from the Molonglo Reference Catalogue (Large et al. 1981) in a 10° declination strip (-20°> h > -30° and Ibl > 20°). Complete optical identifications have been obtained down to the UK Schmidt IIIaJ survey plate limit (bJ ~ 22.5) and also from R band CCD images. Spectroscopic redshifts are available for 92 out of 101 MQS quasars to date and span the range z = 0.1-2.9.

Low resolution (FWHM 25 A) optical spectra, covering 3400-10000 A, have been obtained for 72 MQS quasars with the RGO spectrograph and Faint Object Red Spectrograph (FORS) on the Anglo-Australian Telescope (AAT). Observations were made with the slit at parallactic angle to ensure accurate relative spectrophotometry. Core-to-Iobe flux density ratios, R, have been measured from 5 GHz VLA maps at rv I" resolution. The R values have been K-corrected to an emitted frequency of 10 GHz using spectral indices calculated between 408 MHz and 5 GHz. Individual radio maps and optical spectra for the MQS will be published elsewhere.

3. Assembling the Composite Spectra

Composite spectra for MQS quasars have been constructed from 60 individual AAT spectra divided into four subsets: R 2: 1, 1 > R 2: 0.1, R < 0.1 and CSS (see Baker & Hunstead 1995). These composites include 13, 18, 16 and 13 spectra respectively.

Before combining, the spectra were each shifted to the quasar restframe, normalised at 3000A and noisy edges trimmed. By normalising the spectra

COMPOSITE QUASAR SPECTRA 393

......... ==

......... -tS == -~ S Ql. - tS

!X "" E:.., E:.., u u ::a :I: :x::

'l(

R~l

.1~R<1

R<O.l

css

1000 2000 5000

Arest (A)

Fzgure 1. Composite spectra for the M QS, separated according to their radio core dominance, R (see text). Best fit power law spectral slopes are shown as dashed lines (Baker & Hunstead 1995). Note that the spectral shapes may be less reliable near the edges (below 1600 A and above 5000A) where few quasars contribute.

at one point, rather than dividing by a fitted continuum, we have sought to preserve spectral slope characteristics. For each subset, pairs of spectra were co-added in order of increasing redshift, then intermediate pairs were combined until the final composite was created. The resulting composite spectra are displayed in Fig.I, shown on a log-log scale to emphasize the differences.

394 JOANNE C. BAKER AND RICHARD W. HUNSTEAD

4. Composite Comparison

The set of composite spectra displayed in Fig.1 reveals clear differences between quasars of different R and the CSS quasars. A more detailed quantitative comparison of the average properties of the spectra is made in Baker & Hunstead (1995). In summary, we find that with decreasing R (i) the optical continuum steepens, (ii) the 3000A broad emission feature decreases in relative strength, and (iii) the narrow-line equivalent widths, broad line widths and Balmer decrements increase.

The composite for the CSS quasars in the MQS is especially revealing. The average continuum slope is very steep and shows no deviation from a power law. Broad features, such as around 3000A are missing. Enhanced narrow-line emission is present, particularly low ionisation species such as [0 IIj)'3727 and [0 Ij.\6300. Also, the self-absorption of Mg II).2798 and low Ly alC IV ratio points to heavy absorption in these sources.

5. A Consistent Picture

Although red quasars have been noted individually in the past (e.g. Smith & Spinrad 1980; Mathur 1994), our findings for the MQS suggest further that reddening is highly aspect-dependent, occurring preferentially in lobe-dominated quasars highly inclined to the jet axis. One picture which springs to mind is that of a geometrically thick, dusty torus the outer regions of which partially obscure the light from the optical continuum source and broad-line region. Such a geometry could reproduce the red continua. and large Balmer decrements common in lobe-dominated quasars. Furthermore, aspect-dependent reddening could go much of the way in explaining the differences in narrow-line equivalent widths between core- and lobedominated quasars previously attributed to relativistic beaming and disk emission (Baker et at. in preparation).

Viewing-angle dependent reddening is entirely consistent with the unification by orientation of powerful, radio-loud quasars and galaxies (e.g. Simpson et at. 1995; Rawlings et at. 1995) In addition, the probable lack of a sharp viewing angle cutoff between reddened lobe-dominated quasars and radio galaxies suggests that misclassification may also be a problem. Another important consequence is that red, lobe-dominated quasars are more likely to be missed on blue survey plates, introducing an additional colour bias into nearly all quasar samples.

The MQS composites have revealed striking new differences between the average spectra of CSS and non-CSS quasars, raising again the question of their role in the AGN family. Dust reddening, as inferred from the steep continuum and large Balmer decrement, may be significant in these sources. Also, thermal optical components, e.g. the big blue bump, may be absorbed

COMPOSITE QUASAR SPECTRA 395

or suppressed. The large equivalent widths of the low ionisation narrow lines points to strong jet-ISM interactions, perhaps mediated by shocks. The presence of a dense ISM in CSS quasars is also supported by broad-line self-absorption.

6. Conclusions

The continuous trends in optical properties illustrated in Fig.l for MQS quasars support strongly the unification of core- and lobe-dominated quasars predominantly by orientation. Moreover, they demonstrate that the nuclei of many lobe-dominated quasars may be viewed through layers of dust. The aspect-dependence inferred for reddening in radio-loud quasars points to an intrinsic origin for the dust, perhaps associated with the hazy outer regions of a torus.

CSS quasars in the MQS are also found to be red on average, but the new optical characteristics presented here for these sources point to a more complex picture, whereby absorption and jet-ISM interactions are important.

References

Antonucci, R. R. J. 1993, ARAA, 31, 473 Baker J. C. & Hunstead R. W. 1995, ApJL, 452, L95 Browne 1. W. A. & Murphy D. W. 1987, MNRAS, 226,601 Browne, 1. W. A. & Wright, A. E. 1985, MNRAS, 213, 97 Fanti, C. & Fanti, R. 1994, in ASP Conf. Ser., Vol. 54, The Physics Of AGN, ed. G. V.

Bicknell, M. A. Dopita & P. J. Quinn (San Francisco: ASP), p341 Gelderman R. & Whittle M. 1994, ApJS, 91, 491 Jackson, N. & Browne, 1. W. A. 1989, Nature, 338, 485 Jackson, N. & Browne, I. W. A. 1991, MNRAS, 250, 422 Kapahi, V. K. & Saikia D. 1982, A&A, 3, 465 Large, M. I., Mills, B. Y., Little, A. G., Crawford, D. F. & Sutton, J. M. 1981, MNRAS,

194,693 Mathur, S. 1994, ApJL, 431, L75 O'Dea C. et aZ. 1994, in ASP Conf. Ser., Vol. 54, The Physics Of AGN, ed. G. V. Bicknell,

M. A. Dopita & P. J. Quinn (San Francisco: ASP), p209 Orr, M. & Browne, I. W. A. 1982, MNRAS, 200, 1067 Rawlings, S., Lacy, M., Sivia, D. S. & Eales, S. A. 1995, MNRAS, 274, 428 Simpson, C., Ward, M. J. & Wilson, A. S. 1995, ApJ, in press Smith, H. E. & Spinrad, H. 1980, ApJ, 236,419 Wills, B. J. & Browne, I. W. A. 1986, ApJ, 302, 56

EFFECTS OF DUST AND RESONANCE SCATTERING ON THE UV SPECTRUM OF RADIO GALAXIES

M . VILLAR-MARTiN ST-ECF, Karl-Schwarschild-Str 2, D-85748 Garching Germany, and Max-Planck-Institute fur extraterrestrische Physik Giessenbachstrasse, D-85740 Garching, Germany

L.BINETTE Observatoire de Lyon, 9 avo Charles Andre F-69561 Saint-Genis-Laval Cedex, France

AND

R.A.E. FOSBURY ST-ECF, Karl-Schwarschild-Str 2, D-85748 Garching Germany Affiliated to the A strophysics Division, Space Science Department, European Space Agency

Abstract. In the powerful, high red shift radio galaxies, it is believed that the dominant source of ionization for the interstellar gas is the hard radiation field associated with the active nucleus. The photon source is generally external to the clouds being ionized and so the geometrical perspective from which the gas is observed and the presence and distribution of dust must be properly accounted for in the diagnostic process. In this work, we examine the formation of the three strong lines: C IV .\1549, Lyoo and C III] .\1909. We find that the observed trends, in particular the high C IV .\1549fLyoo ratio, are often better explained by geometrical (viewing angle) effects than by the presence of large quantities of dust either within or outside the excited clouds. We show that neutral condensations along the line-of-sight, by reflecting photons near the wavelength of Lyoo, can increase the observed C IV fLy a ratio. The existence of H I absorption clouds (i.e., mirrors) external to the emission region lead also to the prediction of large, diffuse haloes of what appears to be narrow Lyoo emission.

397


398 M.VILLAR-MARTiN ET AL.

1. Models and comparison with observed UV lines

The strong, rest-frame ultraviolet emission lines (shifted into the optical) observed in high redshift radio galaxies provide one of the principal diagnostics in establishing the state of the interstellar medium in galaxies at early epochs. For objects of high z, the UV rest-frame spectrum is generally dominated by Lya, e IV >'1549, He II >'1640 and e III] >.1909. We have collected from the literature the observed ratios of several high z narrow line radiogalaxies. We have built the diagnostic diagram e IV ILy a VS. e IV Ie III], in which we compare the position of the objects with photoionization models that not only consider the effects of internal dust but also the effects of the observer's perspective.

10 I:- 10 I:-

.. 1 r

~ 1 I:-

u

---------·0--0.1 t-::============== 0.1 I:-

0.1 1 10 0.1 1 10 elv/em] elY/em]

Figure 1. Observed UV emission line ratios. Filled triangles are observed ratios of objects for which the measurement of the three lines was possible. The open diamond is the average radiogalaxy spectrum of McCarthy (1993).

A notable observation is the very small dispersion of the e IV Ie III values com pared to the e IV I Ly a and e IV IHe II ratios (see Fig. 1). For some objects, Lya is observed to be fainter with respect to e IV than predicted by dust-free photoionization models. It is usually claimed that the destruction of Ly a photons by resonance scattering in the presence of dust is the explanation for its faintness, but why does not the same process reduce e IV which is also a resonance line? Is there an alternative explanation for this selective dimming of Lya?

UV SPECTRUM OF RADIO GALAXIES 399

!:,!~~g~~~ PlZ

-----~ 'C " "

<l .~Q) 0

<l<l i»'C ..

00 :=leu" -;;

<1: .~ " .!S.~ d 1::

~ .... <l 0 "'0" " ... ~ QI

'fn-l' '" <l

-----~ H' ,Ho & H' HO

-----~

1> 'Ixu/o'

10 •

..... .... Sack

• 1

_ Swn

---!'rODt .. . !

~ <J

--- ----------,:'- - -------. ----------o 1 ----------~:---------- ___ L~ ___ • . ---------------------- ---7-------

0.1

: /

1 elV/ClUj

I /

10

Figure 2. Left: Adopted slab geometry. Right: Influence of viewing geometry on the UV line ratios (see text). This is the dust-free case.

1.1. THE ROLE OF GEOMETRY

To answer these questions, we have examined the geometrical aspects of the line formation process. Each emitting cloud is approximated as a plane parallel slab which contains a fully ionized region and a partially ionized zone where HO and H+ coexist. In principle there can be an additional neutral zone which does not contribute to the emission line intensities (Fig. 2 left). When the line opacity is important as it is for C IV and Ly 0:, line scattering occurs and the line photons do not escape isotropically but in the direction of highest escape probability which can be shown to be the front for the photoionized slab depicted in Fig. 2 (left) (Villar-Martin, Binette & Fosbury, 1996). In Fig. 2 (right) we present detailed photoionization calculations for which we have used the photoionization code MAPPINGS (Binette et al. 1993). We present a sequence of dust-free models where the ionizing parameter U = J 1" dv I hvc nH, is the variable. As a guide in choosing the input parameters, we have assumed that the excited gas of high z radio galaxies has quite similar properties to that of the low red shift objects: a power law of index 0: ~ -1.4 (1", IX v+ a ) as ionizing energy distribution, solar metallicity (lower abundances do not in any way affect the conclusions reached here), gas density nH = 100 cm-3 . We find that for this input parameters, the U optimum value to reproduce the C IV IC IIIJ

ratio observed is U ~ 0.1. Our diagrams will cover the range of [O.OI,O.IJ in U.

We distinguish between the spectrum seen from the back ( dotted line -equivalent to observing the clouds through the PIZ) from that seen from


the front (dashed line - equivalent to seeing the UV irradiated side) (Fig. 2 (right)). The differences are striking: both Lya and to a lesser extent C IV

are fainter when seen from the back. The CIII] line is isotropic (optically thin) in the dust-free case.

How would this apply to the emitting line regions of the powerful radio galaxies? If this gas is photoionized mainly by the collimated UV radiation emitted by the hiden AGN, it is characterized by an open geometry where many perspectives are possible. In a very simplified scheme, we can imagine that the clouds seen from the nearside cone are seen from the back perspective in our slab calculations while clouds on the far side would be seen from the front. Figure 2 suggests that perspective effects alone (without any dust) is sufficient to explain the weak Lya seen in some objects. Adding a modest neutral zone beyond the PIZ would strongly increase the C IV jLy a ratio without affecting in any way the C IV jC III] ratio. It seem:;, therefore, that a geometry where we see preferentially the ionized gas from the side of the PIZ gives us a clue towards explaining the weakness of Ly a in some objects.

The studies of McCarthy et al. (1991) which emphasized the one-sideness of the line brightness suggest that the observer with limited spatial resolution at very high z will be biased towards either a back or front dominated perspective depending on whether it is the near or the farside illuminated cone which is intrinsically brighter. Our proposed interpretation of the high C IV jLy a ratio of some objects in Fig. 2 is that they correspond to the case where the brightest clouds are seen from behind.

10

• 1

~ " u

0.1 1 eIV/e11J]

.. ..

10

10

Back

_ Sum

~ - - Front

,.' - - - --- - --- - - --.- -- - - - -- -.- --- --- ---

o 1 -=======::::=:~=:::::::'====--::=7'~;-::~::: /

01 1 eIV/e11J]

I /

10

Figure 3. Left: Effects of varying the amount of internal dust as seen from the front. perspective. Right: Effect.s of perspective combined wit.h small quantities of int.ernal dust. The dust-free U sequences of Fig. 2 are repeated here. The dotted almost vertical line corresponds to the model with U = 0.1 (with metallicity 0.3 Z,un) in which the dust content is increased in proportion up to JL = 0.017.

UV SPECTRUM OF RADIO GALAXIES 401

1.2. THE ROLE OF THE DUST

We have observational evidence that indicated the existence of dust internal to the EELR of high z radio galaxies. The aligned blue polarized continuum is the result of scattering by dust of the anisotropic nuclear radiation field (e.g., Cimatti et at. 1993). Therefore, we have investigated the effects that such dust could have on the UV line ratios. The dust content of the photoionized plasma is described by the quantity 11 which is the dust-to-gas ratio of the plasma expressed in units of the solar neighbourhood dust- togas ratio. As shown in Fig. 3, left, even with such a high concentration of internal dust given as by 11 = 1 (equivalent to that in solar neighborhood cold clouds), it is not possible to reproduce the high ratio C IV /Ly 0' ~ 1 which can be observed and has been attributed to dust. When the clouds are viewed from behind, even with 11 = 0.3 they are sufficiently opaque that all the UV lines are severely absorbed. To produce an acceptable spectrum (not reddened Clv/Cm] and Clv/Hell ratios), amounts of dust as little as 11=2 (2% the amount in our ISM) are required (see Fig. 3, right). Such a process can reproduce the weak Ly 0' objects although this result is obtained only for the back spectrum, emphasizing that perspective is the more important factor.

2. Neutral gas mirrors

So far we have considered the effects of scattering by gas in the line emitting clouds but any extrinsic neutral gas component with a non-negligible covering factor could affect the appearance of the object in the Ly 0' light. Let us suppose that this outer material is broken up into cold gas clumps which are randomly distributed. In such a case, the Ly 0' photons which leave the ionized cones will escape the region through the holes between the external clumps while others will strike the neutral clumps and be immediately scattered away to escape eventually through another hole in a different direction. The bulk of the Ly 0' luminosity would be preserved but redistributed on an apparently larger scale than the true line emitting clouds. Such a geometry would explain the diffuse Ly 0' ha.loes observed in some radio galaxies, like PKS2104-242 (McCarthy et al. 1990b). Only a more closed geometry would result in a significant destruction of Ly 0' by dust.

Also, the reflection of Ly a photons by intervening neutral screens could explain, with no need of dust, the absorption features observed in the Ly 0'

profile of some high z radio galaxies (Rottgering et al. 1995). In some objects, there are regions where the only emission line detected

is Ly 0', like in 3C294 (McCarthy et ai. 1990a), while other zones show also C IV, He II and C III. We suggest that Lya corresponds to scattered


continuum and BLR Lya photons by neutral hydrogen that could lie in some cases in H I clumps or in regions where the gas has so low densities that the clouds are totally ionized, but the lines are not strong enough to be detected. Calculations show that the small column densities of HO expected in this last case, would very efficiently reflect the nuclear Lya emission.

References

Binette, 1., Wang, J., Villar-Martin, M., Martin, P.G., Magris, C.M., 1993, ApJ, 414, 535

Cimatti, A., di Serego Alighieri, S., Fosbury, R., Salvati, M., Taylor, D., 1993, MNRAS, 264, 421

McCarthy, P.J., Spinrad, H., van Breugel, W.J.M., Liebert, J., Dickinson, M., Djorgovski, S., Eisenhardt, P., 1990a, ApJ, 365,487

McCarthy, P.J., Kapahi, V.K., van Breugel, W.J.M., Subrahmanya, C.R., 1990b, AJ, 100, 1014

McCarthy, P.J., van Breugel, W.J.M., 1991, ApJ, 371, 478 McCarthy, P.J., 1993, ARA&A, 31, 639 Rottgering, H.J.A., Hunstead, R.W., Miley, G.K., van Ojik, R., Wieringa M.H., 1995,

MNRAS 277, 389 Villar-Martin M., Binette 1., Fosbury R.A.E., 1996, in preparation

HST OBSERVATIONS OF RADIO GALAXIES AT Z '" 1

P.N. BEST

Cavendish Laboratory Madingiey Road Cambridge CBS OHE United Kingdom

Abstract. High redshift radio galaxies provide a unique opportunity to study the distant Universe, enabling investigation of both the formation and evolution of their stellar populations and, by the strong interactions of the radio components with their environments, the structure of the interstellar and intergalactic medium at cosmological redshifts. In this contribution, Hubble Space Telescope (HST) observations are presented for the eight 3CR radio galaxies in the redshift range 1 ;S z ;S 1.3. The optical morphologies of these sources are highly dependent upon their radio properties, and it is argued that they show evidence for large-scale star formation induced by the radio jets.

1. Introduction

The tendency for the optical emission of powerful radio galaxies with redshifts z ~ 0.6 to be elongated and aligned along the radio axis was first noted in 1987 by McCarthy et ai. and by Chambers et ai. Many models have been proposed to explain this effect, but none is entirely satisfactory. The most promising theories involve massive star formation associated with the passage of the radio components - a phenomenon observed on smaller scales in the relatively nearby Minkowski's Object - and scattering of light from an obscured active nucleus either by dust or by electrons. The detection of optical polarisations of up to 10% in some of these sources indicates that scattering (or some other non-thermal process) must be partly responsible, although it need not be the process which dominates the observed alignments. See McCarthy (1993) for a review.

403


404 P.N. BEST

An almost complete sample of 28 3CR radio galaxies with redshifts between 0.6 and 1.8 has recently been imaged using the HST for one orbit in each oftwo filters, selected to be at rest-frame U and V bands l . The galaxies were also observed using the VLA at 8.4 GHz in A-array (at comparable resolution to the HST) and using UKIRT in J and K wavebands.

Eight of the galaxies lie in the redshift range 1 ~ z ~ 1.3, the range in which the image quality is highest. By considering only this narrow range, redshift and radio luminosity effects are eliminated. In other respects these sources are representative of the complete sample. HST images of these galaxies are displayed in Figs. 1( a-h), in order of increasing radio size2 .

Overlaid upon them are contours of radio emission or, in the case of the largest radio sources, lines indicating the direction of the radio axis.

2. Discussion

The spectacular HST images confirm the suggestion of Hammer et al. (1991) that 3C368 (Fig.1b) is contaminated by a foreground M-dwarfstar (the bright unresolved knot at the centre). Comparison with the UKIRT image shows the nucleus of this galaxy to coincide with the smaller emission region one arcsec to the north. This galaxy and 3C 324 (Fig. 1c) were discussed by Longair et al. (1995). When the HST image of 3C 324 is convolved to 1 arcsec resolution and a cut made along its axis, it appears fiattopped, whilst the infrared image is sharply peaked. This suggests that the central region of optical emission may be obscured by a large dust lane. "Dumbbell"-shaped structures seen in many other sources, including 3C 252 and 3C 356 (Figs. 19 and 1h), suggest that dust obscuration of the nuclear regions may be a common phenomenum.

The most striking feature of these images is that whilst most show alignment of the optical emission along the radio axis, the form of this alignment differs greatly from source to source. In particular, there is a clear evolution of the optical morphology as the radio size increases: the smallest radio sources consist of a string of bright knots tightly aligned along the radio axis, whilst the larger ones have fewer (generally no more than two) bright components, which are more highly separated - they also display more diffuse emission. To quantify this result, the parameter "Component Alignment Strength" is defined as:

ac = (N -1)((1- !:::.fJ/90)

1 In general, one filter contained the strong [011]3727 line, whilst the other was free of major emission lines. The morphological similarity of the two images indicates that the results are unaffected by line emission.

2Strictly "projected radio size" is meant. According to unification schemes of radio galaxies and radio loud quasars, galaxies have their radio axis orientated within 45 degrees of the plane of the sky, so the projected and actual physical sizes should be similar.

RADIO GALAXIES AT Z ~ 1 405

11 U1 28

1ft'SlI.S 6.4

41

40

4'-'1

24

Figure 1. In order of increasing radio size: (a) 3C266 (z = 1.272, upper left), (b) 3C368 (z = 1.132, upper right), (c) 3C324 (z = 1.21, centre), (d) 3C280 (z = 0.996, bottom). One arcsec corresponds to approximately 8.5 kpc

where N is the number of bright components, f. is the ellipticity of the optical emission and 6.0 is the position angle difference between the bright components and the radio axis. The inverse correlation between this parameter and radio size, significant at the 99% level, is shown in Fig. 2.

If radio size is a measure of the age of the radio source, this result in-

406 P.N. BEST

54

52 ,.

40 0050 -.....

0223 44.25 44.0 43.75 43.5 43.25 43.0 42.75 52.S

. "".

211.0

18

. ,

-45

.' ':~" .. ;",

124716 '. , ,:',

11 48 57.0 $6.' 56.6 5&.4 40

44

,". ::~.;i~t: . ..... ::;.{t·,

::it~.{.:. ;" . 35

42

.. " .. ~

40 .. ~. '30

~:

35 40 3tI +-...... _...;,....;,....-___ ---...:...--.---' .. ;.,,',.--,..-+ 3;1.6 111133.4 33.2 33.0 2419.75 19.5 19.25 19.0 18.75

Figure 1 continued. (e) 3C65 (z = 1.176, top), (f) 3C267 (z = 1.144, centre left),

(g) 3C 252 (z = 1.1 OS, lower left), (h) 3C 356 (z = 1.079, right).

,.,

I * IN

* ~ *

* *

i- * I 0 * *

--I"". Figure 2. The correlation between "Component Alignment Strength" and radio size.

RADIO GALAXIES AT Z ~ 1 407

dicates that there must be a very active phase as the radio components pass through the host galaxy, and following this phase the system appears to "relax". It is difficult to reconcile this picture with many theories of the alignment effect, for example scattering, but it is consistent with jetinduced star formation models. Shocks induced by the passage of the radio hot-spots through the host galaxy cause massive knots of bright young stars to be formed, which then disperse over the lifetime of the radio source. In this scenario, scattering from dust and ionised gas associated with the star-forming regions would be responsible for producing the observed optical polarisations. There would be no need for dust and ionised gas to be distributed throughout the whole galaxy as required by scattering models.

The alternative hypothesis, which on its own seems unlikely to account for the complete range of observed radio sizes (particularly as the spectral aging analysis of Liu et al. (1992) indicates that 3C 266 and 3C 280, two of the smallest sources, are young radio sources), is to associate the physical size of the radio source not with its age, but instead with the density of the surrounding medium. Sources would be small because of slower hot-spot advance speeds due to the increased resistance of denser X-ray gas within the cluster. This same denser gas could lead to an increase in scattering, producing bright knots of emission in small sources - see Malcolm Bremer's contribution (this volume) for further discussion of this model.

3. Conclusions

The high resolution HST images have provided the first clear indication of the nature of the alignment effect in powerful radio galaxies. The dominant alignment mechanism appears to be massive star formation associated with the passage of the radio beams. Other effects, generally present at lower levels, may, however, still be pronounced in individual sources.

This has profound implications for the star-formation histories of these galaxies. Further modelling of the interaction of the radio beam with its surroundings is currently in progress to determine the masses and distributions of ionised gas and dust throughout and surrounding these galaxies.

References

Chambers K. C., Miley G. K., van Breugel W. J. M., 1987, Nature, 329, 604 Hammer F., Le Fevre 0., Proust D., 1991, Astrophys. J., 374, 91 Liu R., Pooley G., Riley J. M., 1992, Mon. Not. R. astr. Soc., 257, 545 Longair M. S., Best P. N., Riittgering H. J. A., 1995, Mon. Not. R. astr. Soc., 275, L47 McCarthy P. J., 1993, Ann. Rev. Astron. Astrophys., 31, 639 McCarthy P. J., van Breugel W. J. M., Spinrad H., Djorgovski S., 1987, Astrophys. J.

Lett., 321, L29

INSTRUMENTAL DEVELOPMENTS

STUDIES OF COLD GAS IN THE EARLY UNIVERSE WITH LARGE MILLIMETER ARRAYS

ROBERT L. BROWN National Radio Astronomy Observatory 520 Edgemont Road Charlottesville, VA 22903-2475 U.S.A.

Abstract. Our knowledge of the process by which cold gas in the early universe accumulates and forms stars is limited by our inability to image the gas. The next generation of millimeter and submillimeter wavelength arrays will allow us to explore the gas content of forming galaxies at the same resolution at which HST and the new 8 m class optical/lR telescopes will reveal early generations of stars. The dust that obscures our view of the early universe at optical wavelengths becomes an asset at millimeter/submillimeter wavelengths where the thermal dust emission can be imaged. With the new millimeter arrays astronomers will: (1) image the mass segregation and kinematics of hierarchical galaxy formation; (2) have the ability to detect thermal dust continuum emission from more than 15 million galaxies with an observation time of less than a minute per galaxy (indeed, such cosmological lR-luminous galaxies will be a source of confusion to every continuum observation); and (3) detect atomic and molecular spectral line emission from normal galaxies such as the Milky Way at redshifts greater than one.

1. Introduction

Understanding the process by which galaxies form is one of the most active areas of astronomical research, and it is also one of the most observationally challenging. Theoretical models make specific predictions about the way mass accumulates in the early universe but that mass is principally gaseous while the observational tools we are accustomed to using for our studies capture starlight, re-radiated starlight from dust, or an indirect

411


412 ROBERT L. BROWN

manifestation of stars such as radio continuum emission from relativistic electrons accelerated in supernovae or expelled by a supermassive nuclear black hole. We have not been able to image the gas directly. The utility of optical searches for starlight from protogalaxies are also limited by extinction from whatever dust may exist in these systems (Cowie, Songaila, and Hu 1991; Djorgovski and Thompson 1992; Thompson, Djorgovski, and Trauger 1993; De Propris et al. 1993). Fortunately the rapid advance of millimeter wavelength technology has now made it possible for us to observe molecular species in cosmologically distant objects and to use this tool to assess not only the nature of the galaxy formation process, but also to investigate the chemical enrichment of gaseous matter in the early universe. Moreover, at millimeter wavelengths the dust is not a hindrance but an asset, it too can be imaged. Such explorations are among the principal scientific goals of the next generation of millimeter wavelength imaging arrays. In order to establish a framework for discussion of the earliest evolution of objects galaxies and clusters of galaxies as a function of cosmological epoch let us outline the galaxy formation process in general terms as a progression of five phases:

I. gravitational accumulation of gaseous mass II. fragmentation and mass segregation

a. gas filaments collide

h. first stars form (?) III. gravitational coalescence of fragments leading to gaseous "protogalax

ies" IV. luminous "protogalaxies"

a. tidal interactions, cannibalization

h. formation of central gaseous concentration

V. interacting "protogalaxies"

a. formation of a supermassive black hole in the nucleus

h. AGN, QSO manifestations

We can now compare this progression with the observational tools we have at our disposal for study of the process. In Phase I the gaseous mass accumulated is essentially all H 1. The accumulation process itself, as the computer simulations presented at this meeting show, is via a complex web of overlapping filaments of extremely large angular scale (tens of degrees) occurring at z > 5. Observationally, the redshifted HI 21 em line and the H I fine-structure lines are the only possible probes. In Phase IIb, and all later phases, where successive generations of stars are present, we can expect chemically enriched gaseous

STUDIES WITH LARGE MILLIMETER ARRAYS 413

matter and dust, both of which are amenable to observation with sensitive millimeter wavelength synthesis telescopes. It is these observations that will comprise the basis of the present discussion. Of particular interest to us is Phase III where gaseous fragments begin to coalesce until the point when a galactic mass, in gas, is present but few stars have as yet formed. It is this stage of galactic evolution that millimeter and sub-millimeter astronomy is uniquely able to explore and it is also the phase most discriminating for us in helping us understand the physics of galaxy formation. In the latter two phases of galactic evolution we infer that tidal interaction among the high concentration of gas-rich galaxies is the dominant effect driving galaxy evolution. Gas clouds in the young galaxies see the interactions as dissipative, causing the gas to lose angular momentum and allow it to sink into the galactic nuclear potential. The rapid accumulation of an extremely high gas surface density, "" 103 - 104 M0 pc- 2 , leads to phenomena we describe as starburst, AGN or even QSO activity, the details of which will remain poorly understood until we have an opportunity to resolve the gas kinematics in a large sample of objects with the next generation of large millimeter wavelength synthesis telescopes. The first CO observations of the brightest of such objects are summarized elsewhere in this meeting. In the material presented below I will concentrate on three subjects that will be rich areas of research for future large millimeter arrays: (1) the role of CO observations in helping us understand the earliest phases of galaxy formation, Phases II and III above; (2) identification of luminous protogalaxies by means of their dust continuum emission; and (3) an observational methodology for detecting an individual "normal" galaxy such as the Milky Way at cosmological distances.

2. CO observations of hierarchical galaxy formation

If we understand correctly that galaxies form hierarchically through gravitational accumulation of gaseous mass fragments, which may be described by a mass spectrum such as that seen in the distribution of absorption line column densities f( N) oc N-3 / 2 (Tytler 1987), then we expect that in the vicinity of each forming galaxy of mass M we will find approximately three forming galaxies of mass M /2 and 30 or so gas clouds of mass M /10. These ideas have been quantified by Wolfe and his collaborators (e.g., Wolfe 1993), who further demonstrate that the spatial extent of a region of hierarchical galaxy formation may be many hundreds of kiloparsecs to several megaparsecs. The corresponding angular scales are from one to several arcminutes. Since any of the gas clouds that have experienced a episode of star formation will be enriched in heavy elements, we can expect molecular line emission from the hierarchical ensemble of gas clouds. Detection and

414 ROBERT L. BROWN

imaging of this emission will be an important goal, and a feasible goal, of the next generation of millimeter wavelength array telescopes. Presently, the only region of the sky for which we have an indication for CO emission associated with hierarchcical galaxy formation is the damped Ly 0' absorption line region at z = 3.137 toward the QSO PC 1643+4631A (Frayer et al. 1994). The redshifted CO(1-0) and CO(3-2) lines detected toward PC 1643+4631A both have line widths of nearly 700 km S-1 and an integrated line flux of 10 Jy km s-1 (giving a line flux density across the line profile of 15 mJy). The corresponding CO line luminosities are'" 5 X 107 to 2 X 108 L0 , nearly 1000 times the CO luminosity of the Milky Way. The total mass of gas implied by the observations if the z = 3.137 molecular clouds are similar to galactic molecular clouds is 5 X 1011 - 5 X 1012h-2 M0 . Since the CO observations were made with single dish telescopes of beamwidths '" 1', we cannot determine the spatial extent of the emission region nor can we image the distribution or kinematics of the molecular clouds involved. These tasks await future millimeter-wave synthesis arrays. We expect, however, from Wolfe's (1993) analysis that there are multiple CO emission regions corresponding to the location of galactic gas fragments spread over an area as large as, or larger than, the beam. It is exactly this conclusion that needs to be tested because it is so fundamental to our understanding of galaxy evolution. Lacking the sensitive CO interferometry needed, attempts have been made both at infrared and radio wavelengths to image the gas and galaxies that may comprise the z = 3.137 damped Lyman alpha region toward PC 1643+4631A. In an I-band image of this field limited at 25.9 magnitudes, Hu and Ridgeway (1994) find approximately 50 separate objects visible within l' of the QSO position. This is an uncommonly densely populated field. Interestingly, from broadband photometry of the PC 1643+4631A region Hu and Ridgeway identify two of these objects with I - ]( '" 6.5. These extremely red objects, the two reddest objects known anywhere, are not only within one arcminute of each other but they line in the same field from which the high redshift CO emission has been seen (Frayer et al. 1994). While no redshift is known for either red object, it is important that future narrow-beam CO observations are targeted at these two galaxies, each of which must have a substantial dust content. They are therefore the objects most likely to have a significant gas content also.

A second approach to image the mass concentrations associated with this damped Lyman alpha absorption system is to search for non-thermal synchrotron emission resulting from the cumulative supernovae of prior starburst episodes. Radio images at 8.4 GHz of the PC 1643+4631A field have been made by Frayer et al. (1996) as shown in Fig. 1. The positions of all the galaxies found by Hu and Ridgeway (1994) are also shown in


(

311 36 34 RIGHT ASCENSION (B1SS0)

Figure 1. VLA 8.4GHz map of the region near the QSO PC 1643+4631A. Very red galaxies, I - K ? 3 are shown as triangles, while those with I - ]( < 3 are shown as crosses. The galaxies are from H u and Ridgeway (1994) numbered sequentially in increasing K magnitude. The VLA beam is 116" x 94"; the contour levels are -15, 15, 20,25,30,35,40, and 50 Jdy beam-1 .

Fig. 1. The reddest such galaxies (I - J( > 3) are plotted as triangles, the others as crosses. With but two exceptions, all the Hu and Ridgeway galaxies coincide with regions of radio continuum emission; the positional agreement is best for the reddest galaxies. Although none of the Hu and Ridgeway galaxies has a measured redshift, we can use the correlation between CO flux, radio continuum, and infrared flux to constrain the nature of this distribution of sources. The correlation of radio emission from a disk population of supernovae and the infrared emission from dust heated by disk stars that are the progenitors of the supernovae is expressed as (Helou et al. 1985; Condon 1992)

_ 10 ( FIR ) -10 (S(1.4 GHZ)) q- g 3.75x1012 Wm-2 g Wm- 2

(1)

416 ROBERT L. BROWN

and q = 2.3 for a wide range of galaxies. Here

( FIR ) = 1.26 X 10-4 (2.5SS60 /Lm + SlOOJtm) Wm- 2 Jy

(2)

where S60Jtm and SlOOJtm are the rest frame flux densities at 60/-Lm and 100/-Lm, respectively. Assuming that all the flux density in Fig. 1 is from galaxies at z = 3.14 and extrapolating it to 1.4 GHz rest using a spectral index of -O.S, we find the observed 100/-Lm flux density will be less than 400 mJy, below the IRAS limits. However, we can also use this value and the relation between 100/-Lm flux density and CO integrated line strength

log ( SeQ ) = -1.76 + 10 (S100 Jtm) (3) Jykms-1 g mJy

to determine that we should expect to see an integrated CO flux of SeQ "" 7.3 Jy km s-1 from this ensemble of radio galaxies. This is in very good agreement with the observed value of SeQ"" 10 Jy km s-1 for the red shifted CO(1-0) line (Frayer et al. 1994). Finally, the radio continuum emission seen in the VLA image is mostly confined to a ring-like structure 45" in diameter. At z = 3.14 this corresponds to a linear scale Rkpc = SO h- 1

(qo = 0.5) which, taken together with the CO line width and using an inclination of 45°gives a dynamical mass of M "" 6 X 1012 MG. This compares well with the gas mass inferred from the CO observations and indicates that the mass of the system is dominated by gas, not stars. It is also what we expect for a young galactic system forming from a hierarchy of galactic and subgalactic mass concentrations. The observations of the damped Lyman alpha absorption region toward PC 1643+4631A provide important criteria for the design of future millimeter arrays. These criteria include:

need for wide-field imaging: hierarchical galaxy formation occurs over regions in the sky at least 45" in extent; need for high sensitivity: the peak CO emission line flux densities from a region many arcsecond in extent are not likely to be greater than "" 10mJy; need for exceptionally good surface brightness sensitivity: we can expect to need to image lines with brightness less than 0.1 mJ y I arcsec2.

3. Dust continuum observations of luminous protogalaxies

One of the principal scientific goals for the U.S. Millimeter Array was the desire to search for the redshifted thermal dust continuum emission from evolving galaxies. This idea built upon the IRAS observations of the existence of a large population of ultraluminous infrared galaxies, the prototypes for which are MS2 and Arp 220. In both these galaxies the bulk of


the bolometric luminosity is radiated at infrared wavelengths. This means that as such objects are moved to much greater distances they become increasingly difficult to observe in the near and mid infrared. But longward of the thermal peak, on the Rayleigh-Jeans portion of the spectrum sampled at millimeter and submillimeter wavelengths, the steep "K-correction" fully compensates for the 1/r2 dimunition with increasing distance. The result is that if we can detect the millimeter-wave thermal continuum emission from a galaxy such as M82 at a redshift of a few tenths, we can also detect that same galaxy at the same observing wavelength when the galaxy is moved to redshifts as high as 10. The observed flux density will be essentially unchanged as the object is moved to high z. All this is very encouraging since it provides a very simple way for us to identify luminous galaxies at all cosmological epoches. The challenge lies in the recognition that the galaxies will be faint by contemporary standards: most galaxies will be "" 1 mJy at observing wavelengths of 1100-850 micrometers. The faintest objects now detectable are at the few milli-J ansky level. Fortunately the next generation of millimeter arrays will be several orders of magnitude more sensitive. Calculations of the number of IR-Iuminous galaxies that are detectable by means of observations of their thermal continuum emission at millimeter and submillimeter wavelengths have been done by Franceschini et al. (1991) and Blain and Longair (1993, 1996). The estimates are based on the IRAS counts ofIR-luminous galaxies in the local universe, extrapolated backward in time to earlier cosmological epochs using models of evolving stellar populations and an evolving metallicity. Since the ultimate source of energy in these models is stars, we have constraints provided by CO BE observations and by the optical colors of galaxies. The estimates of the thermal re-radiation should be reasonably secure. Counts at millimeter wavelengths of the population of evolving IR galaxies will constrain the details of competing models of stellar evolution over cosmological epochs from 2 to 5 or even earlier (Blair and Longair 1996).

The number of galaxies potentially detectable with the sensitivity of the next generation of millimeter wavelength arrays is illustrated in Fig. 2. This figure, from Franceschini et al. (1991), shows the integral counts of galaxies per steradian as a function of flux density for wavelengths of 25, 60, 180, 800, and 1300 microns. For millimeter arrays the counts at 800 and 1300 microns are germane and here we can see how steep the cumulative curves are. With an instrumental limiting flux density of 10 mJy, there are only a few thousand galaxies per steradian to be seen at 1300 microns, but for an instrumental limiting flux density two orders of magnitude lower, 0.1 mJy, there are 10,000 times more galaxies per steradian. It is this enormous leverage with increasing sensitivity that the future millimeter arrays will

418 ROBERT L. BROWN

10-3

S (Jy)

Figure 2. Integral counts of galaxies at the wavelengths (microns) shown (from Franceschini et al. 1991).

exploit; this is illustrated in Table 1.

TABLE 1. Cumulative numbers of thermal IR galaxies over the sky

Flux density (mly)

1.0

0.1

0.01

N urn ber of galaxies 230 GHz(l) 345 GHZ(2)

1.5 X 107

3.8 X 108

5.0 X 109

7.5 X 107

1.6 x 109

1.5 X 1010

(1) 1300 ILm; (2) 850 ILm

The sensitivity goals for planned millimeter wavelength arrays are still being refined, but to be specific, the goals for the U.S. MMA project (Brown 1996) are shown for these same two wavelengths, 1300 and 870 microns, in Table 2 for integration times of one minute to ten hours.

We see that at an observing frequency of 230 GHz, the MMA can detect the thermal dust continuum emission from 13 million galaxies at a signal to noise of ten or greater in one minute of integration time for each galaxy. At 345 GHz, 75 million galaxies are detectable in a minute at a signal to


TABLE 2. MMA continuum sensitivity goals (mJy/beam)

Integration time ~S(230 GHz) ~S(345 GHz)

1 minute 0.10 0.17 1 hour 0.017 0.028

10 hours 0.004 0.007

noise greater than five. In fact, so many galaxies at cosmological redshifts are detectable that they will be a significant source of "confusion". From the numbers above we expect to find at least one confusing galaxy at a flux density of 0.1 mJy in each MMA primary beam. Since 0.1 mJy is the MMA sensitivity in a minute of observation such a source will be present in every MMA continuum observation of a minute's duration made at the full bandwidth. In a hour of integration there will be ten such confusing sources detectable in each beam. This is a very rich field of research, one in which the ease of detection allows us to ask quite subtle scientific questions both cosmological (source clustering) and evolutionary (stellar luminosity and metallicity as function of epoch). It awaits the proper next generation arrays.

TABLE 3. Luminosity of lines detected by COBE from the Milky Way

Species A (/-1m) 1/ (GHz) L/L0

CO(J=1-0) 2601 115 < 5 x 104 CO(J=2-1) 1302 230 7.9 x 104 CO(J=3-2) 867.2 345 1.3 x 105

CO(J=4-3) 650.4 461 1.3 x 105

CIe PI _ 3 Po) 690.1 493 2.0 x 105

CO(J=5-4) 519.8 577 1.0 x 105

CIe P2 _ 3 PI) 370.4 810 3.2 x 105

Nile PI _ 3 Po) 205.3 1460 5.0 x 106

C lIe P3/2 _ 2 PI /2 ) 157.7 1900 5.0 x 107

Ole PO_3 PI) 145.5 2060 6.3 x 105

Nile P2 - 3 PI) 121.9 2460 7.9 x 106

420 ROBERT L. BROWN

l I I I I I ! I I I I I

I

l """1

J ~

~ ~ 0 I

§2 :z J ~ :z I ~ I (jj I 0 I z :E I ::l I .... w I > § I w C I a:

6 I (!) I 0

~ .... I 9: ..,

v I C? c ± 6 I

6- 0 c I u t,,;

8 0 I u 0

Figure 3. The relative luminosity of spectral line emission from the Milky Way as obtained from COBE observations (Wright et al. 1991).

4. Observing the Milky Way at cosmological distances

In the sections above we have looked at observations that we expect to make with future millimeter arrays addressed both to the earliest stages of galaxy formation where massive objects, protogalaxies, are built up through the accumulation of lower mass fragments and observations of the ultraluminous phase of galaxy evolution where starburst luminosity is re-radiated in the infrared by dust. Both of these observations refer to brief phases in the evolution of a galaxy, neither is sustained. Let us finally ask, is it possible for us to observe a normal galaxy such as the Milky Way at cosmological distances? This question is important because our knowledge of galaxies throughout the cosmos begins with our knowledge of the Milky Way. If we could observe this normal galaxy back in time, we would have an important framework onto which we could graft our understanding of galaxies in active or ultraluminous phases. The COBE observations provide us with the information we need to estimate the appearance of the Milky Way were we to observe it at increasing distances. From this information we can select a method to guide our search for it. Table 3 summarizes the luminosities of the spectral lines seen by CO BE from the Milky Way as derived by Wright et al. (1991).


In Fig. 3 we illustrate these luminosities normalized to the luminosity of the CO(2-1) line. Here we can see that the luminosity of the CO(3-2) and CO( 4-3) lines are greater than that of the lower lines corresponding to the fact that the temperature of the ensemble of molecular clouds in the Milky Way is greater than the 5 - 10 K excitation temperature of the lowest CO lines. However, as can be seen in Fig. 3, the luminosity of the C I lines is greater than any of the CO lines, and the fine-structure lines of NIl, elI, and 0 I are greater still by one or more orders of magnitude. When observing any of these lines the observable quantity is not luminosity but flux density or equivalent width which is the integrated line flux density expressed as the line integral in Jy km s-l. For all the lines of transition x in Table 3 the line integral is expressed as

5' - 5 A _ 0.01 (Lx/106 L8 ) 2 4Q-2 - xL.l.V - Hoqo

vrest/(l + z) (4)

where Q = qoz + (qo - 1)[(1 + 2qOZ)I/2 - 1] and 5 is in Jykms-I, Vrest is in GHz, Ho is in km s-1 Mpc-1, Sx is in Jy, and v in km s-l. Assuming v = 250 km s-1 we can compute the flux density in the line from the COBE line luminosities and the equation above. The results are given in Table 4.

TABLE 4. Flux density of spectral lines from the Milky Way

Transition Line flux density (mJy)

z = 0.5 z=l z=2 z = 3 z=4 z = 5

eO(J=1-0) 1.9 0.6 0.2 0.1 0.06 0.05 eO(J=2-1) 1.5 0.5 0.2 0.08 0.05 0.04 eO(J=3-2) 1.6 0.5 0.2 0.08 0.05 0.04 eO(J=4-3) 1.2 0.4 0.1 0.06 0.04 0.03 eIe P l _ 3 Po) 1.8 0.5 0.2 0.09 0.06 0.04 eO(J=5-4) 0.8 0.2 0.1 0.04 0.03 0.02 eIe P2 _ 3 PI) 1.7 0.5 0.2 0.09 0.06 0.04 N lIe PI - 3 Po) 15.2 4.5 1.4 0.8 0.05 0.4 e II(2 P3/ 2 _ 2 PI / 2 ) 34.6 11 5.9 3.9 2.8 Ole PO_3 P1 ) 0.4 0.2 0.07 0.05 0.03 N lIe P2 - 3 PI ) 1.4 0.7 0.5 0.3

For the purpose of Table 4 we have computed the line flux density only for those lines that are potentially detectable from the Earth. The Earth's atmosphere becomes essentially opaque beyond 1000 GHz, even from very good high elevation sites such as those now being studied for the future millimeter arrays. Those which are not red shifted below 1000 GHz are not

422 ROBERT L. BROWN

tabulated in Table 4. It is important to emphasize that in Table 4 the flux densities given are the line flux densities that would be observed at the red shift frequency. Therefore if we examine the columns corresponding to observations of the Milky Way at redshift of z = 1 or 2, we see that the millimeter/submillimeter array that we need to observe these lines must either have the sensitivity to detect sub-milliJansky spectral lines at millimeter wavelengths or it should have the sensitivity to detect lines of milliJansky flux density at the shortest sub millimeter wavelengths that can be seen from the ground. That is, either one chooses to study the weak red shifted lower CO transitions at millimeter wavelengths or one concentrates on the stronger redshifted finestructure lines of C II and N II at sub millimeter wavelengths. The next generation arrays have chosen complementary approaches to this problem: the European Large Southern Array (LSA) will build great sensitivity at millimeter wavelengths sacrificing the submillimeter performance whereas the u.S. Millimeter Array (MMA) and the Japanese Large Millimeter and Submillimeter Array (LMSA) are emphasizing good submillimeter performance throughout the available atmospheric windows rather than maximizing the millimeter-wavelength sensitivity. Together these instruments will provide scientists with the tools needed to image the process by which galaxies formed from cold gas in the early universe.

References

Blain, A.W. and Longair, M.S. 1993, MNRAS 264, 509. Blain, A.W. and Longair, M.S. 1996, MNRAS (to be published). Brown, R.L. 1996, MMA Memo. Condon, J.J. 1992, ARAA 30,575. Cowie, L.L., Songalia, A. in First Light in the Universe: Stars or QSOs? ed. B. Rocca

Volmerange et al. (Gif-sur-Yvette: Editors Frontieres), 77. De Propris, R., Pritchet, C.J., Hartwick, F.D.A., and Hickson, P. 1993, AJ 105, 1243. Djorgovski, S. and Thompson D.J. 1992 in The Stellar Populations of Galaxies, ed. S.

Barbury and A. Renzini (Dordrecht: Kluwer), 337. Franceschini, A., Toffolatti, L., Mazzei, P., Danese, L., and De Zotti, G. 1991, A&AS 89,

285. Frayer, D.T., Brown, L.B. and Van den Bout, P.A., 1994, ApjL, 433, 5. Helou, G., Soifer, B.T., and Rowan-Robinson, M. 1985, ApJ 298, L7. Hu, E.M. and Ridgeway, S.E. 1994, AJ 107, 1303. Thompson, D.J., Djorgovski, S., and Trauger, J. 1993, in the Evolution of Galaxies and

Their Environment, ed. D. Hollenbach, H. Thronson, and J.M. Shull (NASA Conf. Pub. 3190), 23.

Tytler, D. 1987, ApJ 321, 49. Wolfe, A.M. 1993, in First Light in the Universe: Stars of QSOs? ed. B. Rocca-Volmerange

et al. (Gif-sur-Yvette: Editions Frontiers), 77. Wright, E.L. et al. 1991, ApJ 381, 200.

STRATEGIES FOR GALAXY SURVEYS IN THE SUBMILLIMETRE WAVEBAND

A.W. BLAIN

Cavendish Laboratory M adingiey Rd. Cambridge CB30HE UK

1. Introduction

A large fraction of the bolometric luminosity of nearby star-forming galaxies is emitted by dust grains in the far-infrared waveband, and so the equivalent emission of distant sources is expected to be redshifted into the submillimetre waveband. There are excellent prospects for detecting this continuum radiation from galaxies at large redshifts (e.g. Dunlop et ai., 1994; Isaak et ai., 1994), because the modified blackbody emission spectra of dust grains produces large negative K-corrections which lead to flat flux densityredshift relations at red shifts between about 1 and 10 in the submillimetre waveband (Franceschini et ai., 1991; Blain & Longair, 1993a). Sources at large redshifts can therefore be detected using an instrument which is sufficiently sensitive to detect sources at redshifts z ~ 1. However, observing faint extragalactic sources in the submillimetre waveband is difficult using curren instruments, and mapping blank fields to faint flux density limits is not feasible at present.

The advent of array bolometer receiver systems, for example the Submm Common-User Bolometer Array (SCUBA) for the JCMT (Gear & Cunningham, 1990; Cunningham et al., 1994), will provide the required sensitivity, and will also increase mapping speeds sufficiently to make a blank-field search for distant star-forming galaxies a practical possibility for the first time. The best strategy for such a survey can be determined by combining models of the distribution of sources on the sky with practical details of observing in the submillimetre waveband.

423

M. N. Bremer etal. (eds.). Cold Gas at High Redshijt. 423-427. © 1996 Kluwer Academic Publishers.

424 A.W. BLAIN

2. Observations in the submillimetre waveband

The speed of survey observations in the submillimetre waveband is limited by the time required to integrate down the noise from the instrument used and from the atmosphere. Atmospheric emission limits ground-based observations to several narrow wavelength windows, within which careful subtraction ofthe atmospheric signal is still essential. Careful calibration is also required to correct the observations for the effects of atmospheric attenuation. At 850 f..Lm, the wavelength at which the effects ofthe atmosphere are least significant at wavelengths A < 1 mm, SCUBA has an hexagonal array of 37 bolometers, which confers three significant advantages over the current single-element bolometer system UKT14 on the JCMT:

1. The area of the sky which is observed in a single telescope pointing is increased by a factor equal to the number of array elements.

2. SCUBA's bolometers are cooled to rv 0.1 K, improving the sensitivity by a factor of order 10 to about 10 mJy Hz-1/ 2 at 850 f..Lm.

3. SCUBA will allow much more reliable and accurate calibration and reduce the importance of favourable atmospheric conditions for sensitive observations.

These factors combine to predict an increase in mapping speed of between 3 and 4 orders of magnitude, enough to allow a blank-field survey in a practical integration time.

3. Models of the submillimetre sky

The derivation of source counts in the submillimetre waveband was discussed by Blain & Longair (1993a). The form of faint counts is modified from the Euclidean slope expected for bright sources by the combined effects of redshifting the spectral energy distributions of sources, of evolution and of the large-scale curvature of spacetime. In the optical and near-infrared wavebands the faint counts flatten out and fall below the extrapolated Euclidean count; however, in the submillimetre waveband the large negative K-corrections cause the faint counts to "invert", and rise above the extrapolated Euclidean slope. Inverted counts are the key to the feasibility and potential utility of survey observations in the submillimetre waveband for observing very distant dusty galaxies: a large fraction of the sources in the inverted region is expected to be at moderate and large redshifts.

Fig. 1 demonstrates the effect for two models of galaxy formation which predict identical densities of bright sources: in one, a population of IRAS galaxies undergoes luminosity evolution of the form (1 + z) 1.2 to a limiting redshift Zo = 5; in the other, a simple model of hierarchical structure formation is adopted (Blain & Longair, 1993b), in which the comoving vol-

SUBMILLIMETRE SURVEYS

Hierarchical (solid)

:-...

" " " IRAS (dash) , Euclidean , , , , ,

0.1

Limiting Flux Density I Jy

425

6

4

2

o

-2

-4

FIgure 1. Counts at 850 p.m in two models of galaxy formation and a Euclidean count.

ume emissivity of sources increases as (1 + z)1.5 to Zo = 10. Both models are consistent with the limits to the intensity of extragalactic background radiation in the millimetre waveband determined by FIRAS on the COBE satellite (Mather et al., 1994). Details of the models and the constraints imposed by observations are discussed by Blain & Longair (1995).

4. The number of detected sources

The counts in Fig. 1 can be combined with the specifications of SCUBA to estimate the number of detections expected in a blank-field survey. The limiting flux density in a survey covering an area A in an integration time t is Smin ex: J Aft if the sky and instrument noise are gaussian. The product of the source count at Smin and A then gives an estimate of the number of detected sources. For example, a 30' sensitivity limit of Smin = 1.6 mJy would be expected in a survey at 850 J.Lm lasting 24 hr and covering 0.1 deg2 ,

corresponding to about 15 detections in the IRAS-based model, showing that a successful survey can be realised.

The choice of A and t is limited by several factors. Most important, t cannot reasonably exceed about 3 X 105 s because of pressure for telescope time. A lower limit to A of about 4 X 10-3 deg2 is imposed by the area subtended by the SCUBA array when chopping between three positions, and a weaker upper limit of about 1 deg2 is imposed in a single field by the need to avoid bright point sources and strong cirrus emission. Uniform calibration of the image is also easier over a smaller area. The survey depth

426 A.W. BLAIN

is also limited by confusion effects. The limits are set both by the surface density of extragalactic point sources and by structure in the galactic cirrus emission, and at 850 flm are expected to be important at flux densities of about 0.1 mJy and 0.2 mJy respectively (Franceschini et at., 1989; Gautier et at., 1992). These are not particularly severe limits; however, if the strong evolution of IRAS galaxies observed at low redshifts continues at higher redshifts (Rowan-Robinson, 1995) then the point source confusion limit will be a real concern.

Contours describing the number of 30" detections expected in a survey as a function of A and t are shown in Fig. 2 for both the IRAS-based and hierarchical models of galaxy formation. The dramatic difference between the form of the contours in each model is entirely due to the differences in the corresponding faint counts. This demonstrates the key point that the results of a SCUBA survey are very sensitive to the shape of the faint counts in the submillimetre waveband: therefore if SCUBA can determine the shape of the crucial inverted regions of the counts, then the population and evolution of far-infrared luminous galaxies can be investigated at large redshifts. In order to exploit this capability, the probability of detecting useful numbers of sources in a submillimetre waveband survey must be maximised by a careful choice of the observing parameters. The optimal observing strategy can be determined from Fig. 2, and depends on the underlying form of galaxy evolution. For a fixed integration time t = 3 X

105 s, and if the modestly evolving IRAS picture is correct, then surveying an area of 0.1-1 deg2 would result in the detection of between 20 and 50 sources with flux densities greater than 3 mJy. On the other hand, if an hierarchical picture of galaxy formation is correct, a survey with these parameters would be unlikely to detect any sources. In this case, a smaller area of, say, 0.01 deg2 would be required for the same integration time in order to reach a lower flux density limit of about 0.3 mJy in order to detect about 20 sources in the extremely steep region of the hierarchical counts.

5. Conclusion

A blank-field survey to determine the faint source counts in the submillimetre waveband, complemented by statistical analyses to the confusion limit of the survey, would provide a direct probe of star-formation activity through the redshift interval 1 ~ z ~ 10, and should readily discriminate between different models of galaxy formation. In deciding the best strategy for such a survey the counter-intuitive effects of the inverted counts expected in the sub millimetre waveband must be taken into account, and the optimal strategy depends on the shape of the counts of the source population which is the object of investigation. In the case of hierarchical clustering models, the

SUBMILLIMETRE SURVEYS

0.1 mJy...-, ,. , .' , .' v' .........

.::.. -----~

."--- -.-

0.01 0.1

Survey Area / deg2

1

427

mJy

100

10

10

Figure 2. A contour plot of the numbers of sources expected in a SCUBA survey at 850 /lm for both the IRAS model (dashed lines) and the hierarchical model (solid lines). Three labelled contour levels are presented for each model. The two dotted lines labelled in mly show the depth attained by a survey. Regions of parameter space which are inaccessible because of source confusion and minimum survey area are lightly shaded.

depth of a survey is probably the critical factor which determines whether or not a survey will be successful.

References

Blain, A.W. and Longair, M.S. (1995), MNRAS, submitted Blain, A.W. and Longair, M.S. (1993a), MNRAS, 264, p. 509 Blain, A.W. and Longair, M.S. (1993b), MNRAS, 265, p. L21 Cunningham, C.R., Gear, W.K, Duncan, W.D., Hastings, P.R. and Holland, W.S. (1994),

Instrumentation in Astronomy VIII. Proc. SPIE, 2198, p. 638 Dunlop, 1.S., Hughes, D.H., Rawlings, S., Eales, S.A. and Ward, M.l. (1994), Nature,

370, p. 347 Franceschini, A., Toffolatti, L., Mazzei, P., Danese, L. and de Zotti, G. (1991), At9AS,

89, p. 285 Franceschini, A., Toffolatti, L., Danese, L. and de Zotti, G. (1989), Apl, 344, p. 35 Gautier, T.N., Boulanger, F., Perault M. and Puget, l.L. (1992), Al, 103, p. 1313 Gear, W.K and Cunningham, C.R. (1990), in From ground-based to space-borne sub-

millimetre astronomy ed. Kaldeich, B. ESA 314, p. 291, ESA publications Isaak, KG., McMahon, R.G., Hills, R.E. and Withington, S. (1994), MNRAS, 269, p. L28 Mather, 1.C. et al. (1994), Apl, 420, p. 439 Rowan-Robinson, M. (1995), this volume

CONSIDERATIONS FOR DETECTING CO IN HIGH REDSHIFT GALAXIES

FRANK P. ISRAEL AND PAUL P. VAN DER WERF Leiden Observatory P.O. Box 9513 NL-2300 RA Leiden The Netherlands

1. Introduction

The detection of CO in the z=2.286 galaxy IRAS F10214+4724 (Brown & Vanden Bout 1991; Solomon et al. 1992b) marked the start of intensive searches for molecular gas in other high redshift objects (e.g., Wiklind & Combes 1994; Evans et al. 1996; Van Ojik et al. 1996; Ivison et al. 1996), but the only other confirmed detection is in the gravitationally lensed "Cloverleaf quasar" H1413+117 (Barvainis et al. 1994). With the subsequent discovery that IRAS F10214+4724 is lensed as well (Serjeant et al. 1995; Broadhurst & Lehar 1995; Eisenhardt et al. 1996), the implied molecular gas mass was revised downwards by a factor of about 20. Even after accounting for gravitational lensing the molecular gas content of these two objects exceeds that of the most gas-rich objects in the local universe which shows that galaxies with large quantities of dense, enriched molecular gas exist at high redshifts. With current instrumentation, CO in such objects can only be detected if amplified by a fortuitous foreground gravitational lens. However, future facilities, planned or under construction, will be able to detect the presumably much vaster population of unlensed objects.

In this paper we examine the possibilities and optimum observing strategies for detecting CO in distant galaxies of various types with such future large millimeter-wave facilities. The CO molecule has the advantage of having a large number of rotational lines relatively closely spaced in frequency. As a consequence, a particular (sub )millimetre wavelength band contains at least one CO transition regardless of the redshift of the emitting galaxy. Determining which transitions are likely to yield a detection is a nontriv-

429


430 FRANK P. ISRAEL AND PAUL P. VAN DER WERF

ial exercise involving the (expected) intrinsic line ratios of the target, in addition to the parameters of the instrument being used and the atmospheric properties of its site. These considerations will be different for different redshifts, since a particular line will be redshifted to spectral regions with widely varying atmospheric properties, depending on the precise redshift. In the following we first examine line ratios expected for high-redshift galaxies of various types. As a specific application, we then discuss the observability of these objects as a function of red shift with the planned NRAO Millimeter Array (MMA). Throughout this paper we adopt a value Ho = 75kms-1 Mpc-1 for the Hubble constant.

2. CO line ratios in low redshift galaxies

In order to compare the strengths of different transitions of CO, it is convenient to express line luminosities on the L' scale, which can be easily related to observable quantities, as follows

(1)

Here Tb is the intrinsic Rayleigh-Jeans brightness temperature in the line, ns is the source solid angle and DAis the angular size distance. In Table 1 we present a compilation of L' luminosities of the lowest CO lines detectable from the ground for the inner regions of a variety of nearby galaxies and high-redshift galaxies. With the exception of the Milky Way ratios, which were derived from COBE satellite data, the entries in Table 1 are based on single-dish telescope data; observed main beam brightness temperature ratios have been corrected for beam size differences in order to derive ratios of L' luminosities according to Eq. (1). This equation shows that lines with equal intrinsic brightness temperatures Tb have equal L' luminosities, as long as source dimensions and linewidth do not change between transitions. Since Tb is constant for opaque, thermalized lines (often true for the lower CO transitions), L' will also be constant under these conditions. Note that the line luminosities Leo in the usual units of erg s-1 still do increase in these circumstances, as can be seen from the equation relating Land L' luminosities:

c3 L~o = Leo-g k 3' (2)

7r Vo

where Vo is the rest frequency of the transition. The values listed in Table 1 are easily understood. In starburst nuclei

the average temperature is 50 K or higher (Harris et al. 1991; Wall et al. 1991; Israel et al. 1995) so that levels up to J = 6 (116 K above the ground state) are still relatively easily excited. Due to a fairly high average density, the levels stay thermalized up to about J = 4. In quiescent spiral galaxies

DETECTING HIGH REDSHIFT CO 431

TABLE 1. CO intrinsic line luminosities L' (see text) in nearby galaxies, normalized to CO J = 1-+0, except for the high-redshift objects, where values have been normalized to CO J = 3-+2.

Galaxy 1-+0 2-+1 3-+2 4-+3 5-+4 6-+5 7-+6 Reference

Normal galaxies: Milky Way 1 0.4 0.3 0.2 0.07 0.01 NGC7331 1 0.5 0.13 2

Local CO-rich galactic nuclei: IC342 1 0.9 0.8 0.5 0.4: 3, 4 NGC3628 1 1.1 1.0 0.5 3 NGC6946 1 1.3 1.0 0.8 3

Local starburst nuclei: NGC253 1.0 0.9 0.6 0.4 3,4 NGC4826 1 0.9 1.0 1.0 3

Locallow-metallicity starburst regions: He 2-10 1 1.1 1.4 5 Hubble V 1.2 2.0 3.0 6

High-redshift galaxies: IRAS Fl0214+4724 1 0.8 0.6 7 H1413+117 1 1.1 0.6 0.3 8

References: (1) Bennett et al. (1994); (2) Israel & Baas (1996); (3) Israel & Baas (in preparation); (4) Harris et al. (1991); (5) Baas et al. (1994); (6) Baas & Israel (in preparation); (7) Solomon et al. (1992a); (8) Barvainis (1996).

such as the Milky Way however, where average densities are lower, the gas becomes subthermally excited already at much lower rotational levels. Spirals rich in molecular gas are intermediate between these two cases. Interestingly, the starburst region Hubble V in the nearby low-metallicity magellanic irregular NGC 6822 must be both hot and dense enough to yield CO transition luminosities increasing with level up to at least J = 4.

3. Detectability of CO lines from high-redshift galaxies

In order to estimate the observability of similar objects at high redshift, we assume the same intrinsic line ratios as in corresponding low-redshift galaxies. For Milky Way type objects, where even the low-J levels are subthermally excited, additional radiative excitation by the cosmic microwave background has to be taken into account (e.g., Solomon et al. 1992a). Radiative transfer calculations based on a Large Velocity Gradient (LVG) model


show that for low densities a cosmic microwave background TCMB rv 10 K (valid for z rv 2) significantly enhances the populations of the J = 2 - 4 levels over the low-redshift (TCMB = 2.7 K) values, with a corresponding increase in L' luminosity in these lines. An exciting prospect is that lowmetallicity starbursts may also have significantly increased detect abilities with increasing J level. Low metallicities must occur in the initial bursts of star formation in high redshift objects, so that line ratios similar to those in the low-metallicity starburst region Hubble V may be expected.

For point source detection, we express the expected signal in terms of the flux density integrated over the line:

Sco~v -8 Leo (va) 2 ( DL )-2 Jy km S-l = 3.1 X 10 K km S-l pc2 GHz Mpc (1 + z), (3)

where DL = (1 + z)2DA is the luminosity distance and Va is the rest frequency of the line.

TABLE 2. Adopted luminosities of template galaxies

Galaxy LJR L~o 1-0 Leo 1-0

[L8] [K km S-l pc2 ] [L8]

Milky Way 1.0 X 1010 3.7 X 108 1.8 X 104

M51 3.8 x 1010 1.5 X 109 7.5 X 104

Arp 220 1.5 x 1012 4.6 X 109 2.3 X 105

Based on this equation and the line ratios described above, we now calculate the expected signal for starburst galaxies, CO-rich galaxies, and Milky Way type galaxies. To fix the absolute line strengths, we use the prototypical ultraluminous infrared galaxy Arp 220 as a template starburst galaxy, and M51 as a template CO-rich spiral. Predictions for fainter or brighter objects can be derived by scaling with respect to one of the luminosities specified in Table 2. Note that normal galaxies such as NGC 7331 will have both lower luminosities and line ratios dropping more rapidly than the Milky Way.

The expected linestrengths as a function of redshift are plotted in Fig. 1, for the 3 mm and 1.3 mm atmospheric windows, and for qa = o.s and 0.1. The benefit of having a number of lines available is seen most clearly in the top lefthand panel of Fig. 1, which shows the remarkable result that a starburst galaxy detectable at z,-..., 1 can in the same integration time be detected out to ZrvS, and, in principle, with only a few times more integration time out to much higher redshifts. More speculatively, if the early,

~


10rrTT~~~7r5r-T1T15~G~H~z~'r~r=TO~.5 I I I I

1 ~

:~~~~ 0.1 \ " ":"

·1 .. 0' .. : ....... . , '" 0.01 2 .. 1 ""'- 4 .. 3

3 .. 2 ......

IOrrTT~~r7~5-rrlIT5~G~H~z·~~~=rO~.1 I I I I

I ~

o}~ ~~ "----" .......... .

0.01 " " .. ....

rn8 0.001 ......

0.001 .... .... ...... ............ __ o .. e; -- .........

0.0001 ..... M51 0.0001 .................. - Arp 220 ---- - Milky Way

I I I I 10-·0!:-'-........ -:!::2~'-'-4~ ........ -:!::8 ......... ....L...:8!::-'-~ 10-.0 I I I I

10 2 4 8 8 10

200-280 GH •• ~-0.5 200-280 GHz. ~-O.I

10 ~'~-.... \ .... ' ~

\2:1 .. :: ........... . \ , ,~

3 .. 2' '" 4 .. 3

0.1

..... ..... ... -~ ... 0.01 .... ...... ... .... --- 8 .. 6

-- ... 0.001 0.001 -.

1.5 2 2.50.00010 0.5 1.5 2 2.5 z z

Figure 1. Integrated strengths of CO lines as a function of red shift for various types of galaxies, in the 3 mm window (top panels) and the 1.3 mm window (bottom panels) for qo = 0.5 (left panels) and qo = 0.1 (right panels)


low-metallicity starbursts in galaxies at high redshifts reached luminosities similar to that of Arp 220 now, they would be easier to detect than conventional starburst galaxies. In addition, they would be easier to detect with increasing CO J level, hence with increasing redshift, as long as the corresponding J levels remain thermalized! For Milky-Way-type galaxies the situation is somewhat less favourable because significant flux is lost by going to lines higher than J = 4-3.

These results can easily be understood by noting that as long as the levels involved are thermalized, the luminosity (in erg s-1) that is to be detected increases as J3, where J is the upper level of the transition. This effect partly compensates for the increased distance, until such high lines are reached that the upper level is strongly subthermally excited.

3.1. DETECTION LIMITS WITH THE MMA

We place our point source detection limit at the 50' level for a velocity resolution of 30 km s-1. This resolution is sufficient to obtain detailed line profiles, allowing applications such as CO Tully-Fisher studies (see below). With current sensitivity specifications, this detection limit corresponds to line integrals of about 0.05 Jy km s-1 in the 3 mm and 1.3 mm bands respectively for a 300 km S-1 wide line in an 8 hour synthesis with the planned MMA. For imaging, sensitivity limits will depend on the details of the brightness temperature distribution on the sky (and of the positions of the antennas in the array), but a crude estimate is given by dividing the total flux into ten independent beams, so that the detection limit for imaging is about a factor of ten higher than that for point source detection at the same velocity resolution.

For detection purposes, the source should not be resolved by the array. This poses limits on the maximum baseline that should be used. If the CO emission is essentially contained within the inner 2 kpc, which is a somewhat conservative upper limit for starburst galaxies, the source angular size will reach a minimum of about 0~'3 at z=2.5 for qo=O.1 and then very slowly increase with z. For qo=0.5 a minimum of 0~'4 is reached at z=1.5, after which the angular size increases to 0~'8 at z=10. An angular resolution of 0~'4 corresponds to maximum array baselines of 1400 m and 800 mat 3 mm and 1.3 mm respectively. If the CO emission is more broadly distributed within a diameter of 10 kpc, as in the Milky Way, angular sizes range from 3" at z=0.25 to slightly less than 2" at Z= 1. Maximum array baselines corresponding to 3" are 190 m and 110 mat 3 mm and 1.3 mm respectively.

Can em-wavelength arrays such as the VLA playa similar role? To this end we have performed calculations similar to those illustrated in Fig. 1. In the 22 - 24 GHz band, the VLA could in principle find CO J = 1-0


emission from galaxies around z=4. However, the current VLA sensitivity falls short by an order of magnitude to detect even luminous starburst galaxies in an 8-hour synthesis. A similar result is obtained for the VLA in the 40 - 48 GHz band which covers redshift ranges z=1.4 - 1.9 (J = 1-+0) and 3.8-4.8 (J = 2-+1). Until a major upgrade of the VLA high-frequency capabilities has occurred, it is unlikely to playa role in this field.

Returning to mm arrays, inspection of Fig. 1 now reveals the possibilities and optimum observing strategies with the MMA, of which we summarize the most important points below.

For starburst galaxies at z < 2.5 the highest SIN ratio is obtained in the 1.3 mm band, where M51-type galaxies or starburst galaxies 10 times fainter than Arp 220 can be imaged out to z"'-' 1, while Arp 220-like objects can be imaged out to at least z"'-'2.5. Starburst galaxies fainter than Arp 220 by factors 10 to 50 (depending on the value of qo) will be detectable as point sources out to z"'-'2.5 in the 1.3 mm band. For z>2.5 the 1.3 mm band is less favourable, since only fairly high rotational lines, which may not be efficiently excited, are available. Therefore, at z>2.5 the 3 mm band is the better choice, and starburst galaxies with Arp 220 luminosities will be detectable throughout the z<10 universe in this band for qo=0.5 (out to z",-,6 for qo=O.l). If our speculation that early low-metallicity starbursts yield line strengths sharply increasing with J is correct, the 3 mm band is still preferred, because this effect will only playa role at z>5. In the 1.3 mm band only transitions with upper level J > 12 correspond to these redshifts, and is doubtful whether such high levels are still thermalized. For Milky-Way-type galaxies it is necessary to observe low-J lines, and as a result the 3 mm band is favoured over the 1.3 mm band. Point-source detection of Milky-Way-type galaxies is possible in the 3 mm band out to z"'-'l for qo=0.5. Out to z"'-'l, fairly low lines are still available in the 1.3 mm band as well, and for these redshifts the 3 mm and 1.3 mm bands yield about the same SIN ratio. Imaging is restricted to about z<0.25, for both the 3 mm and the 1.3 mm band. Given a choice of transition, for any red shift the higher transition should be the more easily detectable up to J =6 for starburst galaxies and up to J =4 for Milky-Way-type galaxies. Depending on z and qo, attempts to detect starburst galaxies should be carried out with maximum baselines not exceeding 1400 m at 3 mm, and Milky-Way-type galaxies with baselines not exceeding 200 m. In searching for low-metallicity starbursts at high z it is probably wise to consider maximum baselines of 700 m or less. In shorter-wavelength bands, the maximum allowable baseline is correspondingly shorter, emphasizing the need for relatively compact configurations.


Since the initial starburst in galaxies forming the bulk of their stellar population may well resemble Arp 220 and similar objects in luminosity and observable properties (e.g., Scoville et al. 1996; Van der Werf 1996), these considerations show the immense potential of future large millimetre wave arrays for observing early stages of galactic evolution. As another application, we point out the possibility offered by the MMA to extend Tully-Fisher studies to much larger distances than is possible with atomic hydrogen. CO is less sensitive than H I to large-scale distortions in the outer disks, and corresponds more closely to the luminous matter in galaxies. With the MMA the CO Tully-Fisher relation can therefore be used to measure accurately perturbations to the general Hubble flow.

Acknowledgements. Most of the observations used to derive the line ratios in Table 1 were obtained with the JCMT, operated by the Observatories on behalf of the Particle Physics and Astrophysichs Council of the UK, the Netherlands Organization for Scientific Research and the National Research Council of Canada. The research of Van der Werf has been made possible by a fellowship of the Royal Netherlands Academy of Arts and Sciences.

References

Baas, F., Israel, F.P., & Koornneef, J. 1994, A&A, 284, 403 Barvainis, R. 1996, these proceedings Barvainis, R., Tacconi, L., Antonucci, R., Alloin, D., & Coleman, P. 1994, Nat, 371, 586 Bennett, C.L., et al. 1994, ApJ, 434, 587 Broadhurst, T., & Lehar, J. 1995, ApJ, 450, L41 Brown, R.L., & Vanden Bout, P.A. 1991, AJ, 102, 1956 Eisenhardt, P.R., Armus, L., Hogg, D.W., Soifer, B.T., Neugebauer, G., & Werner, M.W.,

1996, ApJ, in press Evans, A.S., Sanders, D.B., Mazzarella, J.M., Solomon, P.M., Downes, D., Kramer, C.,

& Radford, S.J.E., 1996, ApJ, in press Harris, A.I., Hills, R.E., Stutzki, J., Graf, U.U., Russell, A.P.G., & Genzel, R. 1991, ApJ,

382, L75 Israel, F.P., & Baas, F. 1996, to be submitted to A&A Israel, F.P., White, G.J., & Baas, F. 1995, A&A, 302, 343 Ivison, R.J., Papadopulos, P., Seaquist, E.R., & Eales, S.A. 1996, these proceedings Scoville, N.Z., Yun, M.S., & Bryant, P.M. 1996, these proceedings Serjeant, S., Lacy, M., Rawlings, S., King, L.J., & Clements, D.L. 1995, MNRAS, 276,

L31 Solomon, P.M., Downes, D., & Radford, S.J.E. 1992a, ApJ, 398, L29 Solomon, P.M., Radford, S.J.E., & Downes, D. 1992b, Nat, 356, 318 Van der Werf, P.P. 1996, these proceedings Van Ojik, R., et aI., 1996, submitted to A&A Wall, W.F., Jaffe, D.T., Israel, F.P., & Bash, F.N. 1991, ApJ, 380, 384 Wiklind, T., & Combes, F. 1994, A&A, 288, L41

FUTURE POSSIBILITIES FOR DETECTING HI AT HIGH REDSHIFT

ROBERT BRAUN Netherlands Foundation for Research in Astronomy

1. Introduction

The 21-cm line of neutral hydrogen is in many respects the most valuable tracer of neutral gaseous mass in astrophysics. Even though neutral gas becomes predominantly molecular at high densities, the 21-cm line emission of the associated atomic component allows the total gaseous mass of (proto) galactic concentrations to be estimated to about a factor of two, even in the most extreme cases observed to date. This is in marked contrast to, for example, the luminosity of carbon monoxide emission lines originating in the molecular component. For the same total gaseous mass, these emission lines are observed to vary in luminosity over a factor of about 104 depending on the abundance of heavy elements and the intensity of illumination to which they are subjected.

It is also important to draw the distinction between the sensitivity to the neutral gaseous mass of distinct concentrations with the sensitivity to column density along discrete lines-of-sight. The extremely large absorption cross-section of neutral hydrogen to Lyman-a radiation guarantees a correspondingly high sensitivity to neutral hydrogen columns along linesof-sight to suitable background sources of (red-shifted) ultra-violet light. However, the paucity and compact nature of such background probes implies that gas masses can never be determined for individual objects via such observations of absorption. An emission line tracer of the neutral gas remains essential for this purpose.

In this paper we will briefly review the beginnings of HI astronomy, proceed to an assessment of our current capabilities in this area, and continue by considering what will be necessary to push back the frontier to cosmological distances. We will then consider how such a leap in performance might be realized.

437


438 ROBERT BRAUN

2. Where have we come from?

After the prediction of the existence of the HI 21-cm line by Van de Hulst (1945) it was only a few years before several independent groups detected the line emission from our Galaxy and reported it simultaneously in a 1951 issue of Nature (Ewen and Purcell, Muller and Oort). Even in the preliminary discovery report, important contributions were made by Muller and Oort (1951) to our knowledge of the kinematics and morphology of the gaseous disk of our Galaxy. This was followed, in short order, with major surveys of spiral structure and kinematics of the Galaxy (Van de Hulst, Muller and Oort, 1954)

3. Where are we now?

The years since 1951 have seen a major improvement in our abilities to observe HI in galaxies. An illustration of this improvement is given in Fig. 1 where the continuum sensitivity is plotted after one minute of integration for many radio telescopes as they became available. Since upgraded receiver systems have often been added to existing facilities after construction, these are also indicated in the figure in a number of cases. An exponential improvement in sensitivity over at least 6 orders of magnitude is apparent between about 1940 and 1980. Instruments like the WSRT and the VLA have become available on a schedule which maintained a high rate of discovery. The Arecibo telescope stands out as a major leap in sensitivity performance at a relatively early date.

Sensitivity within the narrow bandwidth of a spectral line observation has evolved in a very similar way to that illustrated in Fig. 1. We have now progressed to the point where we can study the neutral gas content, distribution and kinematics within individual galaxies at recession velocities as high as about 24,000 km s-1. This is perhaps best illustrated by a recent deep integration (about 100 hours) with the VLA obtained by Van Gorkom (1995) to image HI in galaxies of the cD cluster A2670. Some 20 cluster galaxies are detected within an area of about 0.25 square degrees with an rms sensitivity of 80 I-lJy beam -1 per 48 km s-1 channel. The corresponding 50' detection limit for a galaxy spanning 100 km s-1 is about 4x 108h-2 MG).

As impressive as this result is, it also underlines the sad fact that current instrumentation only allows us to determine neutral gas masses for galaxian concentrations out to a red-shift of about 0.1. Given all the evidence for substantial evolution of the gas mass at red-shifts between perhaps 0.2 and 0.5, let alone the dramatic evolution expected between z=0.5 and 2, this is particularly frustrating.

-. S 1000 >. ~

8 -~ ... ::. ::l ... IIJ

~ 1 fIl

§ 0.1

=' :§ 0.01 c 80.001

DETECTING HI AT HIGH REDSHIFT 439

• • I • ............................................................... · . . . I I I • · . . . -:Reber: : : .................................................. " ........... . · . . . · . . . · . . . • I • I · . . . · . . , .... : ........... 'D$rliieTci6' ... ':' ............ :' .......... . · . . . · . . . .. .

····i········· .. ···~odreU .. ····~··········· .. ~······· .. ··· : :eParka: :

.... ; .............. ~ .. I. ,-.BODD •. ~ .............. ~ ............. ..

: : ~ ..... "-.... : .... j ............ ~4m ....... . .... : ............. ~ .............. ~.~ .. .

I I • I .. . " . .. . .. .

•••• ~ ••••••••••••• -: ••••• 0 •••••••• t •. I .......... • ~ •••••••••••• · . . · . . · . . · . . · . .

Figure 1. The time evolution of radio telescope sensitivity. The continuum sensitivity after one minute of integration is indicated for a number of radio telescopes as they became available. Solid lines indicate up-grade paths of particular instruments.

4. Where are we headed?

An overview of how current and upcoming instrumentation measures up to the problem of detecting HI emission from distant systems is given in Fig. 2. The continuum and line emission of the luminous spiral galaxy MI0l has simply been re-scaled to simulate its appearance at the indicated redshifts of 0.12,0.25,0.5, 1, 2 and 4. For this illustration, no time evolution of the emission spectrum has been assumed, even though it is clear that the large stellar mass which is now present was once also in the form of gas. The rms sensitivities of a variety of existing and planned instruments (assuming a spectral resolution of 104 and an integration time of 12 hours) have been overlaid on these spectra. The HI emission line of such a gas-rich system (MHI=2 X 1010 Mev) is easily detectable by the VLA and WSRT near z=O.I, but will probably demand the GMRT for detection at z=0.3.

An important point to note is that it is not merely a question of having enough sensitivity to detect the HI emission line, but also that the appropriate frequency coverage be available. For example, the VLA 20-cm band extends from 1320-1700 MHz at 0.9 times nominal sensitivity, reaching only to z=0.08 in HI. The highest frequency band of the GMRT on the other

440

-0 -II

g:: ..; 'r;; I: f .... -.S 10 >.

:3 1

iii I: 0.1 CIl fIl

CIl 0.01 t:: :::3

0.001

ROBERT BRAUN

Messier 1 Q 1 Spectra ...................... -............................................... .. · .

· . ...................................................................... · . · . · . · . · . ....................... ...: ........................ ..: ............... -_ .... . · . · .

. . . . . ~···················· .. r·····················r .. ········ ············r······················ .. .. . .. .. ....................... -: ........................ : ...... - ................. ;- .... __ ..... ----_ ...... . .. .. . .. .. .. .. , .. .. .. .. , .. ..

10' 1010

Frequency (Hz) 1011

Figure 2. Red-shifted unevolved spectra of MI0l compared to the spectral line sensitivity of current and planned facilities at frequencies between 108 and 1012 Hz. The solid lines are the composite emission spectra red-shifted to z=0.12, 0.25, 0.5, 1, 2 and 4. Note that the continuum and spectral line features of the composite spectra are sampled with different effective bandwidths. Telescope names are placed at the rms sensitivity level after one "transit" of integration at a spectral resolution of 104 .

hand is expected to extend from 1000-1420 MHz, so that red-shifts as high as 0.4 may become accessible.

Finally, it is essential that the frequency in question is not rendered unusable by radio frequency interference. This is in fact the reason that only synthesis arrays have been plotted in Fig. 2. Experience has shown that total power instruments, like Arecibo and the Green Bank 140 foot telescope are unable to achieve noise limited performance in those portions of the spectrum which are in active use. For several reasons, synthesis arrays are much less vulnerable to external interference. This is an issue to which we will return below.

Another way of illustrating upcoming performance is given in Fig. 3. In this case the performance of the upgraded WSRT (as expected in 1997) is illustrated for a long integration of 400 hr duration. The limiting HI mass is plotted as function of red-shift for both the case of "detection" and "imaging". "Detection" is defined here as requiring a 5a signal in a single 50 km s-1 velocity channel, while "imaging" is defined as requiring a 5a


signal in each of six independent velocity channels of 50 km s-1 width. The solid curves, which extend from z=O - 0.2 and z=2.6 - 4.7, indicate the frequency ranges where optimized receiver systems will be available. Almost continuous coverage of the remaining interval, z=0.2 - 2.6, will also be available for the first time with the new receiver system, although at a reduced sensitivity.

e 1012

::s N I

-'= • 1011 -

10e~~~~~--~~~~~1~--~~~~~~

Red-shift

Figure 3. Current and future capabilities for detecting and imaging HI as a function of red-shift for an integration of 400 hour duration. "Detection" is defined to imply a 50' signal in a single 50 km S-1 channel, while "Imaging" implies a 50' signal in each of 6 independent channels of 50 km s-1 width.

The preceding discussion has illustrated how inadequate the current generation of instruments will be to study galaxian gas masses out to cosmological distances. This fact has been one of the major drivers for pursuing a next generation facility with about two orders of magnitude greater sensitivity than what is now available. Since the requirement is basically for about 106 m 2 of collecting area, the proposed facility has come to be called the "Square Kilometer Array Interferometer" .

The astute reader will already have noted that the capabilities of such a new instrument have been overlaid on Figs. 1, 2 and 3. From Fig. 1 it is clear that the SKAI sensitivity is what is required to maintain the

442 ROBERT BRAUN

exponential improvement in performance that accompanied the decades of broad ranging discovery between 1950 and 1990. Returning specifically to the detection of the HI emission line, Fig. 2 illustrates how individual galaxies, like M101, would be within the reach of such an instrument out to red-shifts greater than 2. This capability is placed in a more continuous context in Fig. 3, where it can be seen that SKAI will effectively open much of the universe to direct study of (sub- )galaxian neutral gaseous masses and how they have evolved.

While sufficient sensitivity to detect the integrated signal from gaseous concentrations tells part of the story, another important concern is having sufficient spatial resolution to allow kinematic and morphological studies to be undertaken. The combined imaging and detection capabilities of SKAI are illustrated in Fig. 4, assuming that most of the instrumental collecting area is concentrated in a circular region of about 50 km in diameter. The angular resolution in the red-shifted HI line then varies between about an arcsec locally to 2 arcsec by Z= 1. This combination of collecting area and array size has been chosen to provide about 1 Kelvin of brightness sensitivity for spectral imaging applications within a 24 hour integration. An actual HI data-cube of M101 has been resampled and rescaled to simulate it's appearance at the indicated red-shifts of 0.2, 0.45 and 0.9. The peak observed brightnesses are shown in the left hand panels, while the derived velocity fields are shown on the right. From the figure it is clear that fairly detailed kinematic studies (including kinematic detection of spiral arms, rotation curves, etc.) of "normal" systems will be possible to at least z=0.5, while crude kinematics (basic orientation and rotation parameters) will be possible to Z= 1 or more. In addition, it should be borne in mind that the actual field-of-view and spectral bandpass of a SKAI observation will be many times that shown in Fig. 4. While each panel of the figure is only 230 kpc on a side, the likely SKAI field-of-view will correspond to about 1.5, 3 and 5 Mpc at the three red-shifts shown. At the same time a total observing bandwidth of 100-200 MHz will probe a cylindrical volume about 200 M pc deep. Each pointed observation will therefore provide serendipitous kinematic data on several hundred field galaxies.

So far we have assumed that the gas properties of galaxies at earlier epochs are similar to those of current galaxies in making some predictions of what might be achieved. It would be very surprising if the universe were to behave in such a boring fashion. At the current epoch, the vast majority of the gas that is gravitationally bound by individual galaxies has already been cycled through, and to a great extent locked in the form of, stars. If we can reach back to the time when much of the early activity was taking place, which may correspond to the quasar epoch between z=2 - 3, the current proportions may well be reversed.

DETECTING HI AT HIGH REDSHIFT

SKAI """,. 01 M11J1 Oil: II • 0.1 i¥I U hno 10 to 30 40 110

SKAI I~ 01 IlhIJ1 at z .. 0.11 in 100 hno I .. II 1\

1401 ilUI t1.$ IIttl 21.1 2U

aKA! V.u:Id, 01 AlIII1 Oil: z • 11.2 I .... hi'll ·tOO >$0 0 50 100

M 4S "-.4--.l...-.\.-,!.......,--,--,o......Ji.-I

14 01 31 30 a Ie 21 Ie 2$ )14 :13

SKAt V..tfld. of 1.1101 11\ % .. (US In 24 hno 400 .Jq II 110 100

,.

..

SKAl VIIJ'lcf, 01 MtlJ1 .1 ~ .. U In 1(10 hi'll .1oo.so 0 50 100

'4.4i 40

• 30 .tI • . ..

+ H" ZD

111 ~

10 ~

os 1401 lUi 27$ 2U 211.5 2U

443

Figure 4, Simulated SKAI observations of MIOI as it would appear at red-shifts of 0,2, 0.45 and 0,9. Peak observed brightnesses are shown in the left hand panels and corresponding velocity fields on the right. The assumed integration time is indicated above each panel.

A better indication of what we might expect to find in the early universe is beginning to emerge from extensive numerical simulations of structure and galaxy formation (eg. Weinberg 1995, Ingram 1995). In Figs. 5, 6 and 7 we have taken the simulated neutral hydrogen densities predicted by these simulations at red-shifts of 2,3 and 4 (CDM with n = 1, nB =0.05,

444 ROBERT BRAUN

Ho = 50 km s-l, 0"16 = 0.7) in a co-moving volume that is 22.22/(1+z) Mpc on a side and overlaid the five sigma detection contour of SKAI after a long integration time of 1600 hr. Within a single pointing of the SKAI, and each 2.5 MHz of spectral bandwidth we might expect a handful of detections at z=4, perhaps 50 at z=3 and some hundreds at z=2. Rather than being carried out as separate experiments, the instrumental bandwidth and spectral resolution are likely to be sufficient to observe the entire redshift interval 2-4 simultaneously. Based on the detection frequencies noted above, we would then expect such a single experiment to allow study of some 8000 high red-shift systems.

Figure 5. Simulated HI emission at z=2 with SKAI detections overlaid. The linear grey-scale indicates the predicted peak brightness of HI emission in a 22.2/(1 + z) Mpc cube and extends from log(M0/beam) = 1.7 - 10.B. The single white contour at log(M 0/beam) = 9.22 is the 5eT SKAI detection level after a 1600 hour integration.

Of course the actual number and distribution of detections at these redshifts will probably be quite different than illustrated in Figs. 5-7. However, those differences are likely to make it possible to determine the cosmological model that actually applies to our universe.


.. Ii a 10

120030 15 00 11 5945 RIGHT ASCEl'mON (JiOOO)

Figure 6. Simulated HI emission at z=3 with SKAI detections overlaid. The linear grey-scale indicates the predicted peak brightness of HI emission in a 22.2/(1 + z) Mpc cube and extends from log(M8/beam) = 2.6 - 10.7. The single white contour at log(M8/beam) = 9.72 is the 50' SKAI detection level after a 1600 hour integration.

5. How will we get there?

Some of the basic instrumental parameters ofthe SKAI have already emerged from the previous discussion. The highest possible sensitivity (corresponding to a baseline geometric collecting area of 106 m2 ) is required over frequencies from about 200-1400 MHz. This should be coupled with the highest angular resolution that retains sufficient brightness sensitivity for HI emission line detection. In practise this implies an array distributed over a region of 30-50 km diameter. The instantaneous field-of-view should be as large as possible (from scientific considerations) while not limiting system performance on long integrations at these relatively low frequencies. Considering both ionospheric non-isoplanacity as well as sky model complexity suggests a unit telescope size of between 100 and 300 meter diameter. Instantaneous synthesized image quality must be sufficient to allow adequate

446 ROBERT BRAUN

Figure 7. Simulated HI emission at z=4 with SKAI detections overlaid. The linear grey-scale indicates the predicted peak brightness of HI emission in a 22.2/(1 + z) Mpc cube and extends from log(M8/beam) = 3.4 - 10.7. The single white contour at log(M8/beam) = 10.13 is the 500 SKAI detection level after a 1600 hour integration.

modeling of a time variable sky model (including ground and space-based interfering sources), leading to a minimum requirement of about 32 welldistributed units which would be cross-correlated.

The above requirements are embodied in the schematic configuration shown in Fig. 8. A densely packed elliptical zone accounts for some 80% of the array collecting area. The remaining 20% of the collecting area is distributed over a much larger region to permit sub-arcsec resolution to be employed for other applications like imaging in continuum radiation and HI absorption.

Although the basic parameters of the instrument and its schematic configuration can be derived in a straight forward manner, the method of realizing such an enormous collecting area at an affordable price is less clear. Looking back at Fig. 1, there are indications that some leveling out of sensitivity with time has already set in since about 1980. This is almost

DETECTING HI AT HIGH REDSHIFT

•

.. a···· • • • • • • ••••

447

300 knt

Figure 8. Schematic configuration of SKAI. Note that the unit telescopes are not depicted to scale.

certainly the result of having reached limits in the performance to cost ratio of traditional radio telescope technologies. We have now reached the point where system performance is no longer limited by receiver noise, but primarily by the raw collecting area itself. Traditional technologies have not yet made great progress in reducing the cost of raw collecting area by orders of magnitude. The most cost effective designs from this point of view have been the Arecibo fixed spherical reflector and the GMRT low mass paraboloid. How might we proceed to even greater cost-effectiveness for the unit telescopes?

Several possible element concepts for the SKAI are illustrated in Fig. 9. At the heart of each of these concepts is a much greater reliance than ever before on mass produced and highly integrated receiver systems together with much more extensive digital electronics for beam formation. In the top panel we depict one conceivable extreme in a continuous range of possibilities. In this case the wavefront is detected by individual active elements comparable to a wavelength in size. Each of these is amplified, digitized and combined with the others to form an electronically scan-able beam (or beams) with no moving parts whatsoever. The challenge in this case lies

448 ROBERT BRAUN

Figure 9. Possible element concepts for the SKAI.

in achieving extremely low component and data distribution costs since literally millions of active elements will be required. In the center panel, some degree of field concentration is first achieved with the use of small paraboloids before amplification, digitization and beam formation. In this


case, active element number is reduced to some thousands and greater sky coverage at high sensitivity is also realized, although at the expense of the mechanical complexity of the paraboloid drive and tracking system. In the lower panel we depict the other conceptual extreme, whereby a single large reflector is used for each of the unit telescopes. Extensive arrays of active elements would be employed in this concept to intercept the focal region of the spherical primary in order to efficiently illuminate the surface and allow multiple beams to be formed.

The adaptive beam formation technology which underlies all of the element concepts just considered is extremely attractive for a number of reasons. Real-time beam formation with at least thousands if not millions of active elements provides a comparable number of degrees of freedom for tailoring the beam in a desired way. The basic properties of high gain in some direction and low side-lobe levels elsewhere are fairly obvious and traditional requirements. An additional possibility, which hasn't yet been applied in radio astronomy, is that of placing response minima in other desired directions, such as those of interfering sources. In addition, the way is naturally opened to exploit multiple observing beams on the sky to enhance the astronomical power of the instrument many-fold. These might be used to provide simultaneous instrumental calibration, support multiple, fully independent observing programs or enlarge the instantaneous field-of-view for wide-field applications. Finally, the great potential of adaptive beam formation has led to a strong commercial interest in this technology. This has opened the way to collaborative R&D efforts which are now beginning to take shape.

During the interval 1995-2000, a concerted effort at R&D for SKAI will be undertaken both within the NFRA and at collaborating institutes. The various concepts depicted in Fig. 9 (and potentially new ones) will be worked out in sufficient detail to allow realistic cost estimates to be made. Proto-typing of cost effective technologies as an extension to the WSRT array is planned for the period 2001-2005. Assuming the successful completion of both technical preparations and funding arrangements, construction of the instrument is envisioned for the period 2005-2010.

References

Ewen, H.I. & Purcell, E.M. 1951, Nature 168, 356. Ingram, D. 1995, These Pmc. Muller, C.A. & Oort, J.H. 1951, Nature 168, 357. Van de Hulst, H.C. 1945, Ned.Tijd.Nat. 11,210. Van de Hulst, H.C., Muller, C.A. & Oort, J.H. 1954, Bull.Astron.lnst.Neth. 12,211. Van Gorkom, J.R. 1995, These Pmc. Weinberg, D.H. 1995, These Proc.

SPH SIMULATIONS OF THE EARLY UNIVERSE

Performance of the Dwingeloo Square Kilometer Array

D. R. INGRAM AND N. KATZ University of Washington ASTRONOMY Box :351580 Seattle, WA 98195-1580

D. H. WEINBERG Ohio State University Department of Astronomy Columbus, OH 4:3210

AND

L. HERNQUIST University of California Lick Observatory Santa Cruz, CA 95064

Abstract. Using the results of cosmological simulations evolved with smoothed particle hydrodynamics, we can predict the distribution of neutral hydrogen in a (22.2 Mpc? comoving box at high redshifts. By converting these boxes into two-dimensional images, convolving with a Gaussian beam and adding noise appropriate to sensitivity estimates, we have simulated a series of observations by the proposed Square Kilometer Array Interferometer (SKAI). The capability of the SKAI to easily detect 1010 M8/beam concentrations of Heu tral Hydrogen ("galaxies") should impose significant constraints upon the large scale structure proposed in a variety of cosmological models.

1. Introduction

One of the fundamental problems posed by observational cosmology over the last two decades is the origin of the large scale structure ofthe Universe, determined through extensive surveys in both the plane of the sky and in

451


452 D. R. INGRAM ET AL.

redshift space (e.g. Geller and Huchra 1989). Theoretical cosmology has presented us with a variety of proposed solutions to this problem in the form of competing cosmological models such as cold dark matter (CDM, see Peebles 1993) and models that incorporate a cosmological constant (Peebles 1993, Ratra and Peebles 1988). Recently, with the use of a new generation of computers, such as the Cray C-90, cosmological simulations have begun to attain levels of resolution and robustness sufficient to propose significant observational tests of cosmological model predictions.

The proposed Square Kilometer Array Interferometer (SKAI, see Braun 1996) is a radio telescope that will be capable of groundbreaking work in observational cosmology. Using simulations based upon TreeSPH (Katz et al 1995, KWH hereafter), an algorithm that combines a hierarchical tree method (Barnes and Hut 1986) and smoothed-particle hydrodynamics (SPH, see Hernquist and Katz 1989 and references therein), we have produced three-dimensional maps of neutral Hydrogen in order to predict the 21 centimeter emission that will be observed by the SKAI in its current proposed configuration.

TABLE 1. Simulation Parameters

Parameter Value Comments

Number of particles 2 x 643 half gas

Ho 50 km/secjMpc n 1.0

fh 0.05

U8 0.7 amplitude

Box size (22.2 Mpc? comoving size Gas particle mass 1.5 x 108 Me;) "Dark" particle mass 1.2 x 109 Me;) Background radiation 10-22 ergjsec/cm2 /sr/Hz scales as V-I

Boundary conditions Periodic

2. Parameters

The initial conditions for this simulation are listed in Table 1. Star formation (of necessity, a phenomenological algorithm due to particle sizes) was not included in this simulation. Unfortunately, the simulation runs that included star formation were not completed in time for presentation at this conference, but it is not believed that star formation will have a significant impact upon the general nature of the galaxy distribution (see KWH for a complete discussion).

L 1 n e

SPH SIMULATIONS

600

575

550

525

500

475

450~~~~~~~~~~~~~~~~~~~

500 525 550 575

Column

600 625 650

453

Figure 1. Simulated observation for 100 hours of integration time by the Square Kilometer Array of neutral Hydrogen. The image spans'" 1 Mpc on a side at a redshift of 2. Contours range from 109 to 5 X 1010 M 0/beam. The object at Column 575, Line 500 is a "galaxy" with roughly 1011 M 0.

3. Results

SKAI observations were simulated in the following way. Starting with a simulation data cube of neutral Hydrogen, we then selected an axis along which to view the box. The box was then divided into a number of velocity channels. In order to get a reasonable estimate of the sensitivity of the SKAI in a single observation, we chose to look at frequency channel widths of 100 kHz, which corresponds to a velocity width of 65 kmjs at a redshift of 2. This two-dimensional image of neutral Hydrogen was then converted into solar masses of Hydrogen per velocity channel (Braun 1995):

454 D. R. INGRAM ET AL.

600

@

575

@

550

L i 525 n

f) e

500

0 0 475

450~~-L~~~~~~-L~~~~-L~~~~

500 525 550 575 600 525 650

Column

Figure 2. Simulated observation for 10000 hours of integration time by the Square Kilometer Array of neutral Hydrogen. The image spans ~ 1 Mpc on a side at a redshift of 2. Contours are the same as in Figure 1. Objects of order 109 M 8 in neutral Hydorgen are now clearly seen.

where M H is solar masses of neutral Hydrogen per pixel, N H is column density of neutral Hydrogen, S is the pixel size in centimeters and mp is the mass of a Hydrogen atom. This image is then convolved with a 35 kpc FWHM Gaussian to represent the image of the sky seen by the SKAI beam. After normalization, we are left with an image in units of M H per beam. Using sensitivity estimates from Braun (1995), we then create a Gaussian

1 noise image with magnitude proportional to (exposure time)- 2". This noise image is convolved with a 35 kpc Gaussian (for z = 2), normalized and added to the neutral Hydrogen image. The resultant image is a simulation of data collected in a single observation of the SKAI of a given field for a given exposure time. Figure 1 shows a contour plot of the central region of such an image, for a 100 hour observation of a representative velocity channel of the simulation box. Figure 2 shots a similar contour plot, but

SPH SIMULATIONS 455

for 10000 hours of observation time. As a rule of thumb, we find that the predicted capabilities of the SKAI

should detect a 1010 M 8/beam concentration of neutral Hydrogen at the 5a level in 100 hours. This means that, for the first time, "normal" galaxies will be easily detectable at high redshifts, and the information contained in the general properties of these galaxies (such as the mass distribution and correlation function) should strongly constrain theories of structure formation and cosmological models. Filamentary structures that could give rise to Lyman-alpha forest spectral features unfortunately have brightnesses far too low (of order 105 M8/beam and lower) to be seen, but with the constraints on properties of normal galaxies that should come from the SKAI, we should have a much clearer picture of the nature of the formation of large scale structure in the Universe.

In the coming months, we intend to redo the simulations under a variety of different cosmological models in the hopes of providing realistic observational tests for new instruments such as the Square Kilometer Array. There is also an excellent discusssion on the implications of current and future simulations by Weinberg (1996).

References

Barnes J.E. and Hut, P. (1986), Nature, 324, 446. Braun, R. (1995) [private communication]. Braun, R. (1996) [this conference]. Geller, M.J. and Huchra, J.P. (1989), Science, 246,897. Hernquist, L. and Katz, N. (1989), ApJS, 70, 419. Katz, N., Weinberg, D.H. and Hernquist, 1. (1995), [preprint]. Peebles, P.J.E. (1993), Principles of Physical Cosmology, Princeton University Press,

Princeton. Ratra, B. and Peebles, P.J.E. (1988), Physical Review D, 37, 3406. Weinberg, D.H. (1996) [this conference].

SEARCHES FOR HI EMISSION FROM PROTO CLUSTERS USING THE GIANT METREWAVE RADIO TELESCOPE

Observational Strategies

G. SWARUP

National Centre for Radio Astrophysics Pune University Campus, Pune 411 007, India.

Abstract. The Giant Metrewave Radio Telescope consists of 30 antennas of 45 m diameter. Fourteen of these are located in a '" 1km X 1km area and others along three 14 km long arms of a V-shaped array. GMRT is expected to be completed by mid-96. Parameters of GMRT and of likely HI condensates in the CDM model at high z are discussed. It is likely that these may be detectable with integration times of tens of hours.

1. Introduction

Detection of the 21-cm radiation from neutral hydrogen condensates at high redshifts could provide important constraints on the models of formation of galaxies and clusters in the universe, as suggested initially by Sunyaev and Zeldovich (1972; 1975). Two major scenarios have been proposed for the formation of the large scale structures. If the matter density is dominated by relativistic particles, called the hot dark matter (HDM), massive pancake like structures form first with mass'" 1015- 16 MG. These fragment to form galaxies and clusters. However, N-body simulations including consideration of microwave background fluctuations measured by COBE indicate that the first structures in the HDM model will begin to go non-linear only at the present epoch. Current models favour a hierarchical or bottom-up formation of galaxies and clusters in the presence of Cold Dark Matter (CDM). However, CDM model is not able to explain certain observed features such as the observed superclusters and large drift velocities of clusters. It may be noted that the flux density of the red-shifted 21-cm emission is predicted to be considerably lower for the CDM model than for the HDM model (Scott and Rees, 1990; Subramanian and Padmanabhan 1993; Kumar et al. 1995).

457

M. N. Bremer et al. (eds.), Cold Gas at High Redshijr, 457-462. © 1996 Kluwer Academic Publishers.

458 G. SWARUP

A number of searches have been made for HI condensates at 151, 240 and", 327 MHz (z = 8.4,4.9 and 3.3) giving only upper limits (see Wieringa et al. 1992 and references cited therein). These observations rule out the existence of any neutral hydrogen condensates of ;<; 5 X 1013 Mra at z = 3.3 and", 1015- 16 Mra at z = 8.2. Claim of a positive detection of an HI proto cluster at z = 3.4 by Uson et al. (1991) using the VLA has not been confirmed by subsequent observations at Arecibo and Westerbork.

In Section 2 is described the Giant Metrewave Radio Telescope (GMRT) being constructed in India. It will provide considerably higher sensitivity than the Very Large Array (VLA) and the Westerbork Synthesis Radio Telescope (WSRT) for detection of HI condensates. GMRT is expected to be in operation by mid-96. The prospects for detection of HI at high redshifts and observational strategies for HI searches are discussed in Section 3. Conclusions are given in Section 4.

2. The Giant Metrewave Radio Telescope (GMRT)

Search for the 21-cm line radiation from proto clusters in the red-shift range of about 3 to 8 is one of the primary objectives of the GMRT being set up about 80 km north of Pune in India (lat 19° 06' long 74° 03') (Swarup et at. 1991). Pulsar research is another major programme. Because of its high sensitivity and angular resolution, GMRT will be a valuable instrument for studying a variety of celestial objects, such as radio emission from planets, the Sun, HII regions, supernova remnants, stellar radio sources, the Galactic Centre, nearby galaxies and distant radio galaxies and quasars.

GMRT consists of 30 fully steerable parabolic dishes of 45-m diameter. Twelve antennas are placed more or less randomly in a central array of about 1 km xl km in size and six each along three V-shaped arms of about 14 km length each, including two placed close to the central array (Fig. 1). This configuration was chosen to obtain good sensitivity for both compact and extended sources, say of the order of 10 arcsec and 3 arcmin respectively at 325 MHz. GMRT provides a hybrid configuration corresponding to the Band C arrays of the VLA. An economical design has been developed for the 45-m dishes of GMRT which uses a novel concept nick-named SMART, "Stretched Mesh Attached to Rope Trusses". This design cuts down wind forces on the antennas, yet provides reasonably high antenna efficiency upto a frequency of about 1420 MHz. The reflecting surface of the parabolic dishes consists of 0.55 mm diameter stainless steel welded wire mesh with a mesh size of 10 X 10, 15 X 15 and 20 X 20 mm in the central, middle and outer one-third area of the dish. Measured efficiency of the antennas is about 38% at 21-cm and about 55% at the longer wavelengths.

GMRT operates in six different frequency bands viz. 50 ± 10 MHz,

SEARCH FOR HI EMISSION FROM PROTOCLUSTERS 459

i

• 12 full ... n

~compaetArray

• Central Array -- --. • 'e • . -.

>Om

S,oIe

Figure 1. Locations of the thirty 45 m diameter parabolic dishes of the G M RT

150±20 MHz, 235±8 MHz, 325±20 MHz, 610±20 MHz, 1000-1430 MHz. Antenna feeds are placed on a rotating turret near the prime focus. Output of the feeds are connected to uncooled low-noise amplifiers. Measurements are possible of all the 4 Stokes parameters in the above bands. Polarizers are placed between the antenna feeds and input amplifiers at frequencies upto 610 MHz to give correlated right-handed (RH) and left-handed (LH) outputs. Only linear polarizations in orthogonal directions are amplified and correlated for the 1000 - 1430 MHz band. Although the feeds at 150, 325 and 610 MHz have a much broader bandwidth, analog filters have been placed after the RF amplifiers to minimise any man-made interference. A local oscillator system tied to a central frequency standard converts the received signals to IF centred at 70 MHz where Surface Acoustics Wave filters can be selected to obtain 5.5, 16 or 32 MHz bandwidth. The signals are brought to a central electronics building using optical fibre links. A baseband system converts the received signals to 2 side bands of 0 - 16 MHz for each of the two polarizations thus giving 4 outputs for each antenna. Base-band filters can be selected to restrict the bandwidths of each output in 9 steps ranging from 62.5 kHz to 16 MHz.

Each signal is quantized to 4-bits and then Fast-Fourier-Transformed in hardware to give 256 spectral channels. FFT outputs are then correlated to give either all 4 stokes parameters over only one side band with 128

460 G. SWARUP

TABLE 1. System parameters of GMRT for HI Searches at high values of z.

1. Frequencies (MHz) 153 235 325 610 2. Redshift, z 8.3 5.0 3.37 1.33 3. Field of View of 45-m dish (deg.) 3.1 2.0 1.4 0.75 4. Receiver Bandwidth (MHz) 6 10 32 32 5. Search Volume (106 MpC3)h~o 12 5 5 0.5 6. No. of Abell clusters1 6 2.5 2.5 0.25 7. Resolution (Central Array) (arcmin) 6.9 4.4 3.2 1.7 8. Co-moving size2 1(1 + z) (Mpc) hso 16.2 9.1 5.8 2.1 9. T,y.(K) 580 250 110 100 10. 5 Lm.s. Noise3 (mly) 3.7 1.3 0.5 0.3

1 Calculated, taking abundance of rich Abell clusters as 5 x 10-7 h~o Mpc-3. 2Co-moving size corresponding to the resolution of the Central array. 3Estimated for 10 hr integration; 14 dishes of 45 m dia; !:l.f = f!:l.v/c, where !:l.v = 1000 km S-l; both polarizations. Angular resolution for the Y-array + Central array is about 20 times smaller. hso = Hubble parameter in units of 50 kms- 1 Mpc-1 .

spectral channels or only R-R and L-L products over both side bands with 256 spectral channels each.

GMRT uses the principle of earth's rotation synthesis for mapping celestial radio sources. With its N = 30 antennas, GMRT would measure N(N-l)J2 = 435 Fourier components at any instant. A two-dimensional aperture is synthesized in about 12-hour observations, giving a resolution of about AJ D, where D is the separation between the farthest antennas. GMRT is expected to be in operation by mid-96. System parameters of the GMRT for HI searches for the central array (CN) of 14 antennas are summarized in Table 1.

3. Prospects for detection of HI emission at high redshifts and observational strategies

Searches for HI condensates made so far at 327 and 150 MHz (cf. Wierenga et al. 1992 and references therein; Bebbington 1986) indicate that pancake like structures predicted by the HDM model as precursors to the present day rich clusters may not exist at z = 3.3 and 8.4. On the other hand, in the CDM model, subgalactic scales collapse first at high redshifts and aggregate to form galaxies and clusters. The collapsed gas of Mass> 109 M0 may cool giving rise to neutral hydrogen condensates. Such clumps are indeed observed in the quasar absorption system as damped Lya systems. These systems contain mass in HI at redshifts '" (2-4) comparable to the total stellar content ofthe present day galaxies (Lanzetta et at. 1991). The

SEARCH FOR HI EMISSION FROM PROTOCLUSTERS 461

gaseous mass of a damped Lya system could exceed 1013h5~ Me:) which may evolve into groups of galaxies at later epochs (Wolfe 1993). A cluster of 2:16 galaxies has been detected at z "'" 3.4 by Giavalisco et a1. (1994).

Subramanian and Swarup (1992) noted that a large collection of gas rich protogalaxies could provide a source of HI emission. These protogalaxies could remain neutral in spite of ionization during star formation, due to recombination. Subramanian and Padmanabhan (1993) have derived several observable parameters for the red shifted 21 cm emission from HI condensates in the CDM and HDM models after normalizing power spectra of the primordial density fluctuations based on COBE results. In particular, they have used the COBE normalization to predict the co-moving number density, N, of proto condensates which will have a flux density, S, higher than given values at various redshifts. Kumar et a1. (1995) have worked out HI line profile from a collapsing spherical protocluster for the CDM model. According to above papers, the predicted value of peak flux density S at 325 MHz (z ~ 3.3) is about 0.5 mJy, line width"", 1 MHz and number density, N, of protoclusters "'" 2 X 10-6 Mpc3 for S 2: 0.5 mJy. More massive HI condensates giving S "'" 1mJy in the CDM model have nearly the same line-width but N(S) ~ (10-7 -10-8 Mpc3 ). At 235 MHz, z = 5, for S "'" 1 mJy and 1(1 + z) fV 10 Mpc, N < 10-11 Mpc3 . At 153 MHz corresponding to z = 8.3, number density of protocondensate is negligible in the CDM model. However, there is great uncertainty concerning our expectations of HI condensates at very high redshift. Scott and Rees (1990) have considered the possibility of detecting HI condensates in absorption against the cosmic microwave background radiation.

It is proposed to make extensive search for HI condensates using GMRT at 325 ± 16 MHz in the first instance. Since N increases by a factor of about 6 to 10 for decrease of S by a factor of about 2 for a smaller condensate in the CDM model, it may be preferable to select a few regions, far away from strong radio sources and galactic plane, and to integrate for many tens of hours, say 40 to 50 hours. One may also search in several directions for 10 hours each for detecting condensates in the direction of damped Lya systems, radio galaxies and perhaps a few high z QSOs, particularly if they show any evidence of clustering. In view of considerable uncertainty about the formation of structure at high z, it is desirable also to search at 235, 150 and 610 MHz. A foreground rich cluster may increase the probability of detecting high z HI condensates due to gravitational lensing. Search for absorption by HI condensates of radio emission from distant radio galaxies and quasars will also provide valuable information about the high redshift universe.

462 G. SWARUP

4. Conclusion

Detection of cold HI gas at high redshifts, z "" 3 - 8, will be invaluable for elucidating the problem of galaxy formation. Search for protoclusters is one of the primary objectives of GMRT although being a versatile instrument it will be used for studying variety of celestial objects.

Acknowledgements. I thank K. Subramanian and R. Malik for many valuable discussions.

References

Bebbington, D.H.O.: 1975. Mon. Not. R. Astron. Soc., 218, pp. 577. Giavalisco, M., Steidel, C.C. & Szalay, A.S.: 1994,Astrophys. J., 425, pp. L5. Kumar, A., Padmanabhan, T. & Subramanian, K. : 1995, Mon. Not. R. Astron. Soc.,

272, pp. 544. Lanzetta, K.M., Wolfe, A.M., Turnshek, P.A., Lu, L., McMohan, R.G., Hazard, C. : 1991,

Ap.J.S., 77, pp.1. Scott, D. & Rees, M.J.: 1990 Mon. Not. R. Astron. Soc., 247, pp. 510. Subramanian, K. & Padmanabhan, T. : 1993, Mon. Not. R. Astron. Soc., 265, pp. 101. Subramanian, K. & Swarup, G.: 1992, Nature, 359, pp. 512. Sunyaev, R.A. & Zeldovich, Ya.B. : 1972, Astron. Astrophys., 20, pp. 189. Sunyaev, R.A. & Zeldovich, Ya.B. : 1974, Mon. Not. R. Astron. Soc., 171, pp. 375. Swarup, G., Ananthakrishan, S., Kapahi, V.K., Rao, A.P., Subrahmanya, C.R. &

Kulkarni, V.K.: 1991, Current Science, 60, pp. 95. Uson, J.M., Bagri, D.S. & Cornwell, T.J.: 1991, Phys. Rev. Lett, 67, pp. 3328. Wierenga, M.H., de Bruyn, A.G. & Katgert, P. : 1992, Astron. Astrophys., 256, pp. 331. Wolfe, A.M. : 1993, Astrophys. J., 402, pp. 411.

INDEX

AAT 262, 385 Abell 478 196, 206 Abell 496 202, 207 Abell 2125 186-193 Abell 2645 190-193 Abell 2670 153-156 Abundances 206, 215-224, 243, 256,

280, 346 AO 0235+164219 Arecibo 199-201,285-288,440 Arp 220 25, 29-30, 41-43, 49, 57-59,

65-66, 315, 416, 432-436 Arp 29950 ASCA 207, 208, 346

B0218+347217-223 B20902+34 34,174-175,350 Background radiation, see Microwave

background BALQSOs, see Quasars BBXRT 206 BR 1202-0725 12, 69, 318, 325-329,

331-335 Butcher-Oemler effect 183

radio emission 183-194

Centaurus A 217 CFHT 367-368 Circumstellar dust emmission 80 Cirrus 79 C IV absorbers 243, 249-252 Cloverleaf quasar, see H1413+117 Clustering of IR sources 80 Clusters of galaxies 165-169, 183-194,

199-209 blue galaxy population 184 Butcher-Oemler effect 183-194 cooling flows 195, 199-200,205 formation 105, 460 gas content 105, 145-157 high redshift 184

COBE 65, 96,411-422,425,457,461

463

Cold Dark Matter (CDM) 94-96, 123, 457-458

Compound interferometry 280-281 Cooling 123 Cooling flows 195, 199-200, 205, 373-

378 cold phase temperature 203, 208 dust 210 HI 21 cm measurements 195 molecular gas 195 X-ray absorption 195, 205-207

Correlation function 116, 252 Cosmic microwave background, see Mi

crowave background Cosmological models, 3, 15,94-96,122,

137, 457-458 CDM 94-96, 123, 138, 457-458 HDM 94,457-458

Cosmological simulations 94-96, 451 CTIO 262

Damped Lya: systems (DLAs) 4-6, 102, 121-122, 172-175, 233, 241-243,261-265,267-276,279-280,285,309,312,318

abundances 133 as gas rich dwarf galaxies, 234 as spiral disks, 234 dust content 122, 234, 241-243,

279 molecular gas content 276, 414-

416 Dark matter 9, 123 Density parameter (Q) 110, 121, 171 Density parameter in HI (QH1 ), see

H I integral mass content Dissipation 26-28, 37-44 Dust 11,311-322

formation 11 in cooling flows 210 ill radio galaxies 306-307, 311-

;{20, 379, 400

464

t.emperat.ure 316 t.hermal emission 55,58,78,311-

322,325-327, 337-341,343-347,418-427

Dust. reddening and obscurat.ion 10, 78, 279, 282, 391

in galaxies 285 in quasars 279, 391 in radio galaxies 379, 400

Dust t.ori 78, 344-345 Dwarf galaxies, see Galaxies

EINSTEIN 205 Elliptical galaxies 37 ESO 196, 363-364 European Large Area ISO Survey (ELAIS)

77

Filament.s 96 FIRAS 425 Fluctuat.ion spectrum 4, 104, 110 Fundament.al plane 38

Galactic winds 142 Galaxies

bars 34 colours 188-192 dust. 43-44, 71, 316-317 dwarf 16, 122, 129, 160 elliptical, see Ellipt.ical galaxies evolut.ion 16, 61-63, 71, 79, 85-

90, 121-134 faint blue population 65 FIR luminosity 311 formation 16, 27, 43-44, 71, 78,

121-134, 161, 199, 233-234, 306, 312, 363, 367, 412, 460

hyperluminous, see Hyperlumi-nous galaxies

infrared 61-73 interaction 47,162,191,199,312,

412 low surface brightness (LSB) 160-

161 luminous IR 25-34,37-41,47,55-

56, 77-78, 334 merging 25-34, 37-44, 47-52, 65,

88, 122, 234, 412 spiral, see Spiral galaxies starburst, see Star burst galaxies

ultraluminous infrared, see Ultraluminous infrared galaxIes

G-dwarf problem 129 GINGA 239 GMRT 14-15, 439, 457-461 Gravitational collapse 124, 139 Gravitational lensing 4, 8, 68, 217-

224, 228-231, 249, 261-265, 285,293-298,301,325,333-334,337-341,343-347,429

amplification bias 264 bypass effect 263-264 time delay 231

Gunn-Peterson effect 102, 172 helium 245

H1413+117 8, 25, 31-33, 68, 293, 301-304, 429-431

Ha emission 44 HE 1104-1805262 HI 21 cm

absorption 5,14-15,102,171-175, 221-222, 249, 267-270, 279, 437-449,461

emission 5, 102, 200, 267, 275, 437-449, 457

in cooling flows 195 in mergers 47 spin temperature 13-15, 173-175,

267, 271, 279 H I integral mass content (QHd 104,

121, 261, 270-271, 285-286 Hierarchical clustering 94, 122, 413-

416, 425 Hopkins Ultraviolet Telescope (HUT)

245 Hot Dark Matter (HDM) 94-96,457-

458 HST 12, 17,235,245,379-384,403

Medium Deep Survey 86 Hydra A 196,202 Hydra I cluster 150-153 Hydrodynamical simulat.ions (SPH) 94-

96 Hyperluminous galaxies 61,67-73,337-

341

Intergalactic medium (IGM) 3, 102, 126, 172, 245

Interstellar medium (ISM) 255-256 Intracluster Medium (ICM) 199-200 IRAM 30 m 295, 302, 332-333 IRAS 37, 48, 57-58, 77-80, 86, 331,

337,340,343,346,423 IRAS 09104+4109340 IRAS F10214+4724 8, 25, 30-31, 67,

293-300, 315, 337-341, 343-347, 357,429-431

IRAS galaxies 61-73,160 luminosity evolution 61-62, 424 luminosity function 61-62, 82

ISO 73, 77, 80, 318, 334

JCMT 12, 55-56, 318, 321, 326-328, 332-333, 423

Jeans mass 103 Jets 364, 394,387-389

K-correction 416-420, 423-427 Keck Telescope 12, 17, 85, 138-139,

249

Large scale structure 80, 93, 137 Leiden-Berkeley Deep Survey (LBDS)

86 Line emission

Cr 301, 421-422 C II 328-329, 421-422 CO 301 HCN 301

Low surface brightness galaxies, see Galaxies

Luminosity function 64-65 Lya absorbers 99,109,115,137,172,

246, 249, 268 abundances 249-251 cloud geometry 109-110 cloud size 109, 116 clustering 116, 249-252 damped, see Damped Lya sys-

tems Doppler parameter 139 evolution 116 modelling 109, 137 structure 252, 137

Ly a emission 15 Lya forest, see Lya absorbers Lyman limit systems 102, 172, 263

M51432-436

M82 85, 255, 293, 315, 416 Megamasers 9-10 Mergers 47,191,312

H r morphology 47 star formation 191

Metal depletion 256, 280, 347 Metal production 43, 126-133 Mg II absorbers 126, 219

465

Microwave background 13,65,72,96, 103,227-231,432,457

fluctuations, 457 Milky Way 121, 420-422, 431 Millimeter Array (MMA) 9, 12, 17,

418-422,429,434-436 Millimeter interferometry 25, 418-422,

430-436 Molecular gas

absorption lines 4, 8, 215, 227 abundance 215-217 emission lines 8, 57-58, 202 excitation temperature 216, 431-

436 Molecular tori 224-225 Mrk 231 25, 28, 57-59, 66 Mrk 273 50, 57-59

NGC 253 9, 293,431 NGC 1068347 NGC 1275 196 NGC 161450 NGC4945347 NGC 5128, see Centaurus A NGC 6240 26-29, 57-59 NGC 7252 37, 47

n, see Density parameter OVRO 25-34

Pancakes 3,109-113 PC 1647+4631A 414 PKS 0458-020261 PKS 0528-2508, 234-237,267-275 PKS0745-191196,202 PKS 1413+135217-221 PKS 1830-211 217, 227-231 Photoionization 398 Proto clusters 13-14, 105, 172, 175-176,

275,457-461 Protogalaxies 103,199,235-237,416-

419, 461

466

disks 268, 366 gas mass 238, 437 Lya emission 15, 234-235

Q 1422+2309250 Q 1504+377 217-224 Q 1529-008 262 Quasars, QSOs 3

absorption lines, see Lya absorbers BALQSOs 301, 334 CSS 391 dust 71, 332-334 evolution 242 infrared emission 37 lensing 301 mm continuum 331 mm-UV correlation 333 optical spectra 391 orientation effects 391 radio loud 227, 233, 391 submm continuum 325 submm line emission 328 surveys 394 unification schemes 391 X-ray absorption 239

Radio galaxies 3CR 349-350, 359, 379-383 403 alignment effect 349-359,38'2-383

403 ' dust 71, 306-307,355-361 environment 373-378, 403 evolution 88, 358-361 gas dynamics 352, 364-365, 367-

370, 386-387 gas mass 306, 351 high redshift 305, 385, 397 Lya absorption 364-366 Lya emission 349-361, 363, 367-

368, 385 J-lJy sources 61, 73, 85 molecular gas content 305 opt~cal morphologies 365, 403 optIcal spectrum 397 orientation effects 379, 399 polarisation 355-361 403 spectra 397-401 ' star formation 85, 320, 367 stellar populations 306, 358, 370

403 '

Ram-pressure stripping 199 Recombination 3, 103 Redshift surveys 62, 73, 86-87 93 Reionization 3, 103, 172 ' Resonant scattering 397 ROSAT 206, 239

SCUBA 44,73,318-321,334,423 SEST 199, 202, 229-230 Seyfert galaxies 67, 71, 339 Shocks 38-42 Source counts 61-65 Spectral energy distribution (SED) 64-

70,337-341,343-347 Spiral galaxies

dust 72 luminosity function 185 opacity 72 precursors 268 radio emission 185 radio-IR correlation 185 star formation 78

Square Kilometer Array Interferometer (SKAI, SKA) 5-9, 13-17, 102,439-449,451

Starburst galaxies 25-34 37-44 61 71,82,85,88,312,325 '334' 337-339, 414-416, 423-427 '

Star formation 125 in mergers 191

Stromgren sphere 172 Structure formation 101-103, 111, 122

137 ' bottom-up 15

Submm continuum 312 source counts 424 surveys 423

Superbubbles 255 extragalactic 255 kinematics 258

Supernovae 123, 126, 185, 255-256 Superwinds 255 Surveys

quasars 394 redshift, see Redshift surveys submillimeter 423

Tully-Fisher relation 129, 161-164, 165-169, 436

Two-point correlation function, see Cor-relation function

TX0211-122357

UKIRT 196 Ultraluminous infrared galaxies (ULIRGs)

25-34, 37-44, 47, 55-59, 65-67, 73, 309, 312, 325, 414-416, 423-427

Unification schemes 78,343,391 UV background 110, 123

Virgo cluster 146-150 VLA 12, 48, 85, 185, 199, 202, 306,

438, 458 VLT 17 Voids 93-94, 159-164 VV 114 27-28, 50

WENSS 281 WGACAT 240-241 Winds, galactic 142 WSRT 6-9,12,176-179,267,438,458

X-ray absorption 206, 239-240 gas mass 208 gas temperature 208 in cooling flows 195, 200, 205-

207

Zel'dovitch approximation 111

467

cold gas at high redshift: proceedings of a workshop celebrating the 25th anniversary of the...

Documents