ISO Science Legacy: A Compact Review of ISO Major Achievements (Space Science Reviews) 9781402038433, 1402038437

Citation preview

ISO SCIENCE LEGACY A Compact Review of ISO Major Achievements

Cover figures: Background ISOCAM image of the Rho Ophiuchi Cloud, Abergel et al. Astronomy and Astrophysics 315, L329 Left inserts, from top to bottom: 170 μm ISOPHOT map of the Small Magellanic Cloud (40 pixel size, 1 resolution) from Wilke et al., A&A 401, 873–893 (2003). 2–200 micron composite spectrum of the Circinus galaxy obtained with the SWS and LWS spectrometers showing a plethora of atomic, ionic and molecular spectral, along with various solid-state features from dust grains of different sizes in Verma et al. this volume. Water vapour spectral lines detected in the atmospheres of all four giant planets and Titan, in Cernicharo and Crovisier, this volume. Cristalline silicates detected by ISO in different environments, in stars (young and old) and in comet Hale-Bopp in Molster and Kemper, this volume. Pure rotational hydrogen lines observed towards the molecular hydrogen emission peak of the Rho Ophiuchi filament in Habart, this volume.

ISO SCIENCE LEGACY A Compact Review of ISO Major Achievements

Edited by CATHERINE CESARSKY European Southern Observatory, Garching, Munich, Germany and ALBERTO SALAMA European Space Agency, Madrid, Spain

Reprinted from Space Science Reviews, Volume 119, Nos. 1–4, 2005

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

ISBN: 1-4020-3843-7

Published by Springer P.O. Box 990, 3300 AZ Dordrecht, The Netherlands Sold and distributed in North, Central and South America by Springer, 101 Philip Drive, Norwell, MA 02061, U.S.A. In all other countries, sold and distributed by Springer, P.O. Box 322, 3300 AH Dordrecht, The Netherlands

Printed on acid-free paper

All Rights Reserved c 2005 Springer  No part of the material protected by this copyright notice may be reproduced or utilised 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 Printed in the Netherlands

TABLE OF CONTENTS

Foreword

vii

GENERAL FRANK MOLSTER and CISKA KEMPER / Crystalline Silicates

3–28

JOSE´ CERNICHARO and JACQUES CROVISIER / Water in Space: The Water World of ISO

29–69

EMILIE HABART, MALCOLM WALMSLEY, LAURENT VERSTRAETE, STEPHANIE CAZAUX, ROBERTO MAIOLINO, PIERRE COX, ˆ / FRANCOIS BOULANGER and GUILLAUME PINEAU DES FORETS Molecular Hydrogen

71–91

DAVID ELBAZ / Understanding Galaxy Formation with ISO Deep Surveys

93–119

SOLAR SYSTEM ´ THIERRY FOUCHET, BRUNO BEZARD and THERESE ENCRENAZ / The Planets and Titan Observed by ISO

123–139

¨ ´ ´ ´ and JACQUES CROVISIER / THOMAS G. MULLER, PETER ABRAH AM Comets, Asteroids and Zodiacal Light as Seen by ISO

141–155

STARS and CIRCUMSTELLAR MATTER BRUNELLA NISINI, ANLAUG AMANDA KAAS, EWINE F. VAN DISHOECK and DEREK WARD-THOMPSON / ISO Observations of Pre-Stellar Cores and Young Stellar Objects

159–179

DARIO LORENZETTI / Pre-Main Sequence Stars Seen by ISO

181–199

MARIE JOURDAIN DE MUIZON / Debris Discs Around Stars: The 2004 ISO Legacy

201–214

JORIS A. D. L. BLOMMAERT, JAN CAMI, RYSZARD SZCZERBA and MICHAEL J. BARLOW / Late Stages of Stellar Evolution

215–243

INTERSTELLAR MEDIUM ALAIN ABERGEL, LAURENT VERSTRAETE, CHRISTINE JOBLIN, ˆ RENE´ LAUREIJS and MARC-ANTOINE MIVILLE-DESCHENES / The Cool Interstellar Medium

247–271

´ ELS PEETERS, NIEVES LETICIA MART´IN-HERNANDEZ, NEMESIO J. ´ ´ RODRIGUEZ-FERNANDEZ and XANDER TIELENS / High Excitation ISM and Gas

273–292

EMMANUEL DARTOIS / The Ice Survey Opportunity of ISO

293–310

OUR LOCAL UNIVERSE . . . MARC SAUVAGE, RICHARD J. TUFFS and CRISTINA C. POPESCU / Normal Nearby Galaxies

313–353

APRAJITA VERMA, VASSILIS CHARMANDARIS, ULRICH KLAAS, DIETER LUTZ and MARTIN HAAS / Obscured Activity: AGN, Quasars, Starbursts and ULIGs Observed by the Infrared Space Observatory

355–407

. . . AND BEYOND SEB OLIVER and FRANCESCA POZZI / The European Large Area ISO Survey

411–423

LEO METCALFE, DARIO FADDA and ANDREA BIVIANO / ISO’s Contribution to the Study of Clusters of Galaxies

425–446

FOREWORD Building upon pioneering work in the 1960’s and 1970’s using ground-based, rocket- and balloon-borne systems, the realm of infrared astronomy was fully opened by the first cryogenic telescope in space – IRAS, launched in 1983. Over its ten-month lifetime, IRAS surveyed almost the whole sky in four broad infrared bands. This survey permitted the first evaluations of the total energy emitted by various systems in our galaxy and in the local universe. However, it could not address the detailed mechanisms and processes responsible for the emission detected, nor the exploration of the distant universe. IRAS results graphically illustrated to astronomers the need for sensitive infrared observatories, allowing detailed spatial and spectroscopic study of specific targets. All over the world, high priority was assigned to cooled space infrared telescopes. Following the Japanese IRTS mission, the first major satellite of this type to fly was ESA’s Infrared Space Observatory (ISO). Launched in November 1995, ISO completed almost 30 000 scientific observations in its 2.5-year operational lifetime. Making use of its four sophisticated and versatile scientific instruments (a camera, a photopolarimeter and two spectrometers), ISO provided astronomers with a wealth of data of unprecedented sensitivity at infrared wavelengths from 2.5 to 240 μm. ISO has made, and continues to make, lasting contributions to all areas of astronomy, from the solar system to the frontiers of cosmology, unravelling the history of the universe. Between 1996 and 2004, over 1200 papers appeared in the refereed literature based on ISO data. NASA’s Spitzer Space Telescope, launched eight years later, has enhanced capabilities compared to ISO and, once again, infrared astronomers are offered matchless observing opportunities. However, the published ISO results and the ISO archive remain a valuable resource for research work. They provide guidelines for studies not only with Spitzer but also with future facilities, such as ASTRO- F, SOFIA, Herschel, JWST and ALMA. With the Spitzer Space Telescope now in full operations, we thought that it would be beneficial to the astronomical community to have at hand, in a single volume, a review of the main discoveries owed to the ISO satellite. We did not ask the ISO founding fathers and mothers to write the articles, but instead turned mostly towards younger astronomers whose careers have been strongly influenced by ISO. The articles have been refereed by ourselves or by other scientists at our request. The book is organised as follows: first, overviews of four major themes investigated with ISO (crystalline silicates, molecular hydrogen, deep surveys; water in the universe), and then thirteen chapters reviewing ISO science from the solar system to the distant universe. It is not possible to gather in one book all the advances due to ISO, but we hope that this compendium of over 480 pages will give the essence of the original results obtained by the first full-fledged space infrared observatory. Catherine Cesarsky and Alberto Salama Guest editors

GENERAL

CRYSTALLINE SILICATES FRANK MOLSTER1,∗ and CISKA KEMPER2 1 ESTEC/ESA,

Keplerlaan 1, 2201 AZ Noordwijk, The Netherlands Fellow) Department of Physics and Astronomy, UCLA, 475 Portola Plaza, Los Angeles, CA 90095-4705; Present address: University of Virginia, Department of Astronomy, P.O. Box 3818, Charlottesville, VA 22903-0818, USA (∗ Author for correspondence: E-mail: [email protected]) 2 (Spitzer

(Received 16 July 2004; Accepted in final form 2 November 2004)

Abstract. One of the big surprises of the Infrared Space Observatory (ISO) has been discovery of crystalline silicates outside our own Solar system. It was generally assumed before that all cosmic silicates in space were of amorphous structure. Thanks to ISO we know now that crystalline silicates are ubiquitous in the Galaxy (except for the diffuse ISM) and sometimes even in very large quantities (>50% of the small dust particles). The evolution of the crystalline silicates is still not completely clarified, but the combination of theoretical modeling and observations have already shed light on their life-cycle. The absence of crystalline silicates in the diffuse ISM provides us with information about the dust amorphization rate in the ISM. Keywords: crystalline silicates, infrared astronomy

1. Introduction Before the Infrared Space Observatory (ISO) opened the mid- and far-infrared range for high-resolution spectroscopy, it was generally assumed that cosmic dust silicates were of amorphous structure. The crystalline silicates, the highly ordered counterparts of the amorphous silicates, were only known to be present on earth, in the solar system in comets (Hanner et al., 1994; Hanner, 1996), Interplanetary Dust Particles (IDPs) (MacKinnon and Rietmeijer, 1987; Bradley et al., 1992) and in the dust disk of β-Pictoris (Knacke et al., 1993; Fajardo-Acosta and Knacke, 1995), also a crystalline olivine feature was reported in the polarized 10 μm spectrum of AFGL2591 (Aitken et al., 1988). Apart from the crystalline silicates in the IDP’s, that were found with the aid of transmission electron microscopy and only later confirmed by infrared spectroscopy (Bradley et al., 1992), in all other cases the crystalline silicate features were found by infrared spectro(polari)metry around 10 μm. With the present day knowledge it is relatively easy to understand why crystalline silicates were only discovered to be ubiquitous after ISO was operational. Before ISO was launched the primary MIR/FIR window for observations was around 10 μm. And although the crystalline silicates do have strong features in this area, they are in general overwhelmed by emission from the much more abundant and Space Science Reviews (2005) 119: 3–28 DOI: 10.1007/s11214-005-8066-x

 C

Springer 2005

4

F. MOLSTER AND C. KEMPER

warmer amorphous silicates. Furthermore, most of the crystalline silicates have a (relatively) low temperature ( 1 and all oxygen atoms will be locked into carbon monoxide. Hence, oxygen will be unavailable to build other molecules. Nevertheless, Melnick et al. (2001) have reported the detection of H2 O with SWAS in the direction of IRC+10216, the prototype of carbon-rich AGB stars. The far-IR spectrum of this object as observed by ISO was reported by Cernicharo et al. (1996b) and no clear indication of H2 O could be found as the spectrum is dominated by a forest of emission lines from CO and HCN. Melnick et al. (2001) have suggested that the observed water vapor is arising from the evaporation of cometary bodies orbiting IRC+10216. More sensitive observations with Herschel of several water lines having different excitation conditions could provide a clearer picture about the origin of H2 O in this star. The situation is different for post-AGB carbon-rich stars. Herpin and Cernicharo (2000) have presented ISO LWS observations of the proto-planetary nebula CRL 618, a carbon rich object in a very fast evolutionary phase prior to reaching the planetary nebula stage. The far-infrared spectrum is essentially dominated by CO lines (see Figure 19). All the other species have much lower intensities contrary to that found in the AGB star IRC+10216 where HCN lines (from the ground and vibrationally excited states) are as strong as those of CO (see Cernicharo et al., 1996b). The main difference between both objects has to be found in the physical structure of their CSEs. That of CRL 618 has a central hole in molecular species filled by a bright H II region. In addition to the lines of CO, 13 CO, HCN and HNC, Herpin and Cernicharo (2000) reported the detection of H2 O and OH emission together with the fine structure lines of [O I] at 63 and 145 μm. The abundances of these species relative to CO are 4 × 10−2 , 8 × 10−4 and 4.5 in the regions where they are produced. The authors have suggested that O-bearing species other than CO are produced in the innermost region of the circumstellar envelope. UV photons from the central star photodissociate most of the molecular species produced in the AGB phase and allow a chemistry dominated by standard ion-neutral and neutral-radical reactions. Not only these reactions allow the formation of O-bearing species but also modify the abundances of C-rich molecules like HCN and HNC for which the authors have found an abundance ratio of 1, much lower than what is typical in AGB stars. Herpin et al. (2002) have made a comparative study of three carbon-rich post-AGB objects, CRL618, CRL2688 and NGC7027. In the early stages of the AGB to PN evolution (represented by CRL2688) the far-infrared spectrum is dominated by CO lines. In the intermediate stage, e.g., CRL618, very fast outflows are present which, together with the strong UV field from the central star, dissociate CO. The released atomic oxygen is seen via its atomic lines and allows the formation of new O-bearing species, such as H2 O, H2 CO and OH. At the planetary nebula stage, e.g., NGC7027, a large fraction of the old CO AGB material has been reprocessed. The far-infrared spectrum is dominated by atomic and ionic lines. New species,

WATER IN SPACE

57

Figure 19. ISO/LWS continuum-substracted spectra of CRL618. The lines of CO, 13 CO, HCN, H2 O and OH are indicated by arrows, while those of HNC are indicated by vertical lines (from J = 22–11 at 150.627 μm to J = 17–16 at 194.759 μm). The figure is from Herpin and Cernicharo (2000) and the result of their model is shown as a solid line.

such as CH+ (see Cernicharo et al., 1997d), appear. The water vapor formed during the protoplanetary nebula stage has been photodissociated. The chemical evolution during the protoplanetary nebula stage has been modeled, using only neutral and neutral-radical reactions, by Cernicharo (2004). The radicals are the products of the photodissociation of molecules formed during the AGB phase of the star. The author has shown that H2 O, H2 CO (a molecule detected in CRL618 by Cernicharo et al., 1989), and CO2 are easily formed in the warm and dense gas of the photodissociation region surrounding the hot central star. These models also predict very large abundances for other carbon chains and carbon clusters as those found by Cernicharo et al. (2001a,b). Hence, photochemistry plays an important role in the chemical evolution of the envelopes of post-AGB stars making possible the presence of O-bearing species in a medium where the physical conditions and the atomic abundances would result in very little amounts of these species.

58

J. CERNICHARO AND J. CROVISIER

4. Water in Galaxies Due to the small size of the ISO telescope the observation of water in gas phase toward extragalactic objects is strongly limited in sensitivity. Nevertheless, observation of water ice in 18 galaxies, from a sample of 103 galaxies observed with ISO, has been reported by Spoon et al. (2002). They have found that water ice is present in most ULIRGs, whereas it is weak or absent in the large majority of Seyferts and starburst galaxies. Only a few objects show gas phase molecular emission or absorption: Arp220, NGC1064 and NGC253. Goicoechea, Martin-Pintado and Cernicharo (2005), have analyzed the OH emission/absorption in NGC1064 and NGC253, while Arp220 has been studied in detail by Gonz´alez-Alfonso et al. (2004). Figure 20 shows the far-IR spectrum of Arp220 as observed by ISO. The spectrum looks very similar to that

Figure 20. ISO LWS spectrum of Arp220 where the most prominent line features are identified. The gray line shows the best model for the lines (from Gonz´alez-Alfonso et al., 2004).

WATER IN SPACE

59

Figure 21. Continuum-normalized spectra of SgrB2 (M) and Arp220. The main carriers of some line features are indicated (from Gonz´alez-Alfonso et al., 2004).

of Sgr B2 (see Figure 21). The modeling of the Arp220 far-infrared spectrum (see below), and of the OH lines in NGC1064 and NGC253, was done by comparison of their spectra with that of some well-studied Galactic sources, which provides important clues about the physical conditions in the regions where the lines are formed. In Arp220 Gonz´alez-Alfonso et al. (2004) have observed absorption in the lines of OH, H2 O, NH, NH3 and CH in addition to [O I] at 63 μm and emission in the [C II] line at 158 μm. They have modeled the continuum and the emission/absorption of all observed features by using non-local radiative transfer codes. The continuum from 25 to 1300 μm corresponds to a nuclear region, optically thick in the far-infrared, with a size of 0.4 and a dust temperature of 106 K surrounded by an extended region (2 ) heated mainly through absorption of the nuclear infrared radiation. The OH column densities are high toward the nuclear region (2–6 × 1017 cm−2 ) and in the extended region ( 2 × 1017 cm−2 ) while the water column density is high toward the nucleus (2–10 × 1017 cm−2 ) and lower in the extended region. The column densities in a halo that accounts for the absorption in the lowest lying lines are similar to what is found in the diffuse clouds toward the star-forming regions in the Sgr B2 molecular cloud in the neighborhood of the Galactic Center. One of the most surprising results from Gonz´alez-Alfonso et al. (2004) is the large column density needed to explain the observations of NH and NH3 (1.5 × 1016 cm−2 and 3 × 1016 cm−2 , respectively) while NH2 , a molecule detected in Sgr B2 (M) by Goicoechea and Cernicharo (2001b), has a column density below 2 × 1015 cm−2 .

60

J. CERNICHARO AND J. CROVISIER

The excellent ISO spectrum of Arp220 makes it a reference template for understanding the dusty interstellar medium of ULIRGs. The model of Gonz´alez-Alfonso et al. (2004) shows the important role of the dust photons in pumping the highexcitation lines of OH and H2 O, i.e., a similar situation to than that found in Sgr B2(M) (Goicoechea and Cernicharo, 2002). The modeling of water vapor and OH in Arp220 indicates the prominent role of PDR molecular chemistry in the extended region of Arp220, while chemistry in hot cores and shocks in the nucleus is also contributing to the observed abundance of H2 O. Herschel, with its three instruments (HIFI, PACS and SPIRE), and its large telescope (3.5 m), will permit to observe the molecular content of more distant objects. The full frequency coverage provided by PACS and SPIRE will allow to make more detailed and sensitive studies of H2 O and OH, while HIFI will provide high spectral resolution of the line profiles of these molecules. It is worth noting that the H2 O models for water in Arp 220, and for sources in our galaxy, have required the simultaneous observation and modeling of other species, in particular OH and CO. Moreover, the use of galactic sources as templates as initial input for the interpretation of molecular extragalactic observations could be a mandatory step. Goicoechea et al. (2005) have used Orion and Sgr B2 as templates to interpret the far-infrared spectra of NGC253 and NGC1064. Thanks to ISO, this information will be available when observing extragalactic sources with Herschel. It will be, without any doubt, the only way to interpret correctly the H2 O lines, which will provide important information on the chemistry and physical conditions of the gas.

5. Water in the Solar System Water is ubiquitous in the Solar System. It is present – – – – – –

as ice and liquid on the surface of the Earth; as ice and – presumably, in the past – as liquid on Mars; in various amounts in the atmospheres of most planets; as the main constituent of cometary ices; in primitive meteorites such as carbonaceous chondrites; as ice on numerous planetary satellites and – presumably – on trans-Neptunian objects (TNO); – and even in the Sun’s atmosphere. Water is also a requisite for the apparition and evolution of life. It is thus important to know the cycle of water in the Solar System. Problems that are still open include: – to which extent cometary (and TNO) ices are coming from unprocessed interstellar ice; – what is the origin of water in the giant planets’ atmospheres;

WATER IN SPACE

61

– to which extent did the infall of comets, asteroids and interstellar dust particles (IDPs) contributed to Earth (and Mars) water. The contribution of ISO to cometary and planetary studies has also been reviewed by Crovisier (2000), Encrenaz (2000), Fouchet et al. (2004) and M¨uller et al. (2004). 5.1. COMETS Water is the main constituent of cometary ices. Its sublimation governs cometary activity close to the Sun (r ≤ 4 AU). However, the direct observation of cometary water is very difficult from the ground, where only lines from infrared hot bands can be observed. The water ro-vibrational lines are emitted by fluorescence excited by solar radiation. Their measurement allows us to determine the water production rates in comets. Water was best observed by ISO in the exceptionally bright comet C/1995 O1 (Hale-Bopp), as part of a target-of-opportunity program, in September–October 1996. The comet was then at r = 2.8 AU from the Sun and its water production rate was 3.3×1029 molecules s−1 . The ν1 , ν3 and hot bands around 2.7 μm were resolved in rotation with the SWS (Figure 22; Crovisier et al., 1997a, 1999b). The ν2 band at 6.5 μm was also detected with the SWS, for the first time in a comet (Crovisier et al., 1997b). Several rotational lines (212 –101 and 303 –212 around 180 μm and weaker

Figure 22. The SWS spectrum of water in the 2.7 μm region observed in comet C/1995 O1 (HaleBopp) on 27 September and 6 October at 2.8 AU from the Sun (top). Line assignations are indicated. The synthetic fluorescence spectrum which corresponds to the best fit to the data is shown at the bottom. It corresponds to Q[H2 O] = 3.6 × 1029 molecules s−1 , Trot = 28.5 K and Tspin = 28 K (adapted from Crovisier et al., 1997a).

62

J. CERNICHARO AND J. CROVISIER

Figure 23. Several rotational lines of water observed in comet C/1995 O1 (Hale-Bopp) with the LWS. The C II line is not cometary; it comes from the background (from Crovisier et al., 1999a).

ones) were observed with the LWS (Figure 23). This was also the first detection of rotational lines of water in a comet, before the 110 –101 line at 557 GHz line could be observed by heterodyne techniques with the SWAS and Odin satellites. In comet Hale-Bopp, water was only detected when the comet was at r = 2.9 and 3.9 AU. At farther distances (4.6 AU pre-perihelion and 4.9 AU post-perihelion), only CO and CO2 (which are much more volatile species) were detected, with PHT-S. This shows the evolution between CO-dominated (at large r ’s) and H2 Odominated (at small r ’s) regimes of sublimation of cometary ices. ISO also observed H2 O in two short-period, Jupiter-family comets. In 103P/Hartley 2, which was observed close to its perihelion at r = 1 AU, the water production rate was 1.2 × 1028 molecules s−1 (Colangeli et al., 1999; Crovisier et al., 1999a). In 22P/Kopff, observed in a less productive state after perihelion at r = 1.9 AU, the water production rate was only 3.7 × 1027 molecules s−1 and the water lines at 2.7 μm were just detected by the SWS (Crovisier et al., 1999a). The SWS spectra obtained in the 2.6 μm region on comets Hale-Bopp and Hartley 2 allowed us to investigate the rotational distribution of water, and therefore its excitation conditions. Water excitation is governed by radiative excitation of the fundamental bands of vibration by the Sun IR radiation, collisions in the inner coma, and radiative trapping effects (Bockel´ee-Morvan, 1987). On the other hand,

WATER IN SPACE

63

the kinetic temperature of the coma is governed by the balance between radiative cooling (through emission of water rotational lines) and photolytic heating (principally through water photodissociation). In the field of view of the SWS, cometary water is partly relaxed to its lowest rotational levels. As predicted by modeling, the observed rotational temperatures of water are low: 28 K for Hale-Bopp, 20 K for Hartley 2, < 11 K for Kopff (Crovisier et al., 1997a, 1999a,b). Water exists in two nuclear-spin states according to the spins of its hydrogen atoms: ortho (I = 1) and para (I = 0). Conversions between the two states are forbidden, so that the ortho-to-para population ratio is linked to the temperature (spin temperature Tspin ) of the last re-equilibration of water. The SWS spectra allowed us to determine Tspin of water in comets Hale-Bopp and Hartley 2. Tspin = 28 and 35 K were derived respectively for the two comets (Crovisier et al., 1997a, 1999a,b). A similar value (29 K) was observed in the past for comet 1P/Halley (Mumma et al., 1993). The spin temperature of cometary ammonia (which is indirectly determined from the visible spectrum of NH2 ; Kawakita et al., 2004) has also similar values (25–32 K). These Tspin , all observed in a small range for comets of different orbits and different dynamical history, are still to be understood. It has been argued that Tspin could reflect the temperature of the grains where the molecules formed, but it seems hard to believe that the ortho-to-para ratio could be preserved over cosmological times without re-equilibration. Tspin could reflect the temperature of the inner nucleus ices. But then, one would expect different temperatures for comets with different orbits and history. It would be interesting to investigate by laboratory experiments if Tspin is preserved during the sublimation process, or if some fractionation occurs. It would also be interesting to compare the cometary water ortho-to-para ratio with those observed for interstellar water. 5.2. PLANETS

AND

S ATELLITES

5.2.1. Mars Water on Mars is now studied by a wealth of means. ISO also contributed to the observation of H2 O in Mars’ atmosphere. Ro-vibrational absorption lines were observed in the 2.6 and 6.5 μm regions with the SWS, while many rotational lines were observed all over the 20–200 region with the SWS and LWS (Figure 24; Encrenaz et al., 1999; Burgdorf et al., 2000; Lellouch et al., 2000). From these observations, the mean water concentration at the ground was 4 × 10−4 at that time (31 July 1997) and saturation was reached at an altitude of 10 km. This corresponds to 15 μm of precipitable water. 5.2.2. Giant Planets and Titan An important result of ISO was the discovery of water in the stratospheres of the four giant planets and Titan. These detections are based upon water rotational lines observed in emission with the SWS and LWS, using the Fabry-P´erot (Figure 25;

64

J. CERNICHARO AND J. CROVISIER

Figure 24. Lines of water in the 2.6 μm region of the spectrum of Mars, observed with the SWS grating. A modeled spectrum is superimposed (from Lellouch et al., 2000).

Figure 25. Examples of detections of H2 O lines in all four giant planets and Titan, with the SWS Fabry-P´erot (from Feuchtgruber et al., 1999).

Feuchtgruber et al., 1997, 1999; Coustenis et al., 1998; Lellouch et al., 2002; Encrenaz, 2003). The disk-averaged water column densities are observed in the range ≈1014 cm−2 (Uranus, Neptune, Titan) to ≈2 × 1015 cm−2 (Jupiter, Saturn). Because water condenses at the tropopause, its presence cannot be explained by an internal source of these objects. An external influx has to be invoked. This is the same for CO2 , observed in the stratosphere of Jupiter, Saturn and Uranus. 5.3. THE C YCLE

OF

WATER

IN THE

S OLAR S YSTEM

Observations at infrared (ground-based and from ISO) and radio wavelengths point to the similarity between interstellar and cometary ices (Ehrenfreund et al., 1997;

WATER IN SPACE

65

Bockel´ee-Morvan et al., 2000). This does not prove that comets accreted unprocessed interstellar ices, but rather suggests that similar chemical processes were at work in the primitive Solar Nebula and in interstellar clouds. Impacts are an important phenomenon for the redistribution of water in the Solar System. Shock chemistry following the collisions of large bodies (planetesimals, comets, asteroids) will reshuffle most of the original molecular species. The oxygen compounds of the impactor could then lead to the formation of H2 O (and CO2 ) in the post-impact chemistry. On the other hand, for infall of IDPs, the original molecules could be preserved. Icy IDPs could thus be an important source of water. Several studies have considered asteroids, comets and meteoroids as plausible sources of the water accreted by the Earth (e.g., Morbidelli et al., 2000). Martian water could have a similar origin. The collision of comet Shoemaker-Levy 9 and Jupiter is a well-documented case. Unfortunately, there is still ambiguity on the nature of the impactor (comet or asteroid?); its chemical nature could not be precisely characterized from pre-impact observations (Crovisier, 1996). There is little doubt, however, that the chemical species observed after the impact result from shock chemistry, rather than coming from the impactor (Lellouch, 1996). In this context, three different sources have been invoked to explain the presence of oxygen species (water and carbon dioxide) discovered by ISO in giant planets (e.g., Encrenaz, 2003; Lellouch et al., 2002; Fouchet et al., 2004): – infall of IDPs; – sputtering from icy rings and satellites; this is the preferred source for Saturn and Titan; – impact of small bodies (comets and asteroids); this is certainly the case for Jupiter, for which the fall of comet Shoemaker-Levy 9 in 1994 provided the main source of oxygen.

6. Conclusions Water has been found everywhere in space by ISO. The interpretation of the data is not straightforward but, nevertheless, we had for the first time the opportunity to study the role of water in the chemistry of all kind of objects in space, from solar system bodies to star-forming regions, evolved stars and galaxies. Despite the limited spectral resolution of most water observations with ISO, the possibility to observe its full far and mid infrared spectrum has opened great possibilities in the interpretation of the data. Complex radiative transfer models able to treat the very large opacities of water have been developed, the chemistry of water on dust grains and gas phase has received an extraordinary and unique input from ISO, the presence of water in the upper layers of giant planets and their moons has told us about the chemistry of these objects, the detection of water in extragalactic sources

66

J. CERNICHARO AND J. CROVISIER

could be used as a tracer of the physical and chemical processes prevailing in the nuclei of galaxies, etc. Future missions like Herschel will benefit from the huge information ISO has provided on water vapor. The selection of water lines according to criteria of sensitivity to the physical parameters of the clouds, their interpretation and the prediction of the full spectrum of water in several clouds will be a mandatory step to prepare the Herschel mission. However, little new information will be obtained if collisional rates adapted to the temperatures of the clouds in the ISM and CSM are not available. Several aspects have to be improved for future water observations: we urgently need collisional rates for H2 O–H2 at temperatures going from 10 to 2000 K (dark clouds to evolved stars). For AGB stars collisional rates between the ground and the bending mode are also necessary. These data have to be provided by physicists and chemists doing ab initio calculations or laboratory experiments. A very close collaboration between astronomers and the world of quantum chemistry is needed in order to promote these calculations and to be sure that we will get the maximum scientific output from Herschel when observing water in space. Acknowledgements We would like to thank J.R. Pardo, F. Daniel and J.R. Goicoechea for useful comments and suggestions and a critical reading of the manuscript. They have kindly provided several figures and calculations for this paper. J. Cernicharo thanks the Spanish Ministry of National Education (MEC) for funding support under grants AYA2003-2785, AYA2002-12125E, ESP2002-11650, AYA2002-10113-E, and ESP 2002-01627. Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, the Netherlands and the United Kingdom) and with the participation of ISAS and NASA. References Barlow, M., et al.: 1996, Astron. Astrophys. 315, L241. Beckwith, S., Persson, S. E., Neugebauer, G., and Becklin, E. E.: 1978, Astrophys. J. 223, 484. Benedettini, M., et al.: 2000, Astron. Astrophys. 359, 148. Blake, G. A., Sutton, E. C., Masson, C. R., and Phillips, T. G.: 1987, Astrophys. J. 315, 621. Bockel´ee-Morvan, D.: 1987, Astron. Astrophys. 181, 169. Bockel´ee-Morvan, D., Lis, D. C., Wink, J. E., et al.: 2000, Astron. Astrophys. 353, 1101. Boonman, A. M. S. and van Dishoeck, E.: 2003, Astron. Astrophys. 403, 1003. Burgdorf, M. J., Encrenaz, T., Lellouch, E., et al.: 2000, Icarus 145, 79. Ceccarelli, C., et al.: 1999a, Astron. Astrophys. 331, 372. Ceccarelli, C., et al.: 1999b, Astron. Astrophys. 342, L21. Ceccarelli, C., et al.: 2002, Astron. Astrophys. 383, 603. Cernicharo, J.: 1985, ATM a Program to Compute Atmospheric Transmision Between 0–1000 GHz, IRAM Internal Report.

WATER IN SPACE

67

Cernicharo, J.: 1988, Ph.D. Thesis, Universit´e de Paris VII. Cernicharo, J. and Gu´elin, M.: 1987, Astron. Astrophys. 176, 299. Cernicharo, J., Gu´elin, M., Martin-Pintado, J., and Mauersberger, R.: 1989, Astron. Astrophys. 222, L1. Cernicharo, J., Thum, C., Hein, et al.: 1990, Astron. Astrophys. 231, L15. Cernicharo, J., Gonz´alez-Alfonso, E., Alcolea, J., et al.: 1994, Astrophys. J. 432, L59. Cernicharo, J., Gonz´alez-Alfonso, E., and Bachiller, R.: 1996a, Astron. Astrophys. 305, L5. Cernicharo, J., et al.: 1996b, Astron. Astrophys. 315, L201. Cernicharo, J., Lim, T., Cox, P., et al.: 1997a, Astron. Astrophys. 323, L25. Cernicharo, J.: 1997, The Far Infrared and Submillimeter Universe, 15–17 April 1997, Grenoble, France, ESA SP-401, p. 91. Cernicharo, J., Gonz´alez-Alfonso, E., and Lefloch, B.: 1997b, First Workshop on Analytical Spectroscopy, 6–8 October 1997, ESA SP-419, p. 23. Cernicharo, J., X-Liu, Gonz´alez-Alfonso, E., et al.: 1997c, Astrophys. J. 483, L65. Cernicharo, J.: 1998, Astrophys. Space Sci. 263, 175. Cernicharo, J., Gonz´alez-Alfonso, E., et al.: 1999a, The Universe as Seen by ISO, Paris, 1999. Cernicharo, J., Pardo, E., Serabyn, et al.: 1999b, Astrophys. J. 520, L131. Cernicharo, J., Goicoechea, J. R., and Caux, E.: 2000, Astrophys. J. 534, L199. Cernicharo, J., et al.: 2001a, Astrophys. J. 546, L123. Cernicharo, J., et al.: 2001b, Astrophys. J. 546, L127. Cernicharo, J.: 2004, Astrophys. J. 608, L41. Cernicharo, J., Goicoechea, J. R., Daniel, F., et al.: in press, Astrophys. J. Cheung, A. C., Rank, D. M., Townes, C. H., et al.: 1969, Nature 221, 626. Clegg, P. E., et al.: 1996, Astron. Astrophys. 315, L38. Cohen, M., Barlow, M., Sylvester, R. J., et al.: 1999, Astrophys. J. 513, L135. Colangeli, L., Epifani, E., Brucato, J. R., et al.: 1999, Astron. Astrophys. 343, L87. Comito, C., Schilke, P., Gerin, M., et al.: 2003, Astron. Astrophys. 402, 635. Coustenis, A., Salama, A., Lellouch, E., et al.: 1998, Astron. Astrophys. 336, L85. Crovisier, J.: 1996, in: Noll, K. S., Weaver, H. A., and Feldman, P. D. (eds.), The Collision of Comet Shoemaker-Levy 9 and Jupiter, Cambridge University Press, Cambridge, p. 31. Crovisier, J., Leech, K., Bockel´ee-Morvan, D., et al.: 1997a, Science 275, 1904. Crovisier, J., Leech, K., Bockel´ee-Morvan, D., et al.: 1997b, in: Heras, A. M., Leech, K., Trams, N. R., and Perry, M. (eds.), First ISO Workshop on Analytical Spectroscopy, ESA SP-419, p. 137. Crovisier, J., Encrenaz, T., Lellouch, E., et al.: 1999a, in: Cox, P., and Kessler, M. F. (eds.), The Universe as Seen by ISO, ESA SP-427, p. 161. Crovisier, J., Leech, K., Bockel´ee-Morvan, D., et al.: 1999b, in: Cox, P., and Kessler, M. F. (eds.), The Universe as Seen by ISO, ESA SP-427, p. 137. Crovisier, J.: 2000, in: Mihn, Y. C., and van Dishoeck, E. F. (eds.), Astrochemistry: From Molecular Clouds to Planetary Systems (IAU Symposium 197), ASP, p. 461. Daniel, F., Goicoechea, J. R., Cernicharo, J., et al.: in press, Astrophys. J. de Graauw, T., et al.: 1996, Astron. Astrophys. 315, L49. Encrenaz, T., Feuchtgruber, H., Burgdorf, M., et al.: 1999, in: Cox, P., and Kessler, M. F. (eds.), The Universe as Seen by ISO, ESA SP-427, p. 173. Encrenaz, T.: 2000, in: Casoli, F., Lequeux, J., and David, F. (eds.), Infrared Space Astronomy, Today ´ and Tomorrow (Ecole de Physique des Houches). EDP Sciences, Paris, p. 89. Encrenaz, T.: 2003, Planet. Space Sci. 51, 89. Ehrenfreund, P., d’Hendecourt, L., Dartois, E., et al.: 1997, Icarus 130, 1. Feuchtgruber, H., Lellouch, E., de Graauw, T., et al.: 1997, Nature 389, 159. Feuchtgruber, H., Lellouch, E., Encrenaz, T., et al.: 1999, in: Cox, P., and Kessler, M. F. (eds.), The Universe as Seen by ISO, ESA SP-427, p. 133. Fouchet, T., B´ezard, B., and Encrenaz, T.: 2004, The Planets Observed by ISO, this volume.

68

J. CERNICHARO AND J. CROVISIER

Gensheimer, P. D., Mauersberger, R., and Wilson, T. L.: 1996, Astron. Astrophys. 314, 281. Genzel, R. and Stutzki, J.: 1989, Annu. Rev. Astron. Astrophys. 27, 41. Genzel, R., Reid, M. J., Moran, J. M., and Downes, D.: 1981, Astrophys. J. 244, 884. Gibb, E. L., Whittet, D. C. B., Bougert, A. C. A., and Tielens, A. G. G. M.: 2004, ApJSS 151, 35. Goicoechea, J. R. and Cernicharo, J.: 2001a, Astrophys. J. 554, L213. Goicoechea, J. R. and Cernicharo, J.: 2001b, in: Pilbratt, G. L., Cernicharo, J., Heras, A. M., Prusti, T., and Harris, R. (eds.), The Promise of FIRST, ESA SP-460; Noordwijk: ESA/ESTEC, p. 413. Goicoechea, J. R. and Cernicharo, J.: 2002, Astrophys. J. 576, L77. Goicoechea, J. R., Rodriguez-Fernandez, N. J., and Cernicharo, J.: 2004, Astrophys. J. 600, 214. Goicoechea, J. R., Martin-Pintado, J., and Cernicharo, J.: 2005, Astrophys. J. 619, 291. Gonz´alez-Alfonso, E. and Cernicharo, J.: 1993, Astron. Astrophys. 279, 506. Gonz´alez-Alfonso, E., Cernicharo, J., Bachiller, R., and Fuente, A.: 1995, Astron. Astrophys. 293, L9. Gonz´alez-Alfonso, E. and Cernicharo, J.: 1997, Astron. Astrophys. 322, 938. Gonz´alez-Alfonso, E., Cernicharo, J., Alcolea, J., and Orlandi, M. A.: 1998a, Astron. Astrophys. 334, 1016. Gonz´alez-Alfonso, E., Cernicharo, J., van Dishoeck, E. F., et al.: 1998b, Astrophys. J. 502, L169. Gonz´alez-Alfonso, E. and Cernicharo, J.: 1999, Astrophys. J. 525, 845. Gonz´alez-Alfonso, E., Wright, C. M., Cernicharo, J., et al.: 2002, Astron. Astrophys. 386, 1074. Gonz´alez-Alfonso, E., Smith, H. A., Fischer, J., and Cernicharo, J.: 2004, Astrophys. J. 613, 247. Harwit, M. Neufeld, D. A., Melnick, G. J., and Kaufman, M. J.: 1998, Astrophys. J. 497, L105. Helmich, F. P., van Dishoeck, E. F., and Jansen, D. J.: 1996a, Astron. Astrophys. 313, 657. Helmich, F. P., van Dishoeck, E. F., Black, J. H., et al.: 1996b, Astron. Astrophys. 315, L173. Herpin, F. and Cernicharo, J.: 2000, Astrophys. J. 530, L129. Herpin, F., Goicoechea, J. R., Pardo, J. R., and Cernicharo, J.: 2002, Astrophys. J. 577, 961. Hjalmarson, A., et al.: 2003, Astron. Astrophys. 402, L39. H¨uttemeister, S., Wilson, T. L., Mauersberger, R., Lemme, C., Dahmen, G. and Henkel, C.: 1995, Astron. Astrophys. 294, 667. Jacq, T., Jewell, P. R., Henkel, C., Walmsley, C. M., and Baudry, A.: 1988, Astron. Astrophys. 199, L5. Jacq, T., Walmsley, C. M., Henkel, C., et al.: 1990, Astron. Astrophys. 228, 447. Justtanont, K., de Jong, T., Helmich, F. P., et al.: 1996, Astron. Astrophys. 315, L217. Justtanont, K., de Jong, T., Tielens, A. G. G. M., et al.: 2004, Astron. Astrophys. 417, 625. Kawakita, H., Watanabe, J., Furusho, R., et al.: 2004, Astrophys. J. 601, 1152. Kessler, M. F., et al.: 1996, Astron. Astrophys. 315, L27. Lellouch, E.: 1996, in: Noll, K. S., Weaver, H. A., and Feldman, P. D. (eds.), The Collision of Comet Shoemaker-Levy 9 and Jupiter, Cambridge University Press, Cambridge, p. 213. Lellouch, E., Encrenaz, T., de Graauw, T., et al.: 2000, Planet. Space Sci. 48, 1393. Lellouch, E., B´ezard, B., and Moses, J. I.: 2002, Icarus 159, 112. Lis, D. C. and Goldsmith, P. F.: 1990, Astrophys. J. 356, 195. Liseau, R., Ceccarelli, C., Larsson, B., et al.: 1996, Astron. Astrophys. 315, L181. Maret, S., Ceccarelli, C., Caux, E., Tielens, A. G. G. M., and Castets, A.: 2002, Astron. Astrophys. 395, 573. Melnick, G. J., et al.: 2000, Astrophys. J. 539, L77. Melnick, G. J., Neufeld, D. A., Ford, K. E. S., et al.: 2001, Nature 412, 160. Menten, K. and Melnick, G. J.: 1991, Astrophys. J. 377, 647. Menten, K. L., Melnick, G. J., and Phillips, T. G.: 1990a, Astrophys. J. 350, L41. Menten, K. L., Melnick, G. J., Phillips, T. G., and Neufeld, D. A.: 1990b, Astrophys. J. 363, L27. Molster, T. J., et al.: 2001, Astron. Astrophys. 372, 165. Moneti, A., Cernicharo, J., and Pardo, J. R.: 2000, Astrophys. J. 549, L203. Morbidelli, A., Chambers, J., Lunine, J. I., et al.: 2000, Meteoritics Planet. Sci. 35, 1309.

WATER IN SPACE

69

Moro-Martin, A., Noriega-Crespo, A., Molinari, S., Testi, L., Cernicharo, J., and Sargent, A.: 2001, Astrophys. J. 555, 146. ´ M¨uller, T. G., Abrah´ am, P., and Crovisier, J.: 2004, this volume. Mumma, M. J., Weissman, P. R., and Stern, S. A.: 1993, in: Levy, E. H., and Lunine, J. I. (eds.), Protostars and Planets III, University of Arizona Press, Tucson, p. 1177. Neufeld, D. A. and Kaufman, M. J.: 1993, Astrophys. J. 418, 263. Neufeld, D. A., Lepp, S., and Melnick, G. J.: 1995, Astrophys. J. 100, 132. Neufeld, D. A., Chen, W., Melnick, G. J., et al.: 1996, Astron. Astrophys. 315, L237. Neufeld, D. A., Zmuidzinas, J., Schilke, P., and Phillips, T. G.: 1997, Astrophys. J. 488, L141. Neufeld, D. A., Feuchtgruber, H., Harwit, M., and Melnick, G.: 1999, Astrophys. J. 517, L147. Neufeld, D. A., Ashby, M. L. N., Bergin, E. A., et al.: 2000, Astrophys. J. 539, L111. Neufeld, D. A., Bergin, E. A., Melnick, G. J., and Goldsmith, P. F.: 2003, Astrophys. J. 590, 882. Nisini, B., Lorenzetti, D., Cohen, M., et al.: 1996, Astron. Astrophys. 315, L321. Nisini, B., Benedettini, M., Giannini, T., et al.: 2000, Astron. Astrophys. 360, 297. Nisini, B., Giannini, T., and Lorenzetti, D.: 2002, Astrophys. J. 574, 246. Nisini, B.: 2003, Astrophys. Space Sci. 287, 207. Omont, A., Moseley, S. H., Forveille, T., et al.: 1990, Astrophys. J. 355, L27. Pardo, J. R.: 1996, Ph.D. Thesis, Universit´e Pierre et Marie Curie, Paris, France. Pardo, J. R., Cernicharo, J., and Serabyn, G.: 2001a, IEEE Trans. Antennas Propag. 49, 1683. Pardo, J. R., Cernicharo, J., Herpin, F., et al.: 2001b, Astrophys. J. 562, 799. Phillips, T. G., Kwan, J., and Huggins, P. J.: 1980, in: Andrew, B. H. (ed.), IAU Symposium 87: Interstellar Molecules, Reidel, Dordrecht, p. 21. Phillips, T. G., van Dishoeck, E., and Keene, J.: 1992, Astrophys. J. 399, 533. Plume, R., Kaufman, M. J., Neufeld, D. A., et al.: 2004, Astrophys. J. 605, 247. Polehampton, E. T., Baluteau, J. P., Ceccarelli, C., Swinyard, B. M., and Caux, E.: 2002, Astron. Astrophys. 388, L44. Polehampton, E. T., Menten, K., Brunken, S., Winnewise, G., and Baluteau, J. P.: 2005, Astron. Astrophys. 431, 203. Saraceno, P., Ceccarelli, C., Clegg, P., et al.: 1996, Astron. Astrophys. 315, L293. Sempere, M. J., Cernicharo, J., Lefloch, B., et al.: 2000, Astrophys. J. 530, L123. Snell, R. L., Howe, J. E., Ashby, M. L. N., et al.: 2000, Astrophys. J. 539, L101. Spoon, H. W. W., Keane, J. V., Tielens, A. G. G. M., et al.: 2002, Astron. Astrophys. 385, 1022. Sylvester, R. J., Kemper, F., Barlow, M. J., et al.: 1999, Astron. Astrophys. 352, 587. Truong-Bach, R. J., et al.: 1999, Astron. Astrophys. 345, 925. van der Tak, et al.: 2005, in: Wilson, A. (ed.), Proceedings of the Dusty and Molecular Universe: A Prelude to Herschel and ALMA, 27–29 October 2004, Paris, France, ESA SP-577, Noordwijk, ESA Publications Division, Netherlands, ISBN 92-9092-855-7, p. 431. van Dishoeck, E. and Helmich, F.: 1996, Astron. Astrophys. 315, L177. van Dishoeck, E., Wright, C. M., Cernicharo, J., et al.: 1998, Astrophys. J. 502, L173. van Dishoeck, E.: 2004, Annu. Rev. Astron. Astrophys. 42, 119. Waters, J. W., Gustincic, J. J., Kakar, R. K., et al.: 1980, Astrophys. J. 235, 57. Wright, C. M., van Dishoeck, E. F., Black, J. H., et al.: 2000, Astron. Astrophys. 358, 689. Zmuidzinas, J., Blake, G. A., Carlstrom, J., et al.: 1995, in: Haas, M. R., Davidson, J. A., and Erickson, E. F. (eds.), ASP Conference Series, Vol. 73, p. 33. Zubko, V., et al.: 2004, Astrophys. J. 610, 427–435.

MOLECULAR HYDROGEN EMILIE HABART1,∗ , MALCOLM WALMSLEY1 , LAURENT VERSTRAETE2 , STEPHANIE CAZAUX1 , ROBERTO MAIOLINO1 , PIERRE COX2 , ˆ 2 FRANCOIS BOULANGER2 and GUILLAUME PINEAU DES FORETS 1 Osservatorio 2 Institut

Astrofisico di Arcetri, INAF, Largo E. Fermi 5, I-50125 Firenze, Italy d’Astrophysique Spatiale, Universit´e Paris-Sud, 91405 Orsay, France (∗ Author for correspondence: E-mail: [email protected])

(Received 16 July 2004; Accepted in final form 18 November 2004)

Abstract. Observations of H2 line emission in galactic and extragalactic environments obtained with the Infrared Space Observatory (ISO) are reviewed. The diagnostic capability of H2 observations is illustrated. We discuss what one has learned about such diverse astrophysical sources as photondominated regions, shocks, young stellar objects, planetary nebulae and starburst galaxies from ISO observations of H2 emission. In this context, we emphasise use of measured H2 line intensities to infer important physical quantities such as the gas temperature, gas density and radiation field and we discuss the different possible excitation mechanisms of H2 . We also briefly consider future prospects for observation of H2 from space and from the ground. Keywords: molecular hydrogen, infrared: spectroscopy, molecular processes, ISM: molecules, ISM: clouds, star formation, circumstellar matter, infrared: galaxies

1. Physics of H2 1.1. ASTROPHYSICAL IMPORTANCE

OF

H2

Molecular hydrogen is the most abundant molecule in the Universe and plays a fundamental role in many astrophysical contexts (e.g., Dalgarno, 2000). It is found in all regions where the shielding of the ultra-violet (UV) photons, responsible for its photo-dissociation, is sufficiently large, i.e. where AV  0.01–0.1 mag. H2 makes up the bulk of the mass of the dense gas in galaxies and could represent a significant fraction of the total baryonic mass of the Universe. Further on, H2 plays two main roles that render it key for our understanding of the interstellar medium, in particular of the processes regulating star formation and the evolution of galaxies. Firstly, the formation of H2 on grains initiates the chemistry of interstellar gas; secondly, H2 is recognized as a major contributor to the cooling of astrophysical media. H2 has also specific radiative and collision properties that make it a diagnostic probe of unique capability. Many competing mechanisms could contribute to Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands, and the United Kingdom), and with the participation of ISAS and NASA. Space Science Reviews (2005) 119: 71–91 DOI: 10.1007/s11214-005-8062-1

 C

Springer 2005

72

E. HABART ET AL.

its excitation (see Section 1.2) and H2 can serve as probe of a wide range of physical environments. Thus, as we understand reasonably well its radiative and collision properties we can construct realistic models of the response of H2 to its surroundings.

1.2. EXCITATION

AND

FORMATION

OF

H2

The hydrogen molecule is highly symmetric as it has containing two identical hydrogen atoms. Due to this symmetry, the molecule has no dipole moment and all ro-vibrational transitions within the electronic ground state are quadrupolar with low spontaneous coefficient A values. The molecule exists in two, almost independent states, namely ortho-H2 (spins of H nuclei parallel) and para-H2 (spins antiparallel). There are no radiative transitions between ortho- and para- H2 but ortho-para conversion may occur through proton exchange reactions between H2 and, for example, H0 and H+ . Molecular hydrogen may be excited through several mechanisms: (1) In the ˚ the molecule is radiatively presence of far-ultraviolet radiation (FUV, λ > 912 A), pumped into its electronically excited states. As it decays back into the electronic ground state, it populates the high vibrational levels, and the subsequent cascade to v = 0 gives rise to optical and infrared fluorescent emission, and a characteristic distribution of level populations (e.g., Black and van Dishoeck, 1987, Sternberg, 1989). This excitation mechanism is observed in photon-dominated regions (PDRs) where it dominates the excitation of ro-vibrational and high rotational levels. (2) If the gas density and temperature are high enough, inelastic collisions can be the dominant excitation mechanism, at least for the lower energy levels (e.g., Le Bourlot et al., 1999). In dense PDRs (n H  104 cm−3 ) and shocks, collisions maintain the lowest pure rotational levels in thermal equilibrium. (3) H2 formation in excited states can also contribute to the excitation of the molecule. (4) Finally, in environments such as active galactic nuclei (AGNs) or X-ray emitting young stellar objects where hard X-rays are capable to penetrate deeply into zones opaque to UV photons, X-rays can dominate H2 excitation (Maloney et al., 1996; Tine et al., 1997). Except in the early Universe, most H2 is thought to be produced via surface reactions on interstellar grains (e.g., Gould and Salpeter, 1963; Hollenbach and Salpeter, 1971), since gas-phase reactions are predicted to be too slow. But due to our limited understanding of the relevant properties of interstellar grains (composition, structure and hydrogen coverage) and hence of grain surface reactions, the H2 formation mechanism is not yet understood. Numerous theoretical and experimental studies have thus been dedicated to a better understanding of the H2 formation process (Gould and Salpeter, 1963; Hollenbach and Salpeter, 1971; Sandford and Allamandola, 1993; Duley, 1996; Parneix and Brechignac, 1998; Pirronello et al., 1997; Pirronello et al., 1999; Takahashi et al., 1999; Katz et al., 1999; Williams et al., 2000; Sidis et al., 2000; Biham et al., 2001; Joblin et al., 2001; Cazaux and

MOLECULAR HYDROGEN

73

Tielens, 2002; Cazaux and Tielens, 2003). Another approach to this issue is to use observations to provide estimates of the H2 formation rate in different regions of the ISM in order to see how it depends on the local physical parameters. Recently, combined ISO observations of H2 line emission and dust emission as well as FUSE observations of UV pumping lines in absorption have highlighted the H2 formation process (see Section 3.1).

2. Observations of H2 Direct observation of H2 is difficult. Electronic transitions occur in the ultraviolet region, to which the Earth’s atmosphere is opaque, and observations can only be made from a space-based platform. Ro-vibrational and rotational transitions are faint because of their quadrupolar origin. In the case of rotational lines which occur in the mid-IR, the Earth’s atmosphere is at best only partly transparent. Moreover, most of H2 may hide in cool, shielded regions (e.g., Combes, 2000) where the excitation is too low to be detected through emission lines and whose extinction is too high to allow the detection of the UV pumping lines. Space-based missions in the UV (Copernicus, ORPHEUS, FUSE) and the midIR (ISO, Spitzer) provide the most stringent tests of our current understanding of H2 in space. 2.1. UV ABSORPTION L INES In the 1970s, the Copernicus satellite observed molecular hydrogen electronic lines in absorption towards nearby early type stars (e.g., Spitzer et al., 1974). This experiment yielded reliable column densities for diffuse clouds (AV ≤ 1), the first measurements of the temperature of H2 in diffuse clouds (∼50–100 K) as well as an estimate of the formation rate of H2 on interstellar dust grains (Jura, 1975). Recently, thanks to the greater sensitivity of FUSE, fainter stars with higher extinctions could be observed allowing the study of translucent clouds (e.g., Rachford et al., 2002). However, such observations are limited by the small number of sufficiently luminous background sources against which H2 can be detected in absorption. In practice, UV observations only allow the study of molecular gas in the Solar Neighborhood and provide no spatial information on the distribution of H2 . 2.2. INFRARED OBSERVATIONS Prior to the advent of ISO, infrared emission of H2 was observed in the 2 μm rovibrational lines (most notably the 1-0 S(1) line at 2.12 μm) towards Galactic shocks and PDRs, planetary nebulae, and in some instances the nuclei of active galaxies (for the first studies see e.g. Shull and Beckwith, 1982, Fischer et al., 1987). However, in PDRs and shocks only a small fraction of the H2 is vibrationally excited.

74

E. HABART ET AL.

The first observations of the v = 0 pure rotational lines of H2 have been done from the ground towards a PDR – the Orion bar – by Parmar et al. (1991). The lowest rotational transitions of H2 produced by collisions (see Section 1.2) provide a thermometer for the bulk of the gas above ∼80 K. Also, due to the low A-values of these transitions, optical depth effects are usually unimportant and thus these lines provide accurate measures of the warm (T  80 K) gas mass. With the sensitivity of the ISO instruments (i.e., SWS and CAM) several pure rotational lines of H2 have been observed in a variety of galactic and extragalactic environments. Moreover for each type of environment, numerous objects with a wide range of physical conditions were observed allowing us to adress a number of outstanding questions concerning the physics of H2 . Moreover, the complete ISO wavelength coverage (2–200 μm), which includes many lines due to gas-phase atoms (e.g., C+ , O0 , Si+ , S0 ), molecules (e.g., H2 , CO, H2 O, OH and CO2 ) and bands of polycyclic aromatic hydrocarbons (PAHs), silicates and ices, allows us to study the links between H2 , the dust and the other gas components. A review on ISO spectroscopic results from molecular clouds to disks by van Dishoeck (2004) illustrates the diagnostic capabilities of the various lines and bands detected by ISO. This article summarizes the ISO spectroscopic observations of H2 . The bulk of the data comes from the Short Wavelength Spectrometer (SWS, 2–45 μm, de Graauw et al., 1996). Relevant spectra have also been obtained with the CameraCircular Variable Filter (CAM-CVF, 2.3–16.5 μm, Cesarsky et al., 1996). 3. An Overview of ISO Observations of H2 in Galactic and Extragalactic Environments Infrared Space Observatory has observed H2 emission from sources as diverse as photo-dominated regions (PDRs), shocks associated with outflows or supernovae remnants, circumstellar environment of young stellar objects, planetary nebulae, and external galaxies. Combined with the theoretical models of PDRs and shocks these observations have allowed the derivation of important physical quantities in these objects, namely, the hydrogen density n H , the intensity of the incident farultraviolet (FUV, 6 < hν < 13.6 eV) radiation represented by χ,1 the H2 rotational excitation temperature Trot (a measure of the gas temperature, see below) and the H2 column density NH2 . In this section, these H2 observations and the inferred physical quantities are reviewed (see Table I for a summary). 3.1. PHOTON-DOMINATED REGIONS Photon-dominated regions are one of the most important sources of H2 emission. PDRs formed at the surfaces of molecular clouds where FUV radiation encounters 1 The scaling factor χ represents the intensity of the FUV field in units of the Draine (1978) average interstellar radiation field.

8

Ref.

9

≥105 4 900 0.003 10

103 –106 103 –106 150 10–20

Galactic center

3

0.5–3 × 105 0.5–2.4 × 104 390 1

Orion Bar

11

170 23

102 –105

Normal galaxy NGC 6946

5

6

500 0.1–1

104

IC 443

12

80 100

102 –105

NGC 891-outer disk

688 0.18

630 1.4 4

104 –106

Shocks Cep A W

105 –106

Orion Pk1

7

13

150 1–100

103 –106

Starbursts

720 0.069

105

YSOs LkHα225

7

13

150 1–100

103 –106

Seyferts

104 –106 104 –105 470 0.055

BD+40◦ 4124

b Excitation

a Incident

FUV radiation field expressed in units χ of the Draine (1978) average interstellar radiation field. temperature of the low H2 pure rotational levels. c H column density inferred from the intensity of H rotational lines and assuming that the population distribution of low H rotational levels is essentially 2 2 2 in LTE. References: (1) Habart et al. (2003); (2) Thi et al. (1999) (note that the observations cannot be fitted by a single excitation temperature but two components at ∼100 K and ∼620 K are needed) and Jansen et al., Jansen et al. (1994, 1995); (3) data from Bertoldi, private communication and see references in Habart et al. (2004); (4) Rosenthal et al. (2000); (5) Wright et al. (1996); (6) Cesarsky et al. (1999); (7) van den Ancker et al. (2000b); (8) Herpin et al. (2000); (9) Cox et al. (1998); (10) Rodriguez-Fernandez et al. (2001); (11) Valentijn et al. (1996); (12) Valentijn and van der Werf (1999); (13) Rigopoulou et al. (2002).

750 68

106 –107

PNe Helix

PPN CRL 618

Objects

n H (cm−3 ) χa b Trot (K) NHc 2 (1021 cm−2 )

2

1

Ref.

0.5–1×105 650 620 5

104 250 330 0.6

n H (cm−3 ) χa b Trot (K) NHc 2 (1021 cm−2 )

PDRs IC 63

Oph W

Objects

TABLE I Sample of ISO H2 observations in various astrophysical objects.

MOLECULAR HYDROGEN

75

76

E. HABART ET AL.

and dissociates the dense molecular gas. They are the transition zone between the dense, cold molecular gas and the tenuous, warm, ionized gas (for a recent review see Hollenbach and Tielens, 1999). The thermal and chemical structure of the PDR is governed by the FUV radiation2 and depends mainly on two parameters, namely, the incident FUV flux χ and the density n H . Infrared Space Observatory has observed a series of pure H2 rotational lines towards a number of PDRs sampling a wide range of excitation conditions: high excitation PDRs with χ > 103 such as the Orion Bar (Bertoldi et al., private communication), NGC 2023 (Moutou et al., 1999; Draine and Bertoldi, 2000a), NGC 7023 (Fuente et al., 1999); moderately excited PDRs with 102 < χ < 103 such as IC 63 (Thi et al., 1999), S 140 (Timmermann et al., 1996) and ρ Oph W (Habart et al., 2003); and low excitation PDRs such as the cool PDR Ced 201 (Kemper et al., 1999) or the S 140 extended region (Li et al., 2002). Detailed studies using ISO observations of both H2 and dust emission, in conjunction with ground-based NIR imaging, allows us to bring new insights mainly into (i) the temperature and density structure of PDRs, (ii) the dominant heating and cooling processes and (iii) the H2 formation process. 3.1.1. H2 as a Thermal Probe in PDRs At densities n H  104 cm−3 of bright PDRs, collisions maintain the lowest rotational levels of H2 (v = 0, J  5) in thermal equilibrium. The population of these levels is therefore a good indicator of the gas temperature. This is reflected in their excitation diagrams (see Figure 1) where their populations are consistent with the Boltzmann law. We report in Table I the excitation temperature of the H2 pure rotational levels for some PDRs with different excitation conditions. All the derived Trot are found to be similar between 300 and 700 K. It should be noted that, in spite of their simple appearance, the interpretation of these diagrams has to take account of several factors. First, the gas temperature varies rapidly through the PDR layer from several hundred K at the edge close to the excitation source to less than 30 K when AV > 1. The pure rotational H2 lines arise primarily in the outer warm layer. Second, the excitation temperature of H2 pure rotational levels gives in principle an upper limit to the kinetic temperature as UV pumping may contribute to the excitation of H2 even for low J. For several PDRs observed by ISO, the H2 line intensities and the gas temperature are found to be higher than predicted by current models (Bertoldi, 1997; Draine and Bertoldi, 1999b; Thi et al., 1999; Habart et al., 2003; Li et al., 2002). The cause of this discrepancy is either that, in the models, the gas is not hot enough or alternatively that the column density of H2 is too low in the zones where the 2 The

FUV radiation field sets the PDR chemical structure by ionizing carbon, silicon, etc., and by dissociating H2 , CO and most other molecules. It also heats the gas mainly via the photoelectric effect on dust grains.

MOLECULAR HYDROGEN

77

Figure 1. Upper panels. (a) Left: ISOCAM map, with the LW2 filter (5–8.5 μm), of the bright filament along the western edge of the ρ Ophiuchi main cloud (Abergel et al., 1996). The filament is immersed in the radiation field of the star HD147889 (see the star sign). The four SWS positions (small squares) and the CVF map positions are marked (big squares). (b) Right: ISOCAM-CVF brightness profile of the 0-0 S(3) H2 line (solid line) and PAH emission (dotted lines) along the cut going through the SWS pointings. Lower panels. (a) Left : Pure rotational H2 lines observed towards the H2 emission peak of the ρ Oph filament. The line close to 0 shows the rms noise. (b) Right : Excitation diagram (corrected for dust attenuation) for H2 ; Nu is the column density of the transition upper level, g u is the degeneracy of the upper level and T u is the upper level energy in Kelvin. The solid line is rotational temperature (Trot ) thermal distribution for an H2 ortho-to-para equal at 1.

gas is warm. Several possible explanations have been proposed. An increase of the grain photoelectric heating rate is one possibility. Draine and Bertoldi (2000a) have, in particular, shown that increased photoelectric heating rates based on an enhanced dust-to-gas ratio in the PDR due to gas-grain drift can reproduce the H2 observations in the reflection nebula NGC 2023. Another explanation, namely an increased H2 formation rate, is discussed below. Finally, non-equilibrium processes, e.g. propagation of the ionization and photodissociation fronts which will bring

78

E. HABART ET AL.

fresh H2 into the zone emitting line radiation, can be invoked but detailed models are needed. 3.1.2. H2 Formation Rate Recently, Habart et al. (2004) combined ISO observations of rotational lines of H2 towards a sample of PDRs with observations of ro-vibrational lines made with ground-based telescopes in order to constrain the formation rate of H2 under the physical conditions in the layers responsible for H2 emission. Using as a diagnostic the 0-0 S(3)/1-0 S(1) line ratio, which strongly depends on the value of the H2 formation rate R f , they find that, in regions of moderate excitation (χ ≤1000, such as Oph W, S140 and IC 63), R f is about a factor 5 larger than the standard rate estimated in diffuse clouds. On the other hand, towards regions with higher χ (such as NGC 2023 and the Orion Bar), R f is found to be similar to the standard value. This result can provide at least a partial explanation of the discrepancy discussed above since a larger R f will move the H0 /H2 transition zone closer to the edge of the PDR and consequently increase the temperature of the H2 emitting gas as well as the absolute intensity of the H2 lines. This finding of efficient H2 formation at high gas temperatures (Tgas ≥ 300 K) has also fundamental implications for our understanding of the H2 formation process. Since the residence time of weakly bound H atoms on grains (also called physisorbed) at such high temperatures is very short, a process involving strongly bound H atoms is required. An indirect chemisorption process - where a physisorbed H-atom scans the grain surface to recombine with a chemisorbed H-atom (as recently discussed by Cazaux and Tielens, 2002; Cazaux and Tielens, 2003) – is capable of explaining the ISO data (Habart et al., 2004) and could be the most important mechanism in PDRs. Moreover, small grains (radii < a few 10 nm) which dominate the total grain surface and spend most of their time at relatively low (below 30 K for χ ≤ 3000) temperatures may be the most promising surface for forming H2 in PDRs (Habart et al., 2004). The relationship between aromatic dust particles (referred to as PAHs, radii < 1 nm) and H2 has been addressed both in the laboratory (Joblin et al., 2001) and observationally (Le Coupanec, 1998; Joblin et al., 2000; An and Sellgren, 2001; Habart et al., 2003; Habart et al., 2004). In particular, using ISOCAM and ground-based observations for a sample of PDRs, (Habart et al., 2004) find that the R f /[C/H ]PAH ratio3 is constant. This suggests, but does not prove, that formation of H2 on PAHs could be important in PDRs. This result is supported by the observed spatial correlation between the H2 line emission and the PAH features (see Figure 1 and Le Coupanec, 1998; Joblin et al., 2000; An and Sellgren, 2001; Habart et al., 2003). In the H2 formation process, very small grains or VSGs (radii between a few nm and a few 10 nm) may also be important (e.g., Boulanger et al., 2004; Rappacioli et al., 2004). In the near future, Spitzer will allow us to examine the 3 [C/H] PAH

is the abundance of carbon locked up in PAHs.

MOLECULAR HYDROGEN

79

VSG emission between 15 and 40 μm towards a number of PDRs and we will be able to better understand the nature of VSGs and to determine their role in the H2 formation process. 3.1.3. Ortho-to-Para H2 Ratio In several cases, such as the ρ Oph-West (Habart et al., 2003), NGC 2023 (Moutou et al., 1999) or NGC 7023 PDR (Fuente et al., 1999), the H2 ortho-to-para ratio is significantly smaller than the equilibrium ratio of 3 expected in gas at the temperatures derived from excitation diagrams. This could indicate the presence of (i) dynamical processes, such as the advection of colder gas into the PDR, or (ii) fast conversion of ortho-H2 to para-H2 during the formation on the surface of non magnetic, non metallic grains (Le Bourlot, 2000). It must be emphasized here that, as pointed out by (Sternberg and Neufeld, 1999), care has to be taken with the interpretation of the data since a low ortho/para ratio in excited levels naturally results from the fact that the optical depth in the UV pumping lines is higher for the ortho- than for the para-H2 lines. Moreover, none of the ISO data measure the bulk of ortho- and para-H2 which resides in J = 0 and 1. We note however that for the physical conditions (χ and n H ) of PDRs discussed here, population of the rotational levels by the cascade following the FUV pumping is generally negligible below S(4) and that the lower lying rotational levels are efficiently populated by collisions. In this latter case, the excitation temperature is well defined and the populations of the J = 0 and 1 levels can be derived from a Boltzmann law. 3.2. SHOCKS As in the case of PDRs, shocks are luminous sources of H2 . Shocks in molecular clouds can be caused by a large variety of processes including outflows and jets from young stars, supernovae and expanding H II regions (for theoretical reviews see Draine and McKee, 1993; Hollenbach, 1997). In a shock, the mechanical energy is converted into heat and radiated away through cooling lines of atoms and molecules in the IR. For a wide range of shock conditions, H2 molecules are not dissociated and the gas becomes warm enough for the lowest rotational lines to be excited by collisions. Infrared Space Observatory has detected many lines of H2 in a number of shocks associated with outflows as Orion OMC-1 (Rosenthal et al., 2000), Cep A West (Wright et al., 1996), DR 21 (Wright et al., 1997), HH 54 (Neufeld et al., 1998), HH 2 (Lefloch et al., 2003), L1448 (Nisini et al., 2000) and also in shocks associated with the supernova remnants IC 443 (Cesarsky et al., 1999), 3C 391, W 44 and W 28 (Rho and Reach, 2003). ISO’s main contributions to the study of shocks have been to (i) better characterize the shock structure and physical conditions, especially that found in the slower shocks; (ii) better measure the total cooling power. The ISO data have also convincingly shown that most of the broad-band emission from shocks is due to lines (e.g., Cesarsky et al., 1999). For example, the ISOCAM

80

E. HABART ET AL.

7- and 15-μm bands are dominated by the H2 S(5) and S(1) lines, whereas the IRAS 12, 25, and 60 μm emission is due to H2 S(2), S(0), and [O I] 63 μm, respectively. As for PDRs, the pure rotational H2 transitions dominate the mid-infrared spectra and are characterized by excitation temperatures of Tex ∼ 500–1000 K (see Table I and Figure 2). In contrast, the well-studied H2 ro-vibrational lines usually yield Tex = 2000–3000 K. The most impressive collection of H2 lines is seen for Orion (see Figure 2, Rosenthal et al., 2000), where 56 H2 lines have been seen including the v = 0 S(1)–S(25) lines with upper energy levels up to 42,500 K. Since Tex increases strongly with shock velocity, the observed range of Tex has generally been interpreted in terms of two steady state C-shocks; one with a low density, low velocity ( log(L IR /L  ) ≥ 11) and ultra-luminous (ULIRGs, log(L IR /L  ) ≥ 12) infrared (IR) galaxies. In the past, when galaxies were more gaseous and formed the bulk of their present-day stars, it would have been logical to expect and detect the past star formation events of galaxies in the IR regime and to detect a large population of LIRGs/ULIRGs too. However, prior to the launch of the Infrared Space Observatory (ISO, Kessler et al., 1996), this idea was not widely spread. Partly because of a cultural reason: star formation rates (SFR) were commonly measured from optical emission lines and rest-frame UV light in galaxies. This may explain why the redshift evolution of the SFR density per unit co-moving volume computed by Madau et al. (1996) became famous. However, in the first presentation of the density of UV light per unit co-moving volume (Lilly et al., 1996) the authors were cautious to avoid converting their UV light into a SFR because of the unknown Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands, and the United Kingdom), and with the participation of ISAS and NASA. Space Science Reviews (2005) 119: 93–119 DOI: 10.1007/s11214-005-8060-3

 C

Springer 2005

94

D. ELBAZ

factor to correct for extinction. Already IRAS observations indicated a rapid decline of the co-moving number density of ULIRGs since z ∼ 0.3 (Kim and Sanders, 1998, see also Oliver et al., 1996), but this was over a small redshift range and with small number statistics. In the few years that followed the launch of ISO, several observations showed that galaxy formation could not be understood, at least on an observational basis, without accounting for dust extinction as a major ingredient. The ISO deep surveys played a major role in this process, together with other results summarized below. They clearly established that extreme events such as those taking place in local LIRGs and ULIRGs must have been more common in the past, so much that they can now be considered as a standard phase that most galaxies experienced during their lifetime, at least once, but maybe even several times. The first result of the ISOCAM surveys, as well as the ISOPHOT ones, was the great difference of the counts measured at faint flux densities with respect to local ones from IRAS (Elbaz et al., 1999; Dole et al., 2001). The universe must have been much richer in IR luminous galaxies in the past, either because galaxies were more IR luminous, at fixed galaxy density, and/or because the number density of galaxies was larger in the past, which was partly expected due to the reverse effect of hierarchical galaxy formation through mergers. The strength of the excess of faint objects came as a surprise, but its consequences on the past star formation history of galaxies was confirmed by the convergence of other observations going in the same direction. – The nearly simultaneous discovery of the cosmic infrared background (CIRB, Puget et al., 1996; Fixsen et al., 1998; Hauser and Dwek, 2001 and references therein), at least as strong as the UV-optical-near IR one, whereas local galaxies only radiate about 30% of their bolometric luminosity in the IR above λ ∼ 5 μm . – The 850 μm number counts from the SCUBA sub-millimeter bolometer array at the JCMT (Hughes et al., 1998; Barger et al., 1998; Smail et al., 2002; Chapman et al., 2003, and references therein) which also indicate a strong excess of faint objects in this wavelength range, implying that even at large redshifts dust emission must have been very large in at least the most active galaxies. – The most distant galaxies, individually detected thanks to the photometric redshift technique using their Balmer or Lyman break signature showed the signature of a strong dust extinction. The so-called “β-slope” technique (Meurer et al., 1999) used to derive the intrinsic luminosity of these galaxies and correct their UV luminosity by factors of a few (typically between 3 and 7, Steidel et al., 1999; Adelberger and Steidel, 2000) was later on shown to even underestimate the SFR of LIRGs/ULIRGs (Goldader et al., 2002). – The slope of the sub-mJy deep radio surveys (Haarsma et al., 2000). It has now become clear that the cosmic history of star formation based on rest-frame UV or emission line indicators of star formation such as [0II] or [Hα]

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

95

strongly underestimate the true activity of galaxies in the past if not corrected by strong factors due to dust extinction. Although distant galaxies were less metal rich and much younger, they must have found the time to produce dust rapidly in order to efficiently absorb the UV light of their young stars. We have tried to summarize in the following the role played by the ISO deep surveys in establishing this new perspective (see also Genzel and Cesarsky, 2000). However, 10 years after ISO’s launch we are still trying to understand the consequences of these findings on galaxy formation scenarios. Are these distant LIRGs really similar to local ones? What do they teach us about the connection between star and galaxy formation on one side (IMF, triggering of star formation, conditions of star formation, . . .) and between galaxy and large-scale structure formation on the other side (role of the environment in triggering star formation events, galaxy versus group and cluster formation, ellipticals vs. spirals, . . .)? How much energy radiated by Compton thick embedded active nuclei remains undetected even by the Chandra and XMM-Newton X-ray observatories? These questions together with others that will be discussed in the following demonstrate the liveliness of this field that will continue to feed the next generation of telescopes and instruments to come such as Herschel, ALMA, the James Webb Space Telescope (JWST) or the Spitzer and GALEX space observatories presently in use.

2. The ISO Surveys Extragalactic deep surveys with ISO were performed in the mid- and far-IR with ISOCAM (Cesarsky et al., 1996) and ISOPHOT (Lemke et al., 1996), respectively. In both wavelength ranges, the steep slopes of the number counts indicate that a rapid decline of the IR emission of galaxies must have taken place from around z ∼1 to z = 0. As shown in Figure 1, ISOCAM could detect galaxies at 15 μm in the LIRG regime up to z ∼1.3, while ISOPHOT was limited to either nearby galaxies or moderately distant ULIRGs such as the one of LIR ∼ 4 × 1012 L at a redshift of z = 1 shown in the Figure 1 (normalized SED of Arp 220). The detection limits of ISOCAM and ISOPHOT are compared to SCUBA and Spitzer in the Figure 2. The deepest ISO surveys reached a completeness limit of 0.1 mJy at 15 μm (plain line) and a depth of S15 ∼ 40 μJy (incomplete, Aussel et al., 1999) in blank fields or a twice deeper completeness level in the central part of nearby galaxy clusters using lensing magnification (Altieri et al., 1999; Metcalfe et al., 2003), and 120 mJy at 170 μm (dotted line). The right axis on the plot shows the conversion of the IR detection limit into a SFR detection limit. Any galaxy forming stars at a rate larger than 30M yr−1 could be detected with ISOCAM up to z ∼ 1, assuming that the measured 15 μm flux density of a galaxy can be used to derive its “bolometric” IR luminosity, i.e. L IR = L(8–1000 μm ) (see Section 5.2). Number counts represent the first scientific result of an extragalactic survey. They can be used to constrain the models. Used alone, they leave some degeneracies

96

D. ELBAZ

Figure 1. Effect of the k-correction on the detection of distant LIRGs and ULIRGs by ISOCAM, ISOPHOT and SCUBA. Only ISOCAM at 15 μm can detect LIRGs up to z ∼ 1, while ISOPHOT and SCUBA are sensitive to distant ULIRGs of a few 1012 L .

Figure 2. Sensitivity limits of ISOCAM (15μm, 0.1 mJy), ISOPHOT (170 μm, 120 mJy), SCUBA (850 μm, 2 mJy) and Spitzer (24 μm, 0.45 mJy and 0.022 mJy corresponding to the expected detection limits of the SWIRE and GOODS Legacy Programs). This figure was generated assuming that distant galaxies SEDs are similar to local ones. We used the library of template SEDs constructed by Chary and Elbaz (2001).

unsolved but at least they can demonstrate whether the distant universe was different from the local one in this wavelength range by comparing them to “no evolution” predictions assuming some spectral energy distributions (SEDs) for the k-correction.

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

97

3. The ISOCAM 6.75-μm Deep Surveys Deep images of blank fields at 6.75- and 15-μm done with ISOCAM provide a different view on the distant universe. While the ISOCAM-LW3 filter, centered at 15μm, can probe dust emission up to z ∼ 2 for luminous objects, the redshift range to probe star formation from dust emission with the ISOCAM-LW2 band, centered at 6.75μm, is limited to the relatively nearby universe due to k-correction. However, the emission of the old stellar component, which peaks in the near IR, being brought to this wavelength range for high-z galaxies, their stellar masses are better constrained from this flux density. Sato et al. (2004) derived stellar masses around M ∼ 1011 M for galaxies with z ∼ 0.2–3 from the correlation between rest-frame near IR luminosity and stellar mass. The stellar mass-to-light ratios were derived from model fit of the set of observed magnitudes depending on the galaxies star formation histories. The contribution of the 6.75-μm selected galaxies to the cosmic stellar mass density of galaxies per unit co-moving volume as a function of redshift were estimated to be comparable to those inferred from observations of UV bright galaxies (see Figure 3). Given the narrow mass ranges, these estimates were obtained as simple summations of the detected sources and should therefore be considered as lower limits only. The faint 6.7 μm galaxies generally had red colour. A comparison with a particular population synthesis model suggests that they have experienced vigorous star formation at high redshifts. The derived large stellar masses for the faint 6.7 μm galaxies also support such star forming events at the past. However, these events can only be detected at longer wavelengths using the 15- μm ISOCAM or 90 and 170-μm ISOPHOT bands with ISO (or SCUBA at 850 μm and now the Spitzer Space Telescope). The next sections are devoted to these surveys and their consequences on our understanding of the history of star formation in the universe.

4. The ISOPHOT Far IR Deep Surveys ISOPHOT was used to survey the sky at 90 and 170μm, mainly (plus a few surveys at 60, 120, 150 and 180 μm, see Juvela et al., 2000; Linden-Voernle et al., 2000; cf. review by Dole, 2002). At 90 μm, the 46 arcsec pixels and FWHM of ISOPHOT represented a significant improvement with respect to IRAS, however due to kcorrection and sensitivity limitations, only the local universe could be probed at such wavelengths. The 170-μm band was more favourable for the detection of distant ULIRGs as shown in the Figure 1 where the peak emission of a ULIRG of 4 × 1012 L and redshifted at z = 1 is shown (normalized SED of Arp 220). A major issue at these wavelengths with a 60-cm telescope is obviously the identification of optical counterparts for the determination of a redshift and the separation of local and distant galaxies.

98

D. ELBAZ

Figure 3. (Left) ISOCAM 6.75-μm image of the SSA13 field (Figure 2 from Sato et al., 2003, 23 h observation). This image reaches an 80% completeness limit of 16 μJy in the central 7 arcmin2 region. North is top and east to the left in J2000. The center is (RA, Dec) = (13h 12m 26s , 42◦ 44 24 .8). The map shows signal-to-noise ratio (S/N) per 0.6 arcsec sub-pixel. The signal is an average, weighted by the inverse of the assigned variance, and the noise is a normalized standard deviation. (Right) Stellar mass density in the universe as a function of redshift (Figure 13 from Sato et al., 2004). The right axis shows densities normalized to the critical density of the universe. The contributions of the 6.75-μm galaxies are shown with solid and double circles for the “combined” (including photometric redshifts) and purely spectroscopic samples, respectively. The double circles are plotted at slightly lower redshifts. The horizontal bars represent the redshift ranges of the bins and the vertical bars show one sigma errors, taking account of Poisson noise and uncertainties in stellar mass and Vmax . Several other estimates are overlaid (see Sato et al., 2004 for references). The empty/solid squares and the diamond are obtained from full integration of a Schechter fit to their respective luminosity or stellar mass function at each redshift bin. The X marks and the empty circles are quasi-fully integrated values with a finite integration range from 10 to 1/20 L ∗ , and 10.5 < log(Mstar [h −2 65 M ]) < 11.6, respectively. The triangles are simply summed values of the detected sources. The two dashed curves are deduced by integrating the star formation rate density in the universe, which is derived from the UV luminosity density as a function of redshift (Cole et al., 2001). The upper curve is an extinction corrected case for E(B − V ) = 0.15, and the lower one has no dust correction.

4.1. SOURCE C OUNTS

AND

COSMIC IR B ACKGROUND

Source counts at 170 μm (e.g. Kawara et al., 1998; Puget et al., 1999; Dole et al., 2001) exhibit a steep slope of α = 3.3 ± 0.6 between 180 and 500 mJy and, like in the mid-IR range, show sources in excess by a factor of 10 compared with noevolution scenario. The brightness fluctuations in the Lockman Hole were used by Matsuhara et al. (2000) to constrain galaxy number counts down to 35 mJy at 90 μm and 60 mJy at 170μm, confirming the existence of a strong evolution down to these flux densities. Using a new data reduction method, (Rodighiero and Franceschini, 2004) extended the previous works of Kawara et al. (1998), Efstathiou et al. (2000) and Linden-Voernle et al. (2000) down to lower flux densities (30 mJy at 90 μm ) and found a clear excess of faint objects with respect to no evolution (see Figure 4a). However, the resolved sources account for less than 10% of the Cosmic

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

99

Figure 4. (a) Differential 90 μm counts d N /d S normalized to the Euclidean law (N ∝ S −2.5 ) (extracted from Rodighiero and Franceschini, 2004). Results compared with those from the preliminary analysis of the ISOPHOT ELAIS survey (Efsthatiou et al., 2000, open circles) and with those from Kawara et al. (2004, filled triangles). The long-dashed line shows the expected contribution of nonevolving spirals as in the model of Franceschini et al. (2001). (b) Fluctuations of the CIB in a power spectrum analysis of the FIRBACK/ELAIS N2 field at 170 μm by Puget and Lagache (2000). Observed power spectrum: diamond; straight continuous line: the best fit cirrus power spectrum; dash line: cirrus power spectrum deduced from Miville-Deschˆenes et al. (2002); continuous curve: detector noise.

Infrared Background at 170 μm, which is expected to be resolved into sources in the 1–10 mJy range. Sources below the detection limit of a survey create fluctuations. If the detection limit does not allow to resolve the sources dominating the CIB intensity, which is the case in the far IR with ISO, characterizing these fluctuations can constrain the spatial correlations of these unresolved sources of cosmological significance. An example of the modeled redshift distribution of the unresolved sources at 170 μm can be found in Figure 12 of Lagache et al. (2003); the sources dominating the CIB fluctuations have a redshift distribution peaking at z ∼ 0.9. After the pioneering work of Herbstmeierer et al. (1998) with ISOPHOT, Lagache and Puget (2000) discovered them at 170 μm in the FIRBACK data, followed by other works at 170 and 90 μm (Matsuhara et al., 2000; Puget and Lagache, 2000; Kiss et al., 2001). Figure 4b shows the CIB fluctuations in the FN2 field by Puget and Lagache (2000), at wavenumbers 0.07 < k < 0.4 arcmin−1 . 4.2. NATURE

OF THE

ISOPHOT GALAXIES

Determining the nature of the far IR galaxies has been a longer process than in the mid-IR, mainly because of the difficulty to find the shorter wavelength counterparts in a large beam. Various techniques have been used to overcome this problem, one of the most successful being the identification using 20-cm radio data (e.g. Ciliegi et al., 1999). Another technique is the far IR multiwavelength approach (Juvela et al., 2000) that helps constraining the position and the SED; it also helps separate

100

D. ELBAZ

the cirrus structures from the extragalactic sources. A variation is to use ISOCAM and ISOPHOT data, like the ELAIS Survey (Rowan-Robinson et al., 2004, see also Oliver in this book). Finally, the Serendipity Survey (Stickel et al., 1998, 2000), by covering large and shallow areas, allows to detect many bright objects easier to follow-up or already known. Far-IR ISO galaxies can be sorted schematically into two populations. First, the low-redshift sources, typically z < 0.3 (e.g., Serjeant et al., 2001; Patris et al., 2002; Kakazu et al., 2002), have moderate IR luminosities, below 1011 L  , and are cold (Stickel et al., 2000). Second, sources at higher redshift, z ∼ 0.3 (Patris et al., 2002) and beyond, z ∼ 0.9 (Chapman et al., 2002) are more luminous, typically L > 1011 L  , and appear to be cold. Serjeant et al. (2001) derived the Luminosity Function at 90 μm, and started to detect an evolution compared to the local IRAS 100 μm sample.

5. The ISOCAM 15-μm Deep Surveys A series of surveys were performed within the Guaranteed Time (IGTES, ISOCAM Guaranteed Time Extragalactic Surveys, Elbaz et al., 1999) and Open Time as summarized in the Table I, where the surveys at 7 (6.75 μm ) and 15 μm are sorted by increasing depth irrespective of wavelength. The major strength of ISOCAM is its spatial resolution (PSF FWHM of 4.5 at 15μm, Okumura, 1998) and sensitivity, which permitted to detect galaxies down to the LIRG regime up to z ∼1.3 in the deepest surveys (Figures 1 and 2), well above confusion. Cosmic rays were a stronger limitation than photon or readout noise by inducing ghost sources when they were not perfectly removed, especially those with long-term transients associated to them. Two techniques were developed to solve this issue, the so-called PRETI (Pattern REcognition Technique for Isocam data, Starck et al., 1999) and LARI (Lari et al., 2001) techniques. A third technique was developed by D´esert et al. (1999), in which the mosaic data were analyzed using the beam switching approach, the Three Beam technique. PRETI consists in a multi-scale wavelet decomposition of the signal, while LARI tries to account for the physical processes taking place in the detectors, including the effect of neighboring pixels. The LARI technique was first applied to the ELAIS surveys (see Oliver et al., in these book) and more recently to the IGTES surveys of the Lockman Hole (Rodighiero et al., 2004; Fadda et al., 2004) and of the Marano field (Elbaz et al., in preparation). The quality of the ISOCAM images is shown with the case of the Marano FIRBACK deep field (30 ×30 ) in the Figure 6. The central part of this field was imaged at a deeper level (UDSF) and analyzed with both techniques. The resulting contours are overlayed on a VLT-FORS2 image of the field, demonstrating the consistency and robustness of both source detection algorithms (Figure 7). Lensing appeared to be a very powerful tool to extend the detection of faint objects by a factor 2–3 (see Metcalfe et al., 2003). However, studies such as the

101

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

TABLE I Table of the ISOCAM extragalactic surveys sorted by increasing depth. Name

λ (μm)

Area (’2 )

ELAIS N11 ELAIS N21 ELAIS N31 ELAIS S11 ELAIS S21 Lockman Shallow2,3

15 7;15 7;15 7;15 7;15 15

Comet Field4 CFRS14+525 CFRS03+006 Lockman Deep2,7 Marano DSF2 A3708

12 7;15 7;15 7;15 7;15 7;15

9612 9612;9612 4752;3168 6336;14256 432;432 1944 (80% compl.) 360 100;100 100;100 500;500 900;900 31.3;31.3 (80% compl.) 85;90 89;90 20.5;20.5 (80% compl.) 10;27 28;28 5.3;5.3 (80% compl.) 9 16 7(80% compl.)

Marano UDSR2 Marano UDSF2 A22188

7;15 7;15 7;15

HDFN+FF2,9 HDFS2,10 A239011,8

7;15 7;15 7;15

Lockman PGPQ12 SSA1313

7 7

Int. (min) 0.7;0.7 0.7;0.7 0.7;0.7 0.7;0.7 0.7;0.7 3 10 18;11 6;22 18;11 15.4;15.4 42;42 120;114 114 84;84 116;135 168;168 432;432 744 1264

depth (mJy)

No. of objects

1;0.7 1;0.7 1;0.7 1;0.7 1;0.7 0.25 0.45 0.5 0.3;0.4 0.5;0.3 0.3;0.4 0.19;0.32 0.052;0.21(u) 0.080;0.293(u) 0.18;0.14 0.08;0.14 0.054;0.121(u) 0.079;0.167(u) 0.05;0.1 0.05;0.1 0.038;0.050(u) 0.052;0.092(u) 0.034 0.006 0.016

490 628;566 189;131 304;317 40;43 457 260 37 23;41 – 166 180 4;20 –;142 115;137 18;46 7;44 16;63 10;28 15 65

Notes. Col. (1) Survey name. Col. (2) wavelength of the survey with “7” for the LW2 filter centered at 6.75 μm and covering 5–8.5 μm, and “15” for the LW3 filter centered 15 μm and covering 12–18 μm. Col. (3) total area in square arcmin. Col. (4) integration time per sky position. Col. (5) depth (and 80% completeness limit when indicated). Col. (6) number of objects detected above this depth. References: (1) Oliver et al., 2000; Rowan-Robinson et al., 2004, (2) Elbaz et al., 1999 and in preparation, (3) Rodighiero et al., 2004, (4) Clements et al., 1999, (5) Flores et al., 1999, (6) Flores (private communication), (7) Fadda et al., 2004, (8) Metcalfe et al., 2003, (9) Aussel et al., 1999; Goldschmidt et al., 1997, (10) Oliver et al., 2002, (11) Altieri et al., 1999, (12) Taniguchi et al., 1997, (13) Sato et al., 2003, 2004. Superscript (u) indcates that the depth of the deep surveys in the field of lensing clusters does not include the correction for lensing amplification.

relative clustering of IR galaxies versus field galaxies are better done in larger homogeneous fields, where the role of cosmic variance can also be quantified. As we will see below, the environment of distant LIRGs seems to play a major role in triggering their star formation activity.

102

D. ELBAZ

Figure 5. ISOCAM 15- μm differential counts, with 68% error bars. The counts are normalized to a Euclidean distribution of non-evolving sources, which would have a slope of α = −2.5 in such a universe. This plot is an extension of the Figure 2 from Elbaz et al. (1999) combining the data points from Elbaz et al. (1999), for the ISOCAM Guaranteed Time Extragalactic Surveys (IGTES, filled and empty dots for sources below and above 1 mJy respectively, see text), Gruppioni et al. (2003, bold empty circles) for the ELAIS-S1 field, Rodighiero et al. (2004, large open stars) and Fadda et al. (2004, large open squares), for the Shallow and Deep Surveys of the Lockman Hole, respectively (analysis of the IGTES data using the “Lari” technique, see Section 5). The hatched area materializes the range of possible expectations from models assuming no evolution and normalized to the 15 μm local luminosity function (LLF) from Fang et al. (1998) and the template SED of M51 (with an assumed uncertainty of ±20%; this 15 μm LLF was recently confirmed over a larger sample of local galaxies by H. Aussel, private communication).

5.1. SOURCE C OUNTS

AND

COSMIC IR B ACKGROUND

The source counts at 15 μm exhibit a strong excess of faint sources below S15 ∼ 2 mJy. This excess is usually defined by comparison with model predictions assuming that galaxies behaved similarly in the distant universe as they do today. Such “no evolution” behavior is represented by a shaded area in the Figure 5 (see figure caption). Galaxies above this flux density do fall within this region, as illustrated by the data points from the ELAIS-S1 field (Gruppioni et al., 2003, see also Oliver et al., in this book). In Figure 5, we have separated the data points from the IGTES (Elbaz et al., 1999) between those below and above S15 = 1 mJy, with filled and open dots, respectively. Data above this flux density from Elbaz et al. (1999) appear to be inconsistent with those derived from ELAIS-S1. Most of those points were derived from the Shallow Survey of the Lockman Hole within the IGTES, which suffered from having less redundant observations of a given sky pixel. At that time, ISOCAM data reduction methods were not optimized for such surveys, but since then, they have been improved to better deal with such shallow surveys. Among the techniques that are discussed below, the “Lari” technique is particularly suitable for

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

103

Figure 6. ISOCAM 15-μm image of the Marano FIRBACK Deep Survey (DSF) from the IGTES (Elbaz et al., 1999, and in preparation). Data reduction with PRETI.

such low redundancy surveys (and even more for the very shallow ELAIS surveys, see Oliver et al., in this book) and a recent analysis of the Lockman Hole Deep (Fadda et al., 2004, large open squares) and Shallow (Rodighiero et al., 2004, large open stars) surveys from the IGTES provided new number counts at these flux densities perfectly consistent with those derived from ELAIS-S1 by Gruppioni et al. (2003) also using the same “Lari” technique. Note that the models designed to fit the ISOCAM number counts were constrained by the Elbaz et al. (1999) results, hence overproduce the number of sources above S15 ∼ 2 mJy. As a natural result, they have also overpredicted the number of sources detected in the high flux density regime at 24 μm with Spitzer (see Section 7 and Papovich et al., 2004; Chary et al., 2004). Above the Earth’s atmosphere, the 15-μm light is strongly dominated by the zodiacal emission from interplanetary dust and it has not yet been possible to make a direct measurement of the 15-μm background, or EBL. Individual galaxies contribute to this background and a lower limit to the 15-μm EBL can be obtained by adding up the fluxes of all ISOCAM galaxies detected per unit area down to a given flux limit. The resulting value is called the 15-μm integrated galaxy light (IGL). As in Elbaz et al. (2002), the differential number counts can be converted into a differential contribution to the 15 μm IGL as a function of flux density, estimated

104

D. ELBAZ

Figure 7. ISOCAM 15-μm contours overlayed on the VLT-FORS2 image (7 ×7 , R-band) of the Marano FIRBACK Ultra Deep Survey (UDSF). The LARI (light contours, yellow on screen, grey on paper) and PRETI (dark contours, red on screen) detect the same objects.

from the following formula:   dIGL dN S15 = × × ν15 dS dS 1020

(1)

where d N (sr−1 ) is the surface density of sources with a flux density Sν [15 μm]= S15 (mJy) over a flux density bin d S (mJy) (1 mJy = 10−20 nW m−2 Hz−1 ) and ν15 (Hz) is the frequency of the 15 μm photons. Below S15 ∼ 3 mJy, about 600 galaxies were used to produce the points with errors bars in Figure 8a. Figure 8b shows the 15-μm IGL as a function of depth. It corresponds to the integral of Figure 8a, where the data below 3 mJy were fitted with a polynomial of degree 3 and the 1-σ error bars on dIGL/d S were obtained from the polynomial fit to the upper and lower limits of the data points. The 15-μm IGL does not converge above a sensitivity limit of S15 ∼ 50 μJy, but the flattening of the curve below S15 ∼ 0.4 mJy suggests that most of the 15 μm EBL should arise from the galaxies already unveiled by ISOCAM. Franceschini et al. (2001) and Chary and Elbaz (2001), developed models which reproduce the number counts from ISOCAM at 15 μm, from ISOPHOT at 90 and 170 μm and from SCUBA at 850 μm, as well as the shape of the CIRB from 100

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

(a)

105

(b)

Figure 8. (a) Differential contribution to the 15 μm Integrated Galaxy Light as a function of flux density and AB magnitude. The plain line is a fit to the data: Abell 2390 (Altieri et al., 1999), the ISOCAM Guaranteed Time Extragalactic Surveys (IGTES, Elbaz et al., 1999), the European Large Area Infrared Survey (ELAIS, Serjeant et al., 2000) and the IRAS all sky survey (Rush et al., 1993). (b) Contribution of ISOCAM galaxies to the 15 μm extragalactic background light (EBL), i.e. 15 μm Integrated Galaxy Light (IGL), as a function of sensitivity or AB magnitude (AB = −2.5 log(SmJy ) + 16.4). The plain line is the integral of the fit to dIGL/d S (a). The dashed lines correspond to 1-σ error bars obtained by fitting the 1-σ upper and lower limits of dIGL/d S.

to 1000 μm. These models consistently predict a 15-μm EBL of: EBLmodels (15 μm) ∼ 3.3 nW m−2 sr−1

(2)

If this prediction from the models is correct then about 73 ± 15% of the 15 μm EBL is resolved into individual galaxies by the ISOCAM surveys. This result is consistent with the upper limit on the 15 μm EBL estimated by Stanev and Franceschini (1998) of EBLmax (15 μm) ∼ 5 nW m−2 sr−1

(3)

This upper limit was computed from the 1997 γ -ray outburst of the blazar Mkn 501 (z = 0.034) as a result of the opacity of mid-IR photons to γ -ray photons, which annihilate with them through e+ e− pair production. It was since confirmed by Renault et al. (2001), who found an upper limit of 4.7 nW m−2 s r−1 from 5 to 15 μm. Nearly all the IGL is produced by sources fainter than 3 mJy (94 %) and about 70 % by sources fainter than 0.5 mJy. This means that the nature and redshift distribution of the galaxies producing the bulk of the 15 μm IGL can be determined by studying these faint galaxies only. 5.2. MID-IR

AS A

SFR I NDICATOR

When normalized to the 7.7- μm PAH (polycyclic aromatic hydrocarbon) broad emission line, the spectra of different galaxies exhibit very different 15, 25, 60 or

106

D. ELBAZ

(a)

(b)

(c)

(d)

Figure 9. IR luminosity correlations for local galaxies (from Elbaz et al., 2002). (a) ISOCAMLW3 (15 μm) versus ISOCAM-LW2 (6.75 μm) luminosities (ν L ν ) (56 galaxies). (b) ISOCAMLW3 (15 μm) versus IRAS-12 μm luminosities (45 galaxies). (c) L IR [8–1000 μm] versus ISOCAMLW3 (15 μm) luminosity (120 galaxies). (d) L IR [8–1000 μm] versus LW2-6.75 μm luminosities (91 galaxies). Filled dots: galaxies from the ISOCAM guaranteed time (47 galaxies including the open squares). Open dots: 40 galaxies from Rigopoulou et al. (1999). Empty triangles: four galaxies from Tran et al. (2001). Galaxies below L IR ∼ 1010 L present a flatter slope and have L IR /L B < 1.

100 μm over 7.7 μm ratio. This was often interpreted as an indication that measuring the monochromatic luminosity of a galaxy at one mid-IR wavelength was useless to determine its bolometric IR luminosity, LIR = L(8–1000 μm ). However, the variation of the far over mid-IR ratio is correlated with LIR and local galaxies do exhibit a strong correlation of their mid- and far-IR luminosities (Figure 9). These correlations can be used to construct a family of template SEDs or correlations from which the LIR , and therefore SFR, of a galaxy can be derived from its mid-IR luminosity (Chary and Elbaz, 2001; Elbaz et al., 2002). The LIR derived from this technique are consistent with those derived by the radio-far IR correlation, when radio-mid–far IR data exist (Elbaz et al., 2002; Garrett 2002; Gruppioni et al., 2003). In the Figure 10, we have reproduced the plot from Elbaz et al. (2002) complemented with galaxies detected within the ELAIS survey (Rowan-Robinson et al., 2004). Except at low luminosities where the contribution of cirrus to the IR luminosity becomes non negligible, the 1.4 GHz and 15 μm rest-frame luminosities are correlated up to z ∼ 1 and therefore predict very consistent total IR luminosities.

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

107

Figure 10. 15 μm versus radio continuum (1.4 GHz) rest-frame luminosities. Small filled dots: sample of 109 local galaxies from ISOCAM and NVSS. Filled dots with error bars: 17 HDFN galaxies (z ∼ 0.7, radio from VLA or WSRT). Open dots with error bars: 7 CFRS-14 galaxies (z ∼ 0.7, Flores et al., 1999, radio from VLA). Open diamonds: 137 ELAIS galaxies (z ∼ 0–0.4).

A similar result was later on obtained using the MIPS instrument onboard the Spitzer Space Observatory at 24 μm (Appleton et al., 2004). Several studies compared the SFR derived from the IR luminosity with the optical SFR derived from the Hα emission line (Rigopoulou et al., 2000; Cardiel et al., 2003; Flores et al., 2004; Liang et al., 2004). Rigopoulou et al. (2000) found a large excess of SFR(IR) versus SFR(Hα) even after correcting the latter for extinction. The Balmer decrement was only measured for limited number of objects in the sample and the extinction correction was derived from broadband photometry, which suffers from strong uncertainties in particular due to the degeneracy between age, metallicity and extinction. However, Cardiel et al. (2003) confirmed the direct measure of the SFR(IR) excess using the combination of NIRSPEC and LRIS, for the distant galaxies, and the Echelle Spectrograph and Imager (ESI), for the closer ones, at the Keck telescope. Using high resolution VLT-FORS2 spectra, Flores et al. (2004) and Liang et al. (2004) were able to measure directly the Balmer decrement (using Hα/Hβ or Hβ/Hγ ) and to subtract the underlying nebular emission lines with a fit of the stellar continuum. Although the SFR(IR) exceeds the estimate from emission lines for the most active objects, the data present a clear correlation between SFR(IR) and SFR(Hα) which suggests that the star formation regions responsible for the IR luminosity of distant LIRGs are not completely obscured. This correlations also confirms that the mid-IR is indeed a good SFR estimator. 5.3. N ATURE

OF THE

ISOCAM G ALAXIES

The Hubble Deep Field North and its Flanking Fields (HDFN+FF) provides the best coverage in spectroscopic redshifts and deep optical images for an ISOCAM deep

108

D. ELBAZ

survey. We used the revised version of the Aussel et al. (1999) catalog for which 85% (71%) of the 40 (86) galaxies above 100 μJy (30 μJy) have a spectroscopic redshift to determine the average properties of ISOCAM galaxies summarized in the Figures 11 and 12. Their optical counterparts are relatively bright and their median-mean redshift is close to z ∼ 0.8 (Figure 11). Note the redshift peaks in which the ISOCAM galaxies are located, leaving wide empty spaces in between. Most ISOCAM galaxies are located within large-scale structures, here mainly those at z = 0.848 and z = 1.017, which might be galaxy clusters in formation where galaxy–galaxy interactions are amplified (Elbaz and Cesarsky, 2003, see discussion below). Thanks to the deepest soft to hard X-ray survey ever performed with Chandra in the HDFN, it is possible to pinpoint active galactic nuclei (AGNs) in this field including those affected by dust extinction. Only five sources were classified as

Figure 11. (Left) Histogram of the R(AB) magnitudes of the 15-μm ISOCAM galaxies detected in the HDFN+FF (revised catalog of Aussel et al., down to ∼30 μJy. (Right) Redshift distribution of the HDFN+FF ISOCAM galaxies.

Figure 12. (Left) Distribution of the log(LIR (8–1000 μm)/L ) of the HDFN+FF ISOCAM galaxies derived from their 15-μm flux densities. (Right) Stellar mass as a function of redshift of field HDFN proper galaxies (dark dots) and of ISOCAM galaxies (open circles). All stellar masses were derived using multi-bands SED modelling by Dickinson et al. (2003).

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

109

AGN dominated on the basis of their X-ray properties (Fadda et al., 2002). Hence, unless a large number of AGNs are so dust obscured that they were even missed with the 2 Megaseconds Chandra survey, the majority of ISOCAM galaxies are powered by star formation. This result is consistent with observations of local galaxies which indicate that only the upper luminosity range of ULIRGs are dominated by an AGN (Tran et al., 2001). Using the mid-far IR correlations (Chary and Elbaz, 2001; Elbaz et al., 2002, see also Section 5.2), the LIR distribution of the HDFN mid-IR sources is plotted in Figure 12a. Most of them belong to the LIRG and ULIRG regime, although when including flux densities below completeness down to 30 μJy, one finds also intermediate luminosities. Finally, their stellar masses are among the largest in their redshift range, when compared to the stellar mass estimates by Dickinson et al. (2003) in the HDFN (Figure 12b). In the local universe, both LIRGs and ULIRGs exhibit the typical morphology of major mergers, i.e. mergers of approximately equal mass (Sanders and Mirabel, 1996; Sanders et al., 1999). In the case of LIRGs, the merging galaxies show a larger separation than for ULIRGs, which are mostly in the late phase of the merger. Figure 13 (Elbaz and Moy, 2004) presents the HST-ACS morphology of a sample of z ∼ 0.7 LIRGs in the GOODSN field (extended HDFN). Less than half of these galaxies clone the morphology of local LIRGs, which implies that the physical processes switching on the star formation activity in distant LIRGs might be different than for local ones. The gas mass fraction of younger galaxies

Figure 13. HST-ACS images of LIR galaxies with 11 ≤ log(L IR /L  ) ≤ 12 (LIRGs) and z ∼ 0.7. The double-headed arrow indicates the physical size of 50 kpc. The IR luminosity increases from left to right and from top to bottom.

110

D. ELBAZ

being larger, other types of interactions might generate a LIRG phase in the distant universe, such as minor mergers or even passing-by galaxies producing a tidal effect. The fact that such interactions are more frequent than major mergers could also explain the importance of the LIRG phase for galaxies in general and also the possibility for a galaxy to experience several intense bursts in its lifetime. The appearance of this phase of violent star formation could be facilitated during the formation of groups or clusters of galaxies. A striking example of this is given by the large fraction of LIRGs detected in the distant galaxy cluster J1888.16CL, located at a redshit of z = 0.56 (Duc et al., 2004). Among the 27 objects for which spectra were obtained, six of them belong to the cluster while an extra pair with slightly higher redshifts may lie in infalling groups. All eight galaxies exhibit weak emission lines in their optical spectra, typical of dust enshrouded star forming galaxies, none of these lines being broad enough to indicate the presence of type I AGNs. In this relatively young galaxy cluster, the mechanism that may be triggering SFRs between 20 and 120 M yr−1 for at least eight objects of the cluster could well be tidal collisions within sub-structures or infalling groups. On a more local scale, the mid-IR emission of the Abell 1689 (z = 0.181) galaxies exhibits an excess of the B-[15] colour with respect to richer and closer galaxy clusters, such as Coma and Virgo, which suggests the presence of a mid-IR equivalent to the Butcher–Oemler effect, i.e. the star formation activity of galaxies as reflected by their mid-IR emission increases with increasing redshift (Fadda et al., 2000). The high fraction of blue galaxies initially reported for this cluster by Butcher and Oemler (1984) was later on confirmed by Duc et al. (2002), who also found that the actual SFR for these galaxies was on average 10 times larger than the one derived from the [O II] emission line implying that 90% of the star formation activity taking place in this cluster was hidden by dust. LIRGs could then be a tracer of large-scale structures in formation as suggested by their redshift distribution (Figure 11b, see also Elbaz and Cesarsky, 2003). In order to test this hypothesis (Moy and Elbaz in preparation) compared the fraction of ISOCAM galaxies (above the completeness limit of 0.1 mJy) found in “redshift peaks” to the one obtained when randomly selecting field sources of equal optical and K-band magnitudes in the same redshift range. Redshift peaks of different strengths, measured as N-σ , were defined by smoothing the field galaxies redshift distribution by 15,000 km/s (Figure 14a) and measuring the peaks N-σ above the smoothed distribution. The Monte-Carlo samples of field galaxies to be compared to the redshift distribution of the ISOCAM galaxies were selected within the real redshift distribution and not the smoothed one. The distribution of the Monte-Carlo simulations are shown with error bars at the 68 and 90% level. When considering the whole ISOCAM catalog, i.e. including faint sources below completeness, it appears that ISOCAM galaxies are more clustered than field galaxies. Less than 32% of the simulations present a comparable clustering than the whole ISOCAM sample which contains a large fraction of non LIRGs. Strikingly, when only selecting the brightest galaxies above the completeness limit of 0.1 mJy, one finds that they are LIRGs

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

(a)

111 (b)

Figure 14. (a) Redshift distribution of field galaxies in the GOODSN field. The continuous line is the smoothed distribution with a window of 15,000 km/s. (b) Differential fraction of sources within redshift peaks (see definition in the text) stronger than N-σ . Small filled squares: fraction of sources in peaks for the whole optical catalog of 930 field galaxies (Wirth et al., 2004). Small open squares: median of the fraction of sources in redshift peaks for a series of Monte-Carlo simulations of a sub-sample of the field galaxies corresponding to the same number of galaxies as in the ISOCAM catalog and within the same range of redshifts and optical-near IR magnitudes. Error bars contain 68 and 90% of the simulations. Large Open Circles: total sample of ISOCAM galaxies (75 sources with spectroscopic redshifts and optical-near IR magnitudes). Large open squares: sub-sample of ISOCAM galaxies above the completeness limit of 0.1 mJy (41 sources).

and ULIRGs which fall in the densest redshift peaks, above 6-σ . The probability to randomly select a sample of galaxies from the field (with equal magnitudes and redshift range) in the redshift peaks of 6-σ and above is less than 1%. As a result, mid-IR surveys are very efficient in selecting over-dense regions in the universe, which in return are very efficient in producing a LIRG. In contrast, mid-IR selected galaxies are locally less clustered than field galaxies (Gonzalez-Solares et al., 2004), which could be a natural result of the fact that only the less clustered galaxies still produce IR luminous phases while more clustered galaxies lived their IR luminous phase in the past (see also Elbaz and Cesarsky, 2003). Finally, one question remains to be addressed about distant LIRGs: how long does this starburst phase last and how much stellar mass is produced during that time? Marcillac et al. (2004) used a Bayesian approach and simulated 200,000 virtual high-resolution spectra with the Bruzual and Charlot (2003) code to determine the recent star formation history of distant LIRGs as well as their stellar masses. These ISOCAM galaxies were observed using the VLT-FORS2 (λ/λ ∼ 2000 in the rest-frame) in three different fields. A prototypical LIRG at z ∼ 0.7 is found to have a stellar mass of ∼5×1010 M and to produce about 10% of this stellar mass within about 108 years during the burst. A remarkable result of this study is that the position of distant LIRGs in a diagram showing the value of the H8 Balmer ˚ break signs the absorption line equivalent width versus the strength of the 4000 A presence of a burst of star formation within these galaxies, with an intensity of about 50M  yr−1 as also derived from their mid-IR emission. This result supports the

112

D. ELBAZ

idea that distant LIRGs are not completely opaque to optical light and that one can learn something about their star formation history based on their optical spectra. Liang et al. (2004) compared the gas metallicity of the same sample of objects than Marcillac et al. (2004) with local galaxies of similar absolute magnitudes in the B band. Even accounting for an evolution in the B-band luminosity, the distant LIRGs turn out to be about twice less metal rich. This result suggests that between z ∼1 and today, LIRGs do produce a large fraction of the metals located in their host galaxies in agreement with the strong evolution of the cosmic star formation history found by the models fitting the ISO source counts. 6. Cosmic Evolution, Star-Formation Rate History The combination of surveys at different wavelengths, from ISOCAM, ISOPHOT and SCUBA, together with the the shape and intensity of the CIRB, was used by several authors to constrain the parameters of their backward evolution models assuming a combination of luminosity and density evolution as a function of redshift of the IR luminosity function at 15 or 60 μm : Roche and Eales (1999), Tan et al. (1999), Devriendt and Guiderdoni (2000), Dole et al. (2000), Chary and Elbaz (2001), Franceschini et al. (2001, 2003), Malkan and Stecker (2001), Pearson (2001), Rowan-Robinson (2001), King and Rowan-Robinson (2003), Takeuchi et al. (2001), Xu et al. (2001, 2003), Balland et al. (2002), Lagache et al. (2003), Totani and Takeuchi (2002), Wang (2002). Being limited by the sensitivity of the extragalactic surveys, the major output of these models was to show that LIRGs and ULIRGs were much more common in the past than they are today. Chary and Elbaz (2001) derived that the co-moving IR luminosity due to LIRGs was about 70 times larger at z ∼1 than it is today (Figure 15). Quite logically, it resulted that the contribution of LIRGs to the cosmic star formation history was so large in the past that it dominated the integrated star formation that galaxies experienced in the past. Hence, LIRGs should now be considered not as a type of galaxies but instead as a common phase of intense star formation that any galaxy may have experienced. Elbaz et al. (2002) derived the contribution to the peak of the CIRB at 140 μm from the population of LIRGs around z ∼ 0.8 detected in the ISOCAM 15 μm deep surveys and found that these objects alone can explain more than two thirds of the peak and integrated intensities of the CIRB (see Figure 16). Hence the CIRB is the signature of the strong redshift evolution of LIRGs and the fossil record of star formation which took place in such burst phases. 7. From ISO to Spitzer, Herschel, the JWST and ALMA On 23 August, 2003, NASA’s Spitzer space telescope (formerly SIRTF) was launched. Among its first results came the source counts at 24 μm down to ∼20 μJy

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

113

Figure 15. Star-formation rate from Chary and Elbaz (2001). Upper panel: min and max range from their model, and observed UV/opt data; dots represent Xu et al. (2001) model. Lower panel: 3 different evolution scenarios from their model and data corrected for extinction. Line: pure luminosity; upper dash: pure density; lower dash: luminosity+density.

which confirmed what ISO deep surveys already saw: a strong excess of faint sources indicating a rapid redshift evolution of IR luminous galaxies. When compared to models developed to fit the ISO counts, the faint end of the Spitzer counts are perfectly fitted as shown in the Figure 17 reproduced from Chary et al. (2004). On the high flux density range, around 1 mJy and above, less galaxies are found than predicted by those models (see also Papovich et al., 2004). This is partly, if not integrally, due to the fact that even ISOCAM-15- μm number counts were initially overestimated above S15 ∼ 1 mJy as discussed in Section 5.1, since the data reduction techniques were not optimized for the surveys with little redundancy over a given sky pixel. Although some refinement of the template SEDs used in the models might be considered (as suggested by Lagache et al., 2004), the 24-μm number

114

D. ELBAZ

Figure 16. Integrated Galaxy Light (IGL, filled dots) and Extragalactic Background Light (EBL, open squares, grey area) from the UV to sub-millimeter (from Elbaz et al., 2002). EBL measurements from COBE: 200–1500 μm EBL from COBE-FIRAS (grey area, Lagache et al., 1999), 1.25, 2.2, 3.5, 100, 140 μm EBL from COBE-DIRBE (open squares). IGL in the U,B,V,I,J,H,K bands from Madau and Pozzetti (2000). The upper end of the arrows indicate the revised values suggested by Bernstein et al. (2001, factor two higher). 6.75 μm (ISOCAM-LW2 filter) IGL from Altieri et al. (1999, filled dot). Hatched upper limit from Mkn 501 (Stanev and Franceschini, 1998). The ISOCAM 15 μm IGL (2.4 ± 0.5 nW m−2 s r−1 ) is marked with a star surrounded by a circle. The other star surrounded by a circle is the prediction of the contribution of the 15- μm sources to 140 μm (Elbaz et al., 2002).

counts appear to be perfectly consistent with the up-to-date 15-μm counts. Moreover, the high flux density range does not strongly affect the conclusions of the models based on previous 15-μm counts since bright objects do not contribute significantly to the CIRB and to the cosmic density of star formation. Hence these refinements are not strongly affecting the conclusions derived on galaxy formation and evolution based on the ISO deep surveys and summarized in the previous sections (see also Dole et al., 2004 for the Spitzer counts in the far IR). Another hint on the consistency of Spitzer MIPS-24-μm surveys with ISOCAM15 μm is given by the comparison of the images themselves. Galaxies detected at 15 and 24 μm are clearly visible in both images in Figure 18, although several 24-μm sources do not have a 15-μm counterpart. This results from the combination of the better sensitivity of MIPS, by a factor 2 or slightly more for the deepest surveys (galaxies are detected down to S15 ∼ 40 μJy in the HDFN, see Aussel et al., 1999, without lensing magnification), and of the k-correction. For galaxies above z ∼ 1, the PAH bump centered on the PAH feature at 7.7-μm starts to exit to 15-μm broadband while it remains within the MIPS-24-μm band up to z ∼ 2. The combination of ISOCAM and MIPS can be used to test whether the library of template SEDs that were used to derive “bolometric” IR luminosities from 8 to 1000 μm for the 15-μm galaxies are correct at least in the mid-IR range. One of the most important test is to check whether the 7.7-μm PAH bump is still present at z ∼ 1

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

115

Figure 17. Completeness corrected galaxy counts in the MIPS 24- μm channel from Spitzer observations of the ELAIS-N1 field from Chary et al., (2004). The error bars reflect the Poissonian uncertainty. The horizontal bars represent the minimum and maximum flux density in that bin. The lines show four models for 24- μm counts: King and Rowan-Robinson (2003, KRR), Xu et al. (2001, Xu), Chary and Elbaz (2001, CE), Lagache et al. (2003, LDP). The symbols are plotted at the average of the flux densities of the detected sources in that bin for the data while the lines are plotted at the countsweighted flux average for the models. The lower plot in the figure shows the histogram of the actual number of sources detected in each flux bin without any completeness correction.

Figure 18. ISOCAM 15-μm image (left) of the Ultra-Deep Survey in the Marano FIRBACK field (depth 140 μJy, 80% completeness) versus Spitzer MIPS-24 μm image (right; depth 110 μJy, 80% completeness). The crosses identify 16 galaxies detected at 15 and 24 μm for which VLT-FORS2 spectra were obtained and which SEDs were fitted in Elbaz et al. (2004).

and whether the 24 over 15-μm flux density ratio is consistent with the template SEDs used by the models from which star formation histories were derived. The template SEDs designed by Chary and Elbaz (2001) provide a very good fit to the combination of both mid-IR values for a sample of 16 galaxies detected by

116

D. ELBAZ

ISOCAM and MIPS (crosses on the Figure 18). The L IR (8–1000 μm ) derived from either the 15- or 24- μm luminosities or the combination of both to constrain the SED fit present a 1-σ dispersion of only 20% (Elbaz et al., 2004). For galaxies located around z ∼ 1, the relative 15 and 24- μm luminosities clearly suggest the presence of a bump at 7.7 μm as observed in nearby galaxies and due to PAHs. Many questions remain unsolved that will be addressed by future missions, staring with Spitzer. Only when a fair sample of redshifts will have been determined for the distant LIRGs will we be able to definitely ascertain the redshift evolution of the IR luminosity function. Already for the brightest part of it, campaigns of redshift measurements have started, on ELAIS fields for the nearby objects, and on the Marano field with VIMOS and the Lockman Hole with DEIMOS for more distant objects. The fields selected for Spitzer legacy and Guaranteed time surveys were also carefully selected to be covered at all wavelengths and followed spectroscopically, so that this issue should be addressed in the very near future. Due to confusion and sensitivity limits, direct observations in the far IR will not reach the same depth than mid-IR ones until the launch of Herschel scheduled for 2007. The direct access to the far IR distant universe with Herschel will certainly bring major information on galaxy formation, together with the James Webb Space Telescope (JWST) up to 30 μm and the Atacama Large Millimeter Array (ALMA), which will bring an improved spatial resolution for the z ∼ 2 and above universe. Among the questions to be solved, we do not resist to the temptation of listing some of our favorite ones: what is the connection of large-scale structure formation with LIRG phases in galaxies? Can we probe the formation of distant clusters by the detection of the epoch when galaxies formed stars in such intense starbursts, producing galactic winds and enriching the intra-cluster medium with metals? Have we really resolved the bulk of the hard X-ray background, which peaks around 30 keV, and not left unknown some deeply buried AGNs which could make a larger than 20% fraction of the mid-IR light from distant LIRGs? What is the future of a distant LIRG, is it producing stars in a future bulge or disk? How uncertain is the interpolation to lower luminosities than those observed done in the models used to derive cosmic star formation history scenarios? This is of course only an example of a vast series of questions which indicate that the field opened for a large part by ISO has a long life to come.

References Adelberger, K. L. and Steidel, C. C.: 2000, ApJ 544, 218. Altieri et al.: 1999, A&A 343, L65. Appleton, P. N., Fadda, D. T., Marleau, F. R., et al.: 2004, ApJS 154, 147. Aussel, H., Cesarsky, C. J., Elbaz, D., and Starck, J. L.: 1999, A&A 342, 313. Balland, C., Devriendt, J., Silk, J.: 2003, MNRAS 343, 107. Barger, A. J., Cowie, L. L., Sanders, D. B., et al.: 1998, Nature 394, 248.

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

117

Bernstein, R. A., Freedman, W. L., and Madore, B. F.: 2002, ApJ 571, 56. Bruzual, G. and Charlot, S.: 2003, MNRAS 344, 1000. Butcher, H. and Oemler, A.: 1984, ApJ 285, 426. Cardiel, N., Elbaz, D., Schiavon, R. P., et al.: 2003, ApJ 584, 76. Cesarsky, C., Abergel, A., Agn`ese, P., et al.: 1996, A&A 315, L32. Chapman, S. C., Smail, I., Ivison, R., et al.: 2002, ApJ 573, 66. Chapman, S. C., Blain, A. W., Ivison, R. J., and Smail, I.: 2003, Nature 422, 695. Chary, R. R. and Elbaz, D.: 2001, ApJ 556, 562. Chary, R. R., Casertano, S., Dickinson, M. E., et al.: 2004, ApJS 154, 80. Ciliegi, P., McMahon, R. G., Miley, G., et al.: 1999, MNRAS 302, 222. Clements, D., et al.: 1999, A&A 346, 383. Cohen, J. G., Hogg, D. W., Blandford, R., et al.: 2000, ApJ 538, 29. Cole, S., et al.: 2001, MNRAS 326, 255. D´esert, F.-X., Puget, J.-L., Clements, D., et al.: 1999, A&A 342, 363. Devriendt, J. and Guiderdoni, B.: 2000, A&A 363, 851. Dickinson, M., Papovich, C., Ferguson, H. C., and Buda´ari, T.: 2003, ApJ 587, 25. Dole, H., Gispert, R., Lagache, G., et al.: 2000, Springer Lecture Notes in Ph., v 548, 54 (astroph/0002283). Dole, H., Gispert, R., Lagache, G., et al.: 2001, A&A 372, 364. Dole, H.: 2002, Proceedings of the ESA SP-511 conference in C. Gry, et al.: (eds.) “Exploiting the ISO Data Archive – Infrared Astronomy in the Internet Age”, (astro-ph/0211310). Dole, H., Le Floc’h, E., Perez-Gonzalez, P. G., et al.: 2004, ApJS 154, 87. Duc, P.-A., Poggianti, B., Fadda, D., et al.: 2002, A&A 382, 60. Duc, P.-A., Fadda, D., Poggianti, B., et al.: 2004, Proceedings of the IAU Colloquium 195, Outskirts of Galaxy Clusters: Intense life in the suburbs, A. Diaferio Ed. (astro-ph/0404183). Efstathiou, A., Oliver, S., Rowan-Robinson, M., et al.: 2000, MNRAS 319, 1169. Elbaz, D., Cesarsky, C. J., Fadda, D., et al.: 1999, A&A 351, L37. Elbaz, D., Cesarsky, C. J., Chanial, P., et al.: 2002, A&A 384, 848. Elbaz, D. and Cesarsky, C. J.: 2003, Science 300, 270. Elbaz, D. and Moy, E.: 2004, Proceedgins of the “Multiwavelength Cosmology” Conference, held on Mykonos Island, Greece, M. Plionis Ed., Kluwer Academic Publishers, p.173 (astro-ph/0401617). Elbaz, D., Dole, H., Le F’loch, E., and Marcillac, D.: 2004 (submitted). Fadda, D., Elbaz, D., Duc, P.-A., et al.: 2000, A&A 361, 827. Fadda, D., Flores, H., Hasinger, G., et al.: 2002, A&A 383, 838. Fadda, D., Lari, G., Rodighiero, G., et al.: 2004, A&A (in press) (astro-ph/0407649). Fang, F., Shupe, D., Xu, C., and Hacking, P.: 1998, ApJ 500, 693. Fixsen, D. J., Dwek, E., Mather, J. C., Bennett, C. L., and Shafer, R. A.: 1998, ApJ 508, 123. Flores, H., Hammer, F., Thuan, T. X., et al.: 1999, ApJ 517, 148. Flores, H., Hammer, F., Elbaz, D., et al.: 2004, A&A 415, 885. Franceschini, A., Aussel, H., Cesarsky, C. J., Elbaz, D., and Fadda, D.: 2001, A&A 378, 1. Franceschini, A., Berta, S., Rigopoulou, D., et al.: 2003, A&A 403, 501. Garrett, M.: 2002, A&A 384, L19. Genzel, R. and Cesarsky, C.: 2000, ARAA 38, 761. Goldader, J. D., Meurer, G., Heckman, T. M., et al.: 2002, ApJ 568, 651. Goldschmidt, et al.: 1997, MNRAS 289, 465. Gonzalez-Solares, E. A., Oliver, S., Gruppioni, C., et al.: 2004, MNRAS 352, 44. Gruppioni, C., Pozzi, F., Zamorani, G., et al.: 2003, MNRAS 341, L1. Haarsma, D. B., Partridge, R. B., Windhorst, R. A., and Richards, E. A.: 2000, ApJ 544, 641. Hauser, M. and Dwek, E.: 2001, ARAA 37, 249. Herbstmeierer, U., Abraham, P., Lemke, D., et al.: 1998, A&A 332, 739.

118

D. ELBAZ

Hughes, D. H., Serjeant, S., Dunlop, J., et al.: 1998, Nature 394, 241. Juvela, M., Mattila, K., and Lemke, D.: 2000, A&A 360, 813. Kakazu, Y., Sanders, D., Jpseph, R., et al.: 2002, IAU184, astro-ph/0201326. Kawara, K., Sato, Y., Matsuhara, H., et al.: 1998, A&A 336, L9. Kawara, K., Matsuhara, H., Okuda, H., et al.: 2004, A&A 413, 843. Kessler, M., Steinz, J., Anderegg, M., et al.: 1996, A&A 315, L27. King, A. J. and Rowan-Robinson, M.: 2003, MNRAS 339, 260. Kim, D.-C. and Sanders, D. B.: 1998, ApJS 119, 41. Kiss, C., Abraham, P., Klaas, U., et al.: 2001, A&A 379, 1161. Lagache, G., Abergel, A., Boulanger, F., et al.: 1999, A&A 344, 322. Lagache, G. and Puget J.-L.: 2000 , A&A 355, 17. Lagache, G., Dole, H., and Puget J.-L.: 2003, A&A MNRAS 338, 555. Lagache, G., Dole, H., Puget J.-L., et al.: 2004, ApJS 154, 112. Lari, C., Pozzi, F., Gruppioni, C., et al.: 2001, A&A 325, 1173. Lemke, D., Klaas, U., Abolins, J., et al.: 1996, A&A 315, L64. Liang, Y., Hammer, F., Flores, H., Elbaz, D., and Cesarsky, C. J.: 2004, A&A 423, 867. Lilly, S. J., Le Fevre, O., Hammer, F., and Crampton, D.: 1996, ApJ 460, L1. Linden-Voernle, M. J. D., Norgaard-Nielsen, H. U., Jorgensenet, H. E., et al.: 2000, A&A 359, L51. Madau, P., Ferguson, H. C., Dickinson, M. E., et al.: 1996, MNRAS 283, 1388. Madau, P. and Pozzetti, L.: 2000, MNRAS 312, L9. Malkan, M. and Stecker, F.: 2001, ApJ 555, 641. Marcillac, D., Elbaz, D., Charlot, S., et al.: 2004, submitted A&A (submitted). Matsuhara, H., Kawara, K., Sato, Y., et al.: 2000, A&A 361, 407. Metcalfe, L., Kneib, J.-P., McBreen, B., et al.: 2003, A&A 407, 791. Meurer, G. R., Heckman, T. M., and Calzetti, D.: 1999, ApJ 521, 64. Miville-Deschˆenes M.-A., Lagache, G., and Puget, J.-L.: 2002, 393, 749. Okumura, K.: 1998, ESA ISOCAM PSF Report, www.iso.vilspa.esa.es/users/expl lib/CAM/ psf rep.ps.gz. Oliver, S., et al.: 1996, MNRAS 280, 673. Oliver, S., Rowan-Robinson, M., Alexander, D., et al.: 2000, MNRAS 316, 749. Oliver, S., Mann, R. G., Carballo, R., et al.: 2002, MNRAS 332, 536. Papovich, C., Dole, H., Egami, E., et al.: 2004, ApJS 154, 70. Patris, J., Dennefeld, M., Lagache, G., and Dole, H.: 2003, A&A 412, 349. Pearson, C.: 2001, MNRAS 325, 1511. Puget, J.-L., Abergel, A., Bernard, J.-P., et al.: 1996, A&A 308, L5. Puget, J.-L., Lagache, G., Clements, D., et al.: 1999, A&A 345, 29. Puget J.-L. and Lagache, G.: 2000, IAU204, astro-ph/0101105. Rigopoulou, D., Spoon, H. W. W., Genzel, R., et al.: 1999, AJ 118, 2625. Rigopoulou, D., Franceschini, A., Aussel, H., et al.: 2000, ApJ 537, L85. Roche, N. and Eales, S.: 1999, MNRAS 307, 111. Rodighiero, G. and Franceschini, A.: 2004, A&A 419, L55. Rodighiero, G., Lari, C., Fadda, D., et al.: 2004, A&A (in press) (astro-ph/0407639). Rowan-Robinson, M.: 2001, ApJ 549, 745. Rowan-Robinson, M., Lari, C., Perez-Fournon, I., et al.: 2004, MNRAS 351, 1290. Rush, B., Malkan, M. A., and Spinoglio, L.: 1993, ApJS 89, 1. Sanders, D. B. and Mirabel, I. F.: 1996, ARA&A 34, 749. Sanders, D., Surace, J. A., Ishida, C. M.: 1999, in J. E. Barnes and D. B. Sanders (eds.) “Galaxy Interactions at Low and High Redshift” IAU Symposium 186, Kyoto, Japan, (astro-ph/9909114). Sato, Y., Kawara, K., Cowie, L. L., et al.: 2003, A&A 405, 833. Sato, Y., Cowie, L. L., Kawara, K., et al.: 2004, AJ 127, 1285.

UNDERSTANDING GALAXY FORMATION WITH ISO DEEP SURVEYS

Serjeant, S., Oliver, S., Rowan-Robinson, M., et al.: 2000, MNRAS 316, 768. Serjeant, S., Efstathiou, A., Oliver, S., et al.: 2001, MNRAS 322, 262. Smail, I., Ivison, R. J., Blain, A. W., and Kneib, J.-P.: 2002, MNRAS 331, 495. Stanev, T. and Franceschini, A.: 1998, ApJ 494, L159. Starck, J.-L., Aussel, H., Elbaz, D. et al.: 1999, A&AS 138, 365. Steidel, C. C., Adelberger, K. L., Giavalisco, M., et al.: 1999, ApJ 519, 1. Stickel, M., Bogun, S., Lemke, D., et al.: 1998, A&A 336, 116. Stickel, M., Lemke, D., Klaas, U., et al.: 2000, A&A 359, 865. Takeuchi, T., Ishii, T., Hirashita, H., et al.: 2001, PASJ 53, 37. Tan, J., Silk, J. and Balland, C.: 1999, ApJ 522, 579. Taniguchi, Y., Cowie, L. L., Sato, Y., et al.: 1997, A&A 328, L9. Totani, T. and Takeuchi, T.: 2002, ApJ 570, 470. Tran, Q. D., Lutz, D., Genzel, R., et al.: 2001, ApJ 552, 527. Wang, Y.: 2002, A&A 383, 755. Wirth, G. D., Willmer, C. N. A., Amico, P., et al.: 2004, AJ 127, 3121. Xu, C., Hacking, P., Fang, F., et al.: 1998, ApJ 508, 576. Xu, C., Lonsdale, C., Shupe, D., et al.: 2001, ApJ 562, 179. Xu, C. K., Lonsdale, C. J., Shupe, D. L., et al.: 2003, ApJ 587, 90.

119

SOLAR SYSTEM

THE PLANETS AND TITAN OBSERVED BY ISO ´ THIERRY FOUCHET∗ , BRUNO BEZARD and THERESE ENCRENAZ LESIA, Observatoire de Paris, 5 Place Jules Janssen, 92195 Meudon Cedex, France (∗ Author for correspondence: E-mail: [email protected]) (Received 30 June 2004; Accepted in final form 13 December 2004)

Abstract. Infrared spectroscopic observations of planets and Saturn’s satellite Titan with the Infrared Space Observatory led to many significant discoveries that improved our understanding on the formation, physics and chemistry of these objects. The prime results achieved by ISO are: (1) a new and consistent determination of the D/H ratios on the giant planets and Titan; (2) the first precise measurement of the 15 N/14 N ratio in Jupiter, a valuable indicator of the protosolar nitrogen isotopic ratio; (3) the first detection of an external oxygen flux for all giant planets and Titan; (4) the first detection of some stratospheric hydrocarbons (CH3 , C2 H4 , CH3 C2 H, C4 H2 , C6 H6 ); (5) the first detection of tropospheric water in Saturn; (6) the tentative detection of carbonate minerals on Mars; (7) the first thermal lightcurve of Pluto. Keywords: solar system

1. Introduction At the time of ISO launch, the infrared spectrum of planets and Titan was not a virgin territory. Interplanetary spacecraft had carried infrared spectrometers in the vicinity of Mars, Jupiter, Saturn and Titan, Uranus, and Neptune. These instruments demonstrated the predominance of CO2 in the martian atmosphere, N2 on Titan, and H2 and He on the giant planets, and gave insights on the planetary composition in minor species. Dedicated planetary missions were also specifically designed to provide spatial and temporal resolution on the atmospheric structure, in order to constrain the planets’ dynamics and meteorology. However, instruments aboard planetary spacecraft generally lacked spectral resolution and sensitivity. In this respect, ISO instruments gave planetary scientists a significant improvement that allowed them to obtain important results on the formation, chemistry, and dynamics of planetary objects. Hereafter we detail the major ISO achievements with emphasis on the measurement of the planetary isotopic composition and its significance for our understanding of Solar System formation, on the oxygen exogenic flux in the outer Solar System, and on the hydrocarbon photochemistry. Earlier reviews of ISO results can be found in Lellouch (1999), B´ezard (2000) and Encrenaz (2003). Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands, and the United Kingdom), and with the participation of ISAS and NASA. Space Science Reviews (2005) 119: 123–139 DOI: 10.1007/s11214-005-8061-2

 C

Springer 2005

124

T. FOUCHET ET AL.

2. Isotopic Composition 2.1. DEUTERIUM

IN THE

GIANT P LANETS

The D/H ratio is one of the most useful diagnostic of the formation and evolution of the giant planets. They formed in the protosolar nebula composed of gas, mainly H2 and He, and icy grains, enriched in deuterium. Their current D abundances thus reflect the relative fraction of presolar gas and grains that participated in their formation. In Jupiter and Saturn, the majority of the mass came from the gaseous component of the presolar nebula. Their D/H ratios are therefore regarded as an estimate for the presolar D/H value. In contrast, the masses of Uranus and Neptune are dominated by ices, which have enriched their atmospheres in deuterium. For this reason, measuring their D/H ratios could yield the deuterium abundance in the proto-uranus and proto-neptunian ices, and help to constrain the condensation processes in the presolar nebula. Prior to the ISO mission, the deuterium abundance had been measured in Jupiter and Saturn from near-infrared lines of HD (Smith et al., 1989). However, the interpretation of these lines was complicated by some blending with absorptions from CH4 and uncertainties on the cloud physical properties. Therefore, the most accurate value for Jupiter was obtained in-situ by the mass spectrometer aboard the Galileo probe (Niemann et al., 1998). For the three other giant planets, we had to rely on measurements of the CH3 D/CH4 ratio, and on the knowledge of the fractionation coefficient f induced by the isotopic exchange reaction HD + CH4

H2 + CH3 D. The latter coefficient is quite uncertain as the reaction equilibrium constant and kinetics, and the planetary convection timescales are not well known. As predicted by B´ezard et al. (1986), ISO allowed the first detection of HD rotational lines (Encrenaz et al., 1996; Griffin et al., 1996; Feuchtgruber et al., 1999a; Lellouch et al., 2001). These lines enabled a simple and consistent determination of the deuterium abundance on the four giant planets. The LWS instrument detected the R(0) line on all giant planets. The SWS instrument detected the R(2) line (37.7 μm) on each planet (Figure 1), while the R(3) line was only detected on Saturn. For Jupiter and Saturn, Lellouch et al. (2001) inferred D/H ratios in H2 of −5 (D/H)H2 = (2.4±0.4)×10−5 and (D/H)H2 = 1.85+0.85 −0.60 ×10 , respectively. Using CH3 D and CH4 bands in the 7–9 μm range, Lellouch et al. (2001) also measured the −5 D/H in methane, (D/H)CH4 = (2.2 ± 0.7) × 10−5 and (D/H)CH4 = 2.0+1.4 −0.7 × 10 , respectively for Jupiter and Saturn. Combining these two measurements yielded −5 global D/H ratios of (2.25 ± 0.35) × 10−5 and 1.70+0.75 −0.45 × 10 , respectively. Correcting for a small atmospheric enrichment in deuterium from ices that constituted the proto-cores of these planets, Lellouch et al. (2001) obtained a protosolar ratio of (D/H)ps = (2.1 ± 0.4) × 10−5 . Other estimates of the (D/H)ps ratio had been obtained analysing the evolution of the 3 He/4 He ratio from its protosolar value to its current solar value (Geiss and Gloeckler, 1998). The current solar ratio was measured in the solar wind, while

PLANETS AND TITAN

125

Figure 1. The HD R(2) rotational line observed on the four giant planets.

the protosolar 3 He/4 He ratio was assumed equal to the current Jupiter ratio measured in situ by Galileo (Niemann et al., 1998), or equal to the ratio measured in meteorites (Geiss and Gloeckler, 1998). These values led to protosolar deuterium composition, (D/H)ps = (1.94 ± 0.5) × 10−5 and (D/H)ps = (2.1 ± 0.5) × 10−5 , respectively, in excellent agreement with the ISO measurement. This confirms that Jupiter is a reliable indicator of the protosolar D/H. The protosolar value also indicates a weak decrease of the D/H ratio in the Local Interstellar Medium (LISM) since the formation of the Solar System: (D/H)LISM = (1.5 ± 0.1) × 10−5 (Linsky, 1998). For Uranus and Neptune, the retrieved isotopic ratios were larger than those −5 and (D/H)H2 = 6.5+2.5 measured in Jupiter and Saturn: (D/H)H2 = 5.5+3.5 −1.5 ×10 −1.5 × 10−5 , respectively (Feuchtgruber et al., 1999a). This confirms that at least a fraction of the ices that constituted the proto-cores of these planets was mixed in their atmospheres. Using the interior models of Podolak et al. (1995), Feuchtgruber et al. (1995a) were able to evaluate the deuterium composition of the proto-uranian and −5 −5 and (D/H)ices = 10.8+8.5 proto-neptunian ices: (D/H)ices = 9.4+7.6 −4.2 × 10 −4.7 × 10 , respectively. These values differ by about a factor of three from the D/H ratios measured in cometary ices, ∼30 × 10−5 (see Altwegg and Bockel´ee-Morvan, 2003 for a review). Drouart et al. (1993) and Hersant et al. (2001) used this D/H variability within Solar System objects as a chronometer of their formation. They argued that the various D/H ratios resulted from the different levels of isotopic exchange between

126

T. FOUCHET ET AL.

water enriched in deuterium by interstellar chemistry and the hydrogen gas. The sooner ices condensed in the protosolar nebula, the higher were their D/H ratios. Using turbulent models of the nebula Drouart et al. (1999) and Hersant et al. (2001) calculated that comets formed in ∼105 years, while Uranus and Neptune cores took ∼106 years to agglomerate. 2.2. DEUTERIUM

IN

TITAN

In Titan, methane is the prime deuterium carrier. Using PHT-S and SWS data, Coustenis et al. (2003) observed both CH3 D and CH4 between 7 and 9 μm to −5 derive a D/H ratio of (D/H)CH4 = 8.7+3.2 −1.9 × 10 . The ISO result is smaller than −4 the value inferred from Voyager 1/IRIS spectra, 1.5+1.4 (Coustenis et al., −0.5 × 10 1989), but agrees with the ground-based determination of (7.75 ± 2.25) × 10−5 of Orton et al. (1992). Two different scenarios have been proposed to explain this deuterium enrichment in Titan’s atmosphere compared to that in the protosolar gas. Both scenarios account for the replenishment of Titan’s atmosphere in methane, which is effectively photolysed in heavier hydrocarbons by the solar ultraviolet flux. In the first scenario, Lunine et al. (1999) proposed that methane evaporates from a local, small reservoir, located near the surface. In this case, over 4.5 byr, solar photolysis may have enriched the D/H in atmospheric methane up to a factor of 4. In the second scenario, Mousis et al. (2002) argued that the atmosphere could be replenished by cryovolcanism. In this case Mousis et al. (2002) advocated that solar photolysis did not induce any fractionation, and that the observed deuterium enrichment is primordial. In this case, the D/H ratio in Titan could also be used as a chronometer for the condensation of ices at Saturn’s heliocentric distance in an approach similar to that used by Drouart et al. (1999) and Hersant et al. (2001). 2.3.

15

N/14 N R ATIO

IN J UPITER

The value of the 15 N/14 N in the protosolar nebula had long been an unsolved mystery. Measurements in comets (Altwegg and Bockel´ee-Morvan, 2003), in the solar wind trapped in the lunar regolith (Kerridge, 1993), in situ in the solar wind (Kallenbach et al., 1998), and in meteorites (Pillinger, 1984) gave values ranging between 2.5 × 10−3 and 6.9 × 10−3 that were impossible to explain within the framework of a consistent, global scenario of the nebula. In addition, measurements in Jupiter’s atmosphere yielded a ratio comparable to the terrestrial ratio ((15 N/14 N)⊕ = 3.68 × 10−3 ), but with uncertainties as large as a factor of 2 (Encrenaz et al., 1978; Tokunaga et al., 1980). Using ISO-SWS observations at 10 μm, Fouchet et al. (2000a) refined the measurement in Jupiter, yielding a 15 14 −3 N/ N ratio of (1.9+0.9 −1.0 ) × 10 , much smaller than the terrestrial ratio (Figure 2). A similar result was found by the in situ measurements of Galileo (Owen et al.,

PLANETS AND TITAN

127

Figure 2. Comparison between Jupiter’s ISO spectrum (solid line) and two synhtetic models with terrestrial (dashed line) and half-terrestrial (dotted line) 15 N/14 N.

2001), and latter by the Cassini/CIRS observations (Abbas et al., 2004; Fouchet et al., 2004). Formation models of Jupiter (Owen et al., 2001; Hersant et al., 2004) suggest that the planet has retained most of the nitrogen present in its feeding zone. Therefore, the jovian nitrogen isotopic ratio can be seen as a tracer of the protosolar ratio. This view is supported by a recent reanalysis of lunar samples that gave an upper limit of 2.8 × 10−3 for the trapped solar wind (Hashizume et al., 2000). Such a difference between the protosolar and the terrestrial 15 N/14 N ratios bears important consequences for the origin of the Earth atmosphere. Terrestrial N2 must originate from a minor nitrogen reservoir in the protosolar nebula, highly fractionated (a factor of 2) with respect to the largest reservoir. If interstellar chemistry seems the most probable culprit (Charnley and Rodgers, 2002; Al´eon and Robert, 2004), the exact fractionation mechanism still remains to be identified, along with the main nitrogen carrier to the Earth.

128

T. FOUCHET ET AL.

3. Exogenic Oxygen to the Giant Planets One of the most striking results obtained from ISO observations is the discovery of H2 O in the stratospheres of the four giant planets (Feuchtgruber et al., 1997, 1999b; Lellouch et al., 2002), and Titan (Coustenis et al., 1998), as well as the detection of CO2 in the stratospheres of Jupiter, Saturn and Neptune (Feuchtgruber et al., 1997, 1999b; Lellouch et al., 2002). The presence of water above the tropopause cold trap can only be explained by an external influx. CO2 also condenses at the tropopause of Saturn and Neptune, while internal sources on Jupiter are believed to be weak, since carbon–oxygen bonded species present in the planetary interior are effectively converted to CH4 at shallow atmospheric levels. Therefore, the H2 O and CO2 detections from ISO provided evidences for an exogenic supply of oxygen in the atmospheres of outer Solar System objects. Three different possible sources are possible: (i) infall of interplanetary dust particles (IDPs), (ii) sputtering from icy rings and satellites, and (iii) rare impacts from kilometre-sized bodies. In Jupiter, CO2 may orginates from the Shoemaker-Levy 9 (SL9) impacts that occured in 1994. The shock chemistry induced by the impacts produced CO and H2 O that were deposited by the plumes at atmospheric levels above the 100-μbar level. CO and H2 O can subsequently react to form CO2 , while all the chemical products are slowly transported, both horizontally from the location of the impacts (44◦ S), and vertically. The ISO-SWS 14 arcsec × 27 arcsec aperture allowed Lellouch et al. (2002) to crudely resolve the CO2 latitudinal variation. They derived column densities, (6.3 ± 1.5) × 1014 cm−2 in the southern hemisphere, (3.4 ± 0.7) × 1014 cm−2 in the Equatorial Region, and 8, with a width equal to the instrumental one, and centered at 121.75 μm.

PRE-MAIN SEQUENCE STARS SEEN BY ISO

197

A possible interpretation of the observed FIR spectra is based on the simultaneous presence of two components: (i) a well localized J-shock, responsible for the [O I] emission, and (ii) an extended low density ionized medium, produced by the UV photons from the disk boundary layer (Teff ∼ 3 × 104 K), responsible for the [N II] and [C II] emission. Infact, in FU Ori systems, the central star (F–G type) cannot provide the requisite ionising photons (G o ≈ 0.1 at 0.1 pc).

4. Future Perspectives The FIR spectroscopical surveys presented here have clarified many aspects of the interaction between pre-main sequence objects and their close environment, and have originated new questions which can be hopefully answered by using the forthcoming facilities. Some relevant aspects, addressed in this review, are summarized in the following: – ISO instrumentation has evidenced for the first time the multiplicity of phenomena (outflows, shocks, disk emission, fluorescence) occurring in complex systems or in young binary objects (such as T Tau), but, due to the large beam size, one is often unable to specify which is the origin of the different manifestations. – Surveys of H2 pure rotational lines (in the 7–28 μm range) revealed themselves as crucial to probe small amounts of warm gas (maybe disks/planets) around young embedded objects, but enough spatial resolution (better than 1 arcsec) is required to impose compelling constrains. – The ISO-LWS mapping capability is enough to derive (from [O I] and [C II] line ratios) the gross distribution of G o and density. A better spatial resolution of ∼10 arcsec is in order to investigate the internal structure of the detected PDR’s by tracing the transition zones C+ /C/CO, thus clarifying whether or not clumpiness is an important ingredient. – The observed ratio [OI] 63 μm/[OI] 145 μm is always lesser than predicted from both standard PDR and J-C-shocks models. To find a satisfactory agreement with the observed data, we pushed the PDR models toward the lowest [OI] 63/145 values. Studying whether this large scale behaviour stems from properties intrinsic to the emitting gas or results from averaging different emission vs. absorption components, deserves a spatial resolution much better (by ∼ an order of magnitude) than that of ISO-LWS. – Molecular emission in HAEBE has been detected on a few sources, coming from very compact (≤10 arcsec) regions. A systematic study carried out with a better sensitivity and spatial resolution could both give a definitive answer to the size of the emitting regions and significantly increase the detection rate of molecular emission (which is, at the moment, quite low). – ISO missed the fine structure transition of [N II] at 205 μm. The diagnostic diagram which uses the [N II] line pair (122/205 μm, see e.g. Rubin et al., 1994),

198

D. LORENZETTI

would provide the electron density in the N+ volume and the fractional ionization in the environment around FU Ori objects. This will allow to verify the hypothesis about the [N II] excitation mechanism in that class. Moreover, at higher spectral resolution, the [N II] 205 μm line (expected to be brighter than the 122 μm one) can be resolved, allowing a better constrain on the origin of this line.

References Acke, B. and van den Ancker, M.: in press, Astron. Astrophys. 426, 151. Abraham, P., Leinert, C., Burkert, A., Henning, T., and Lemke, D.: 2000, Astron. Astrophys. 354, 965. Benedettini, M., Nisini, B., Giannini, T., et al.: 1998, Astron. Astrophys. 339, 159. Benedettini, M., Pezzuto, S., Giannini, T., Lorenzetti, D., and Nisini, B.: 2001, Astron. Astrophys. 379, 557. Bouwman, J., de Koeter, A., Dominik, C., and Waters, L. B. F. M.: 2003, Astron. Astrophys. 401, 577. Bouwman, J., de Koeter, A., van den Ancker, M. E., and Waters, L. B. F. M.: 2000, Astron. Astrophys. 360, 213. Bouwman, J., Meeus, G., de Koeter, A., Hony, S., Dominik, C., and Waters, L. B. F. M.: 2001, Astron. Astrophys. 375, 950. Burton, M. G., Hollenbach, D., and Tielens, A. G. G.: 1990, Astrophys. J. 365, 620. Chiang, E. I. and Goldreich, P.: 1997, Astrophys. J. 490, 368. Chiang, E. I., Joung, M. K., Creech-Eakman, M. J., Qi, C., and Kessler, J. E.: 2001, Astrophys. J. 547, 1077. Creech-Eakman, M. J., Chiang, E. I., Joung, R. M. K., Blake, G. A., and van Dishoeck, E. F.: 2002, Astron. Astrophys. 385, 546. Dominik, C., Dullemond, C. P., Waters, L. B. F. M., and Walch, S.: 2003, Astron. Astrophys. 398, 607. Draine, B. T., Roberge, W. G., and Dalgarno, A.: 1983, Astrophys. J. 264, 485. Dullemond, C. P., Dominik, C., and Natta, A.: 2001, Astrophys. J. 560, 957. Elia, D., Strafella, F., Campeggio, L., Giannini, T., Lorenzetti, D., and Nisini, B.: 2004, Astrophys. J. 601, 1000. Fuente, A., Marti´ın-Pintado, J., Bachiller, R., Neri, R., and Palla, F.: 1998, Astron. Astrophys. 334, 253. Fuente, A., Marti´ın-Pintado, J., Rodr´ıguez-Fern´andez, N. J., Chernicharo, J., and Gerin, M.: 2000, Astron. Astrophys. 354, 1053. Fuente, A., Marti´ın-Pintado, J., Rodr´ıguez-Fern´andez, N. J., Rodr´ıguez-Franco, A., de Vicente, P., and Kunze, D.: 1999, Astrophys. J. 518, L45. Giannini, T., Lorenzetti, D., Tommasi, E., et al.: 1999, Astron. Astrophys. 346, 617. G¨urtler, J., Schreyer, K., Henning, T., Lemke, D., and Pfau, W.: 1999, Astron. Astrophys. 346, 205. Hartmann, L. W. and Kenyon, S. J.: 1985, Astrophys. J. 299, 462. Hartmann, L. W. and Kenyon, S. J.: 1996, Ann. Rev. Astron. Astrophys. 34, 207. Hartmann, L. W., Kenyon, S. J., and Hartigan, P.: 1993, in Levy, E. H. and Lunine, J. I. (eds.), Protostars and Planets, Vol. III, University of Arizona Press, Tucson, p. 497. Henning, T., Burkert, A., Launhardt, R., Leinert, C., and Stecklum, B.: 1998, Astron. Astrophys. 291, 546. Hollenbach, D. and McKee, C. F.: 1989, Astrophys. J. 342, 306. Hutsem´eker, D. and van Drom, E.: 1990, Astron. Astrophys. 238, 134. Kaufman, M. J., Wolfire, M. G., Hollenbach, D. J., and Luhman, M. L.: 1999, Astrophys. J. 527, 795. L´eger, A. and Puget, J. L.: 1984, Astron. Astrophys. 137, L5. Lorenzetti, D., Giannini, T., Nisini, B., et al.: 2000, Astron. Astrophys. 357, 1035.

PRE-MAIN SEQUENCE STARS SEEN BY ISO

199

Lorenzetti, D., Giannini, T., Nisini, B., et al.: 2002, Astron. Astrophys. 395, 637. Lorenzetti, D., Tommasi, E., Giannini, T., et al.: 1999, Astron. Astrophys. 346, 604. Malfait, K., Waelkens, C., Bouwman, J., de Koter, A., and Waters, L. B. F. M.: 1999, Astron. Astrophys. 345, 181. Malfait, K., Waelkens, C., Waters, L. B. F. M., Vandenbussche, B., Huygen, E., and de Graauw, M. S.: 1998, Astron. Astrophys. 332, L25. Meeus, G., Waters, L. B. F. M., Bouwman, J., van den Ancker, M. E., Waelkens, C., and Malfait, K.: 2001, Astron. Astrophys. 365, 476. Natta, A., Meyer, M. R., and Beckwith, S. V. W.: 2000, Astrophys. J. 534, 838. Natta, A., Pusti, T., Neri, R., Wooden, D., Grinin, V. P., and Mannings, V.: 2001, Astron. Astrophys. 371, 186. Nisini, B., Giannini, T., and Lorenzetti, D.: 2002, Astrophys. J. 574, 246. Nisini, B., Lorenzetti, D., Cohen, M., et al.: 1996, Astron. Astrophys. 315, L321. Peeters, E., Hony, S., van Kerckhoven, C., et al.: 2002, Astron. Astrophys. 390, 1089. Pezzuto, S., Grillo, F., Benedettini, M., et al.: 2002, MNRAS 330, 1034. Pezzuto, S., Strafella, F., and Lorenzetti, D.: 1997, Astrophys. J. 485, 290. Richter, M. J., Jaffe, D. T., Blake, G. A., and Lacy, J. H.: 2002, Astrophys. J. 572, L161. Rubin, R. H., Simpson, J. P., and Lord, S. D.: 1994, Astrophys. J. 420, 772. Sandell, G. and Weintraub, D. A.: 1994, Astron. Astrophys. 292, L1. Siebenmorgen, R., Natta, A., Kruegel, E., and Prusti, T.: 1998, Astron. Astrophys. 339, 164. Siebenmorgen, R., Prusti, T., Natta, A., and M¨uller, T. G.: 2000, Astrophys. J. 561, 1074. Spinoglio, L., Giannini, T., Nisini, B., et al.: 2000, Astron. Astrophys. 353, 1055. Th´e. P.S., de Winter, D., P´erez, M.R.: 1994, Astrophys. Supp. 104, 315. Thi, W. F., van Dishoeck, E. F., Blake, G. A., van Zadelhoff, G.-J., Hogerheijde, M. R.: 1999, Astrophys. J. 521, L63. Thi, W. F., van Dishoeck, E. F., Blake, G. A., et al.: 2001, Astron. Astrophys. 385, 546. Tielens, A. G. G. M. and Hollenbach, D.: 1985, Astrophys. J. 291, 722. van Boekel, R., Waters, L. B. F. M., Dominik, C., et al.: 2003, Astron. Astrophys. 400, L21. van den Ancker, M. E., Bouwman, J., Wesselius, P. R., Waters, L. B. F. M., Dougherty, S. M., and van Dishoeck, E. F.: 2000a, Astron. Astrophys. 357, 325. van den Ancker, M. E., Wesselius, P. R., and Tielens, A. G. G. M.: 2000b, Astron. Astrophys. 355, 194. van den Ancker, M. E., Wesselius, P. R., Tielens, A. G. G. M., van Dishoeck, E. F., and Spinoglio, L.: 1999, Astron. Astrophys. 348, 877. van den Kerckhoven, C., Tielens, A. G. G. M., and Waelkens, C.: 2002, Astron. Astrophys. 384, 568. van Dishoeck, E. F.: 2004, Ann. Rev. Astron. Astrophys. 42, 119. Wesselius, P. R., van den Ancker, M. E., Young, E. T., et al.: 1996, Astron. Astrophys. 315, L197.

DEBRIS DISCS AROUND STARS: THE 2004 ISO LEGACY MARIE JOURDAIN DE MUIZON Leiden Observatory, PO Box 9513, 2300 RA Leiden, The Netherlands; LESIA, Paris Observatory, 92190 Meudon, France (E-mail: [email protected]) (Received 3 August 2004; Accepted in final form 2 November 2004)

Abstract. Debris discs around stars were first discovered by the Infrared Astronomical Satellite (IRAS) in 1983. For the first time material orbiting another star than the Sun, but distinct from a circumstellar envelope, was observed through its far infrared emission. This major discovery motivated astronomers to investigate those discs by further analyzing the IRAS data, using ground-based telescopes for the hunting of exoplanets, developing several projects using the Infrared Space Observatory (ISO), and now exploiting the ISO Data Archive (IDA). This review presents the main ISO results, statistical as well as individual, on debris discs in orbit around pre-main-sequence and main-sequence stars. Keywords: stars, debris discs, debris disks, planetary systems, infrared astronomy, Vega, β Pic, HD 207129, α PsA, ρ 1 CnC

1. Introduction Discs around stars appear during the early stages of stellar evolution. About 4.6 Gyr ago, the Sun – like any other star – formed in a local contraction of an interstellar cloud of molecular gas and dust. During its first few million years the Sun was surrounded by a warm rotating disc of gas and dust that on one hand fed material onto the forming star and on the other hand led to the formation of comets, planets and planetesimals, asteroids and other objects. This kind of warm disc is also found around other pre-main-sequence stars, such as T Tauri stars, Herbig Ae/Be stars and ZAMS stars (Robberto et al., 1999). They contain some original interstellar dust, are optically thick and extend to a few AU. After a few 107 years, the warm inner part of the solar disc was dissipated while a cooler debris disc remained in the outer part of the newly formed solar system. Such a disc contains mainly interplanetary dust resulting from collisions, is optically thin and extends up to a few hundred AU. It is such a debris disc that was first discovered by Infrared Astronomical Satellite (IRAS; Aumann et al., 1984) around Vega, α Lyr, one of the best calibrated and Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands, and the United Kingdom), and with the participation of ISAS and NASA. Space Science Reviews (2005) 119: 201–214 DOI: 10.1007/s11214-005-8063-0

 C

Springer 2005

202

M. JOURDAIN DE MUIZON

extensively used photometric standard in the visual range, then found to radiate more mid- and far-infrared emission than its photosphere can produce, and also around α PsA, ε Eri and β Pic. A systematic search for Vega-like discs in the IRAS survey led to several other identifications but was not very conclusive due to the limited sensitivity (Backman and Paresce, 1993; Plets and Vynckier, 1999). As follow-up of IRAS, several surveys were undertaken with Infrared Space Observatory (ISO; Kessler et al., 1996). The most unbiased was obtained by Habing et al. (1999, 2001) but others concentrated on stars in open clusters (Spangler et al., 2001), G dwarfs (Decin et al., 2000), or selected MS stars (Fajardo-Acosta et al., 1999). Several case studies on the most ISO observed discs are also presented: Vega and β Pic (Heinrichsen et al., 1998, 1999), ρ 1 Cnc (Dominik et al., 1998), HD 207129 (Jourdain de Muizon et al., 1999), five Vega-like stars including α PsA (Fajardo-Acosta et al., 1997). Finally ISO spectroscopic observations evidenced the presence of amorphous and crystalline silicates, forsterite, PAHs (Meeus et al.; 2001, Bouwman et al., 2001) and molecular hydrogen (Thi et al., 1999, 2001a,b) in the discs of several pre-main-sequence stars. See also two more general reviews on the subject by Lagrange et al. (2000) and Zuckerman (2001).

2. ISO Surveys 2.1. G ENERAL S TATISTICS Habing et al. (1999, 2001) proposed to determine the incidence of Vega-like debris discs in a distance limited sample of main-sequence stars. The stars were carefully selected from Johnson and Wright (1983) who computed far-infrared fluxes for 93% of the 2,150 stars in the “Woolley catalog of stars within 25 pc from the Sun.” After rejecting all stars either too faint for ISO at 60 μm or for which an infrared excess would be ambiguous to interpret as a disc, their final selection consisted of 84 main-sequence stars of spectral types from A to K. Within this range, no spectral type was privileged. The stars were measured with ISO at 25, 60, 90, and 170 μm in a total observing time of 65 h. Based on ISOPHOT (Lemke et al., 1996) C100 3 × 3 minimaps at 60 μm, Habing et al. (1999, 2001) found that 17% of the stars in their sample do have a disc. Lachaume et al. (1999) determined the age of the 84 stars. It appeared that all stars younger than 300 Myr have a disc, 70% of the stars younger than 400 Myr still have a disc but this is the case for only 8% of the stars older than 1 Gyr. Thus, it seems that most stars arrive on the main-sequence surrounded by a disc, and for the majority of them the disc is dissipated within a few hundred Myr (Figure 1). The decay of the discs is attributed to the destruction and escape of planetesimals; indeed the collision of planetesimals is a good source of dust, necessary to replenish the disc because dust particles disappear on a much shorter timescale relative to the lifetime of the discs. The timescale of the dissipation of the disc corresponds to the

DEBRIS DISCS AROUND STARS: THE 2004 ISO LEGACY

203

Figure 1. Cumulative distribution of stars with an infrared excess, as a function of the index, sorted by age. X-axis is rank at the bottom and age at the top of the figure. Y-axis is the number of stars up to that rank which have a disc. Each dot on the figure represents a star of the sample. For each star with an infrared excess, Nex increments by one. The two segments of a continuous straight line correspond to the perfect situation in which most debris discs are no longer detectable after 400 Myr (Habing et al., 2001).

end of the heavy bombardment phase in the Solar System, which is identified by dating the cratering on the Moon, Mercury, Ganymede, and Callisto. Craters are due to impacts from planetesimals and this bombardment appears to have stopped about 4 Gyr ago (Shoemaker and Shoemaker, 1999; Morbidelli et al., 2001). Thus, the timescale of 400 Myr would not trace, properly speaking, the disappearance of the dust discs, but the termination of the production of dust by collisions of planetesimals, hence the drastic reduction in the occurrence of these larger bodies (Jourdain de Muizon et al., 2001). Using 25 μm ISOPHOT data on 81 stars from the Habing et al. (2001) sample, Laureijs et al. (2002) found that 5 stars (6%) have an infrared excess that can be attributed to a disc of dust temperature between 50 and 120 K. The 5 stars are younger than 400 Myr, thus indicating that warm debris discs are relatively rare and concern the younger stars only. The survey confirms that there seems to be an absence of detectable amounts of dust close to the stars (D ≤ 20 AU). From a survey of 38 main-sequence stars using IRAS and ISOPHOT data, Fajardo-Acosta et al. (1999) found no star with a significant excess at 12 μm, and a fraction of ∼14% excess stars at 20 μm. However, this fraction is difficult to interpret since the ISOPHOT data in their study were inconclusive and the detections needed confirmation. The absence of 12 μm excess indicates that the discs are not warmer than 200 K. Finally, Jourdain de Muizon and Laureijs (2005) found three more excess stars at 60 μm and one more at 90 μm by exploring the ISOPHOT chopped data, which had been obtained at an early stage in the ISO mission but were later replaced by minimaps (see Habing et al., 2001).

204

M. JOURDAIN DE MUIZON

2.2. G STARS Decin et al. (2000) studied the incidence of the Vega phenomenon around G dwarfs. The stars were selected from the CORALIE sample (planet-search programme around stars closer than 50 pc and of spectral type from F8 to M1, excluding giants and faint cool dwarfs). Only southern stars were considered and the main selection criteria were observability and detectability by ISO at 60 μm. Confused systems such as multiple stars and those against a high cirrus background were rejected. Finally, this ISO programme consisted of 69 stars, 34 of which were effectively observed and results are given for 30 of them (ISOPHOT minimaps at 60 μm). Of the 30 G dwarfs in Decin et al. (2000), five (17%) have a debris disc and four out of these five are older than 3 Gyr. This is in good agreement with Habing et al. (2001) who have 21 G dwarfs in their sample and find that four (19%) of them have a debris disc; three out of these four are older than 5 Gyr (they are part of their 8% stars older than 1 Gyr and which still have their disc, see Section 2.1). However, two of the excess stars in Decin et al. (2000) have fractional luminosities of their disc comparable to the disc of β Pic which was a unique case so far, whereas the others and all the discs around G stars in Habing et al. (2001) are between one and two orders of magnitude fainter. Also no correlation was found between the existence of a disc and a planet around the stars of the Decin et al. (2000) sample. Thus, for both groups of authors, about 18% of the G stars do have a disc and about 80% of these discs are around G stars older than 3 Gyr. Why do G stars appear to keep their discs longer than A, F, or K stars is not yet understood. Around the Sun, the zodiacal light and the Kuiper Belt are probably some remnants of the earlier disc. 2.3. STARS

IN

OPEN CLUSTERS

Spangler et al. (2001) used ISOPHOT to observe a total of 148 stars, of which 87 young (50–700 Myr), nearby (d < 120 pc) main-sequence stars in open clusters (α Persei, Coma Berenice, Hyades, Pleiades, UMa) of spectral type A to K, 41 T Tauri stars in the clouds Chamaeleon I, Scorpius and Taurus (d ≈ 140–150 pc), and a sample of 19 isolated young nearby field stars (d < 60 pc) and another field star. They obtained ISOPHOT chopped observations with the C100 detector at 60 and 100 μm or raster at 60 and 90 μm. The goal was to determine an evolutionary sequence for circumstellar disc characteristics, hence their choice of well-studied clusters in which stellar ages are fairly well defined. Although spectral types span the range A to K the authors gave a preference to solar-type stars and privileged spectral types F and G in their star selection. They detected 36 stars, of which 33 show evidence for a far-infrared excess, i.e., 22% of their hole sample. More than one third were already thought to have an IR excess in their IRAS data, the rest are new ISO detections. These latter consist of 13 cluster stars, 5 young field stars and 1 other field star, with an ISO excess emission at 60, 90 or 100 μm. Among the

DEBRIS DISCS AROUND STARS: THE 2004 ISO LEGACY

205

main-sequence excess stars, spectral type F clearly dominates. Given the sample size of the statistics, the 22% excess stars in Spangler et al. (2001) is not too far from the 17% in Habing et al. (2001) or in Decin et al. (2000). The difference is most likely due to the selection criteria, the sample of Habing et al. (2001) being the most unbiased and homogeneous. Discs around T Tauri stars are closer to accretion discs (optically thick) and are not properly called debris discs (optically thin) like those around main-sequence stars. We only mention them in this ISO review because they are precursors of the debris discs but they cannot be treated equally when defining the incidence of debris discs. A convenient parameter to describe systems exhibiting excess emission from circumstellar discs is the fractional excess luminosity, f d = L ex /L , where L ex is the luminosity of dust and L is the stellar bolometric luminosity. Although L ex is not easy to estimate as it requires to know the complete infrared spectrum of the excess, it can be done approximately using both the IRAS and ISO data. A very interesting result of Spangler et al. (2001) is how the IR excess evolves with time. They established that the fractional excess luminosity f d decreases with stellar age according to the power law f d ∝ (age)−1.76 . This is compatible with a collisionally replenished disc as suggested in Habing et al. (2001). Spangler et al. (2001) claim they do not see evidence for an abrupt cessation of the debris disc phenomenon as reported in Habing et al. (2001). It is indeed not obvious from their Figure 1, but in fact 27 out of their 33 excess stars (82%) are younger than 400 Myr, and the other 6 are between 400 and 625 Myr. This is in perfect agreement with Habing et al. (2001). It is clear that debris discs are mostly found around young stars. Spangler et al.’s sample is biased towards young stars anyway. All excess stars in their sample are younger than the end of the late heavy bombardment on the Moon. Robberto et al. (1999) present preliminary results on 97 very young stars in 5 open clusters, 3 of which are in common with those in Spangler et al. (2001), namely Chamaeleon, α Per and Pleiades, and they are all younger than 300 Myr. Using ISOPHOT with the C100 detector at 60 μm and P2 detector at 25 μm they detected only 4 stars, i.e., 4.1% of their hole sample. The four stars detected are all T Tauri stars in the Chamaeleon I cluster; three of them are classical T Tauri stars which are still in the accretion phase (there are only six such stars known in this cluster) and the fourth one is probably in the transition phase. Their results show that the transition from an optically thick disc (or accretion disc) to an optically thin one (or debris disc) occurs on a timescale of ∼10 Myr, with a transition phase lasting less than ∼0.3 Myr. 2.4. RE -A NALYSIS

OF THE ABOVE

S URVEYS

Greaves and Wyatt (2003) have addressed the issue of the incidence of debris discs among A- and G-type main-sequence stars from the various published ISO surveys and some IRAS discs (about 200 stars). Their analysis shows a much higher detection rate of debris disks towards A stars (9/20 hence 45%) than G stars (11/180

206

M. JOURDAIN DE MUIZON

hence 6%), and this big difference in disc detection rate applies even when stars of similar age are considered. From the observations, disc lifetime is estimated to about 0.5 Gyr and may occur at any time during the main-sequence life of the star. Disc lifetime and high occurrence of discs among A stars were two conclusions already reached by Habing et al. (2001), but they are now also verified on a much larger sample. The fact that A stars have a much shorter lifetime and a larger disc mass than G-stars, implies the disc lifetime is a much higher fraction of the star lifetime in A stars than in G-stars and thus it is not surprising that A stars discs are more often seen than in G stars and also often seen in the first half of the A star lifetime (i.e., younger than 500 Myr). The disc mass scans a range of about 100, decline with time is slow and follows a power law not steeper than t −1/2 . The conclusion that the debris disc lifetime can take place at any time during the mainsequence, and not necessarily at the beginning of the stellar life, would account for the existence of a few detected debris disks around some old G stars. According to the models (Kenyon and Bromley, 2002a,b), perturbing planetesimals can form slowly at large orbital radii as late as a few Gyr, thus producing dust by collisions and replenishing a disc, but the same models cannot really explain the more or less constant disc duration of about 0.5 Gyr. Also based on the various ISO surveys, the age dependence of the Vega phenomenon has been studied observationally and theoretically, respectively by Decin et al. (2003) and Dominik and Decin (2003). Decin et al. (2003) have critically reexamined the stellar age estimates from the existing ISO surveys. Two aspects are to be considered: (i) the time dependence of the disc dust mass, measured by f d the fractional (i.e., disc/star) luminosity and (ii) the incidence of debris discs versus stellar age. Decin et al. (2003) came to the conclusion that there is no clear trend in the time dependence of f d , in particular no power-law with an index of ≈−2 (e.g., Spangler et al., 2001), but rather a large spread irrespective of stellar age. However, for a given stellar age, the maximum f d is about 10−3 (upper cut-off). They also identify a few cases of very young stars with intermediate or small excess (i.e., below the lower cut-off for f d which is about 10−5 ), and they conclude that Vegalike stars are more common in young stars that in old stars, a conclusion which had already been reached by Habing et al. (1999). The results of Decin et al. (2003) are in agreement with Habing et al. (1999, 2001) who had the most unbiased sample of the ISO Vega-like studies (see Section 2.1). They differ significantly from those of Spangler et al. (2001), but that could be explained because the latter studied mainly stars in clusters, thus of similar age and evolution pattern, hence the rather steep power law they could derive for f d versus time. Dominik and Decin (2003) have tried to provide a coherent theoretical picture of the various ISO observations of debris discs. On one hand, without replenishment, the dust in a debris disc should disappear in few thousands years under the PoyntingRobertson effect, i.e., a drag on interplanetary particles caused by their interaction with solar (respectively stellar) radiation, which causes the particles to lose orbital

DEBRIS DISCS AROUND STARS: THE 2004 ISO LEGACY

207

momentum and spiral into the sun (respectively star). On the other hand the discs last much longer than a few thousands years and some of them are seen in old stars, thus the dust must be replenished. The model must account for both the existence of the disc at a given stage of stellar evolution and also the disc evolution (from its formation to its dissipation). The former needs an accurate knowledge of stellar ages, the latter an accurate estimate of f d . According to Dominik and Decin (2003) the ISO observations can be explained by a simple model of collisional cascades, completed by some stirring in the case of high mass discs (cf. Kenyon and Bromley, 2002b). In a collisional cascade scenario where collisions occur with constant velocities, the decrease of the amount of dust can be described by a power law of index − 1 and this applies to all observed debris discs. In the case of very low mass discs (about 10−3 M⊕ , practically unobserved so far), the slope of this power law can be of the order of −2. In a scenario where the collisional cascade is continuously stirred (because the collision velocities increase with time), the power law slope can be steeper than −1. 3. Case Studies A few stars were observed rather extensively and with several ISO observing modes or instruments because they were particularly interesting cases. The best examples are α Lyr (Vega), β Pic, α PsA (Fomalhaut), ρ 1 CnC and HD 207129.

3.1. VEGA

AND

β P IC

The most detailed ISO studies of the discs around these two prototypes of debris discs are found in Heinrichsen et al. (1998, 1999). They are summarized in Table I. Based on the 25 μm ISO data, the authors argue that the disc around β Pic is in fact much more massive than the cool dust derived from the ISO 60 μm emission. They suggest that there is some warm dust (300 to 500 K), extension of the inner disc seen in the optical, and a significant amount of cool dust in addition to that seen by ISO. The star could be surrounded by a large “Oort” cloud of comets. Walker and Heinrichsen (2000) present ISOPHOT-S spectra (from 6 to 12 μm) of 12 Vega-like stars including the four “prototypes” (Vega, β Pic, Fomalhaut and ε Eri) in search for silicates and PAHs features in debris discs. They found silicate dust emission towards two stars (HD 144432 and HD 139614), emission from carbon-rich molecules towards two others (HD 169142 and HD 34700) and emission from both towards HD 142666. For all their other stars, including the four “prototypes,” either only the photosphere is seen at these mid-infrared wavelengths, or some thermal featureless excess. The authors also present preliminary ISOPHOT maps at 60 and/or 90 μm, thus giving some insight in the extent and structure of the discs. The authors argue that the discs could be in the early stage of planet formation. However, they give no accurate information on stellar age, hence

208

M. JOURDAIN DE MUIZON

TABLE I ISOPHOT results on Vega and β Pic ISOPHOT

Vega

60 μm (P32/C100) Yes: face-on 7.8 pc 86 AU at 60 μm 140 AU at 90 μm Multifilter-photometry 25, 60, 80, 100, 120, 150, 170 and 200 μm Adopted dust emissivity Q(λ) ∝ 1/λ1.1 Dust mass in disc (1–5) × 10−3 M⊕ a (1.3–13) × 10−4 M⊕ Reference Heinrichsen et al. (1998) Proposal: Walker High-resolution scans Disc resolved Distance Disc radius

a Habing

β Pic 25, 60 μm Yes: edge-on 19.3 pc 84 AU at 25 μm 140 AU at 60 μm 4.85, 7.3, 11.3, 12.8, 16, 25, 60, 80, 120, 150 and 170 μm Q(λ) ∝ 1/λ (1.0–3.3) × 10−2 M⊕ a (1.2–12) × 10−2 M⊕ Heinrichsen et al. (1999) Proposal: Heinrichsen

et al. (2001).

the question: “Are the above features emitted by dust in a debris disc or in a much younger and thicker envelope or disc?” 3.2. α P S A Fajardo-Acosta et al. (1997) have obtained P32 maps with ISOPHOT-C100 at 60 μm, with a spatial resolution of about 30 for six Vega-type systems. At 60 μm there is no excess emission detected towards α CrB, σ Her or α Cen. There is a marginal detection of extended emission, out to ∼800 AU or 30 from the star γ Oph. Only α PsA in their sample shows a convincing excess emission in the range ∼30 to 80 , i.e., ∼210 to 560 AU, confirming the IRAS temperature of 58–75 K and suggesting grains up to ∼10 μm in size. They estimate a dust mass of ∼(2–6) × 10−3 M⊕ . A comparison with β Pic shows that both discs are similarly extended at 60 μm (400 AU for β Pic vs. 560 AU for α PsA). 3.3. ρ 1 CN C The star ρ 1 CnC is among the first 10 stars found to host a planet. In 1996, following the news of a planet around ρ 1 CnC, a special set of ISO observations was requested and carried out to search for a disc. The star ρ 1 CnC was observed with ISOPHOT at 25 μm (PHT03), and at 60, 90, 135 and 170 μm (PHT22). Dominik et al. (1998) found an excess of 170 ± 30 mJy at 60 μm. They interpreted it as a debris disc located at about 60 AU from the star and containing at least 4 × 10−5 M⊕ of dusty material. The star is a G8V, slightly metal-rich, located at a distance of 12.5 pc

DEBRIS DISCS AROUND STARS: THE 2004 ISO LEGACY

209

Figure 2. The infrared excess toward HD 207129. Comparison between two different dust compositions. The dashed line indicates a Draine and Lee (1984) interstellar grain model, the solid line a Li and Greenberg (1997) cometary dust model. The two cases differ most around 30–40 μm, but they both fit the remaining far-IR excess equally well (Jourdain de Muizon et al., 1999).

from the Sun. From CaII H and K lines, its age was estimated to be 5 Gyr. Butler et al. (1997) detected the presence of a planet of period 14,65 days, implying a semi-major axis of 0.11 AU. The inferred mass is M2 sin i = 0.84MJup . The farinfrared ISO excess was interpreted as a disc after any other possible origin had been examined and eliminated (companion M5, planet, cirrus knot in the background). The far-infrared spectrum of the excess is best fitted using cometary icy dust grains from the dust model by Li and Greenberg (1997) as seen in Figure 2 of Dominik et al. (1998). These observations brought the first evidence of the coexistence of a disc and a planet around a star other than the Sun. 3.4. HD 207129 This solar-type (G2V) star is particularly interesting because of its cold debris disc, may be one of the coldest observed so far. It caught the attention of Jourdain de Muizon et al. (1999) when they found a clear excess not only at 60 μm, but also at 170 μm in their ISOPHOT-C200 data. Additional discretionary ISO observations were requested and obtained to get a more complete infrared spectrum of the disc. Based on the independent measurements of the CaII K line made by two groups of authors (Pasquini, 1992; Henry et al., 1996), Lachaume et al. (1999) determined a stellar age of 4.7 Gyr similar to that of the Sun, adopted by Jourdain de Muizon et al. (1999) and Habing et al. (2001). Zuckermann and Webb (2000) argue that the star is a member of the Tucanae stream and must be as young as the Tucanae association (10–50 Myr) which is located at ∼45 pc. That is three times further than HD 207129 which is located at 15.6 pc from the Sun, according to the Hipparcos Catalogue (Perryman et al., 1997). This significantly weakens the possibility of its

210

M. JOURDAIN DE MUIZON

belonging to the Tucanae stars. In any case the age of the star is less of an issue if it is young because it would then strengthen the case of Habing et al. (1999, 2001) that debris disc is the privilege of young stars (less than 400 Myr). HD 207129 is indeed one of the few puzzling exceptions in their study. The infrared excess around HD 207129 is shown in Figure 2. The star emits approximately 1.1 × 10−4 of its luminosity longwards of 25 μm. The excess emission is explained by assuming a disc of dust-like material equivalent to 2 × 10−2 M⊕ . The dust temperature ranges from 10 to 50 K, which makes it the coldest debris disc around a Vega-like star known to date. The dust distribution in the disc presents a circular hole at a distance of ∼400 AU from the star. This hole should be filled within 10−3 of the stellar age by particles spiraling inward because of the PoyntingRobertson effect unless an agent sweeps it clean. It could thus be explained by the presence of one or more planets. 4. Dust and Gas Composition of the Discs Debris discs are essentially made of dust particles and larger bodies; however, these latter cannot be detected by ISO. In addition, the younger discs may still have a significant gas component. Before ISO, any attempt to trace the gas via CO molecular bands was unsuccessful. Direct measurement of H2 with ISO–SWS (de Graauw et al., 1996) has made a real break-through toward understanding the gas composition of very young discs. 4.1. DUST C OMPOSITION Amorphous silicates (r < 1 μm) were detected in the ISOPHOT-S spectra of nine classical T Tauri stars in the Chamaleon I dark cloud (Natta et al., 2000). They discuss a model that explains the origin of this material in a hot, optically thin surface layer of a disc around the star, i.e., in the disc atmosphere. Crystalline silicates were detected by van den Ancker et al. (2001) toward the pre-main-sequence (B9.5Ve, A0II–IIIe or AOV?) star 51 Oph, using SWS01 and LWS01 full scans. The solid-state bands and energy distribution indicate that the dust has formed recently; it is a very young disc, not a debris disc. Other emission bands from hot gas (∼350 K) such as CO, CO2 , H2 O and NO dominate the 4– 8 μm spectrum. Both these gas and dust bands are unusual for a young star and are more typical of evolved AGB stars, although 51 Oph does not seem at all to belong to that class. The authors explore various possibilities for the nature of 51 Oph, among them a recent episode of mass loss from a Be star, the collision of two gas-rich planets and the accretion of a solid body as the star expands at the end of its short main-sequence life. A variety of crystalline silicates, forsterite (Mg2 SiO4 ), has been found toward the star HD 100546. Malfait et al. (1998) present SWS01 and LWS01 full scans of this isolated Herbig Ae/Be star. Forsterite is present in the micrometeorites

DEBRIS DISCS AROUND STARS: THE 2004 ISO LEGACY

211

and interplanetary dust of our Solar System. The ISO spectrum of HD 100546 is very similar to that of comet Hale–Bopp published by Crovisier et al. (1997). The amount of forsterite in the disc of HD 100546 is equivalent to 1013 comets Hale– Bopp, strengthening the hypothesis by Grady et al. (1997) that the disc around HD 100546 contains a huge swarm of comets similar to the Oort cloud in the Solar System. Crystalline silicates have also been found in several other comets such as P/Halley. Malfait et al. (1998) argue that the crystallisation process occurs during the early phases of disc evolution around young stars. The ISO-SWS and -LWS spectra of an additional 14 isolated Herbig Ae/Be stars are presented in Meeus et al. (2001). These stars are believed to be the more massive analogs of T Tauri stars; they are seen as the progenitors of Vega-like stars (Waters and Waelkens, 1998). Meeus et al. (2001) obtained a variety of mid-infrared spectra ranging from the amorphous silicates (as in the M supergiant μ Cep) to the crystalline silicates (as in comet Hale–Bopp). The variations in the shape of the 8–14 μm part of the spectra indicate the prominence of one or the other form of silicates. Four out of the fourteen stars have no silicates. In most silicate stars in their sample, crystalline silicates are present; this is confirmed by the shape of the 15–28 μm ISO–SWS spectra of the stars. PAH bands are also identified toward half of the stars, and all features are superposed on a near-IR to mid-IR continuum excess. The authors interpret the continuum in term of disc geometry: an optically thick, geometrically thin, power-law component and an optically thin flare region (black body component). Bouwman et al. (2001) got a further detailed insight into the 6–14 μm spectrum of these stars in order to study the silicate grain processing. The 10 μm silicate profile is modelled using three components: silica (SiO2 ) responsible for the 8–9 μm blue shoulder in the silicate band, forsterite contributing to the 11.3 μm feature, and amorphous olivine with two typical grain sizes of 0.1 and 2.0 μm. They identify two causes for the observed shift in peak position of the silicate band: (i) a change in average grain size from small (0.1 μm) to big (2 μm), as a result of the depletion of small grains in the inner region of the disc, due to coagulation and other effects, (ii) a change in grain composition from amorphous silicate to a mixture of amorphous and Mg-rich crystalline silicate (forsterite) which could be the result of thermal annealing in the inner regions of the disc. However crystallization occurs on a longer timescale than coagulation. There is no correlation between dust composition and disc geometry. 4.2. G AS C OMPOSITION : H 2 Before ISO, any attempt to detect a gas component in Vega-like discs had been unsuccessful or inconclusive (see e.g., Liseau, 1999). The classical method of tracing molecular hydrogen by observing CO, widely applied in the interstellar medium, did not seem to help. Thi et al. (1999) discovered molecular H2 in the ISO–SWS spectra of the T Tauri star GG Tau. The detection of two pure rotational lines at 17.035 and 28.218 μm provided a direct measure of the total amount of warm molecular

212

M. JOURDAIN DE MUIZON

gas in the disc around the star. The same group (Thi et al., 2001a,b) found H2 lines in the SWS spectra of 4 T Tauri stars (spectral type K to M), 7 Herbig Ae stars (sp. type A to F), and 3 main-sequence stars including β Pic (spectral type A to F). Their data suggest the presence of warm gas (T ≈ 100–200 K), which mass ranges from ∼10−4 M up to 8 × 10−3 M , around all the stars they observed. This mass corresponds to 1–10% of the total disc mass inferred from millimeter continuum observations, and a much higher fraction in the case of debris discs. Additional CO observations show that CO is not a good tracer of the gas in circumstellar discs. The amount of CO gas is likely strongly affected by photodissociation due to the stellar and interstellar ultraviolet radiation in the surface layers of the disc and freeze-out onto grain surface in the midplane. Direct measurement of H2 leads to a gas-to-dust ratio of ∼100 in the debris discs of β Pic and HD 135344, similar to the interstellar medium ratio. The bulk of the gas around pre-main-sequence stars is cool (T ≈ 20–80 K), while the warm gas (T ≈ 100–200 K) may constitute the major gaseous component of debris discs around main-sequence stars, thus providing a reservoir for the formation of Jovian planets. 5. Conclusion and Open Questions The ISO Data Archive (IDA) has not yet been fully exploited as far as debris discs and their precursors are concerned. Several aspects of disc evolution are not well understood. Why and how do most discs disappear after a lifetime of about 400 Myr? Why do some stars keep a disc for much longer, up to several Gyr? Do discs coexist with planets? Is it systematic, and if not, what conditions lead to a disc-planets or disc-only system? Does it depend on the stellar spectral type? Which physical and chemical processes occur during the evolution of a disc? Although the ISO data has already provided partial answers to some of these questions, it is clear that our picture is still incomplete. Beyond the IDA, space projects such as Spitzer and the Herschel Space Observatory do and will follow up ISO observations of debris discs. Spitzer has a similar wavelength range as ISO but a higher sensitivity; it is able to probe deeper and, hence, perform similar kind of statistics as ISO but on a larger volume around the Sun, thus expected to detect photometrically hundreds of debris disks. Some preliminary Spitzer results on the few brightest disks also address the structure of the disks, dust grain composition, and search for substellar companions. Herschel will open a new window since it extends to the submillimeter range. This will allow to trace the cold component of the discs such as the dust located in the Kuiper Belt, the Oort cloud and still further from the star. References Aumann, H. H., Gillett, F. C., Beichman, C. A., de Jong, T., Houck, J. R., Low, F. J., et al.: 1984, ApJ 278, L23.

DEBRIS DISCS AROUND STARS: THE 2004 ISO LEGACY

213

Backman, D. and Paresce, F.: 1993, in E. Levy and J. Lunine (eds.), Protostars and Planets III, University of Arizona press, Tucson, p. 1253. Bouwman, J., Meeus, G., de Koter, A., Hony, S., Dominik, C., and Waters, L. B. F. M.: 2001, A&A 375, 950. Butler, R. P., Marcy, G. W., Williams, E., Hause, H., and Shirts, P.: 1997, ApJ 474, L115. Crovisier, J., Leech, K., Bockelee-Morvan, D., Brooke, T. Y., Hanner, M. S., Altieri, B., et al.: 1997, Science 275, 1904. Decin, G., Dominik, C., Malfait, K., Mayor, M., and Waelkens, C.: 2000, A&A 357, 533. Decin, G., Dominik, C., Waters, L. B. F. M., and Waelkens, C.: 2003, A&A 598, 636. de Graauw, Th., Haser, L. N., Beintema, D. A., Roelfsema, P. R., van Agthover, H., Barl, L., et al.: 1996, A&A 315, L49. Dominik, C., Laureijs, R. J., Jourdain de Muizon, M., and Habing, H. J.: 1998, A&A 329, L53. Dominik, C. and Decin, G.: 2003, A&A 598, 626. Draine, B. T. and Lee, H. M.: 1984, ApJ 285, 89. Fajardo-Acosta, S. B., Stencel, R. E., and Backman, D. E.: 1997, ApJ 487, L151. Fajardo-Acosta, S. B., Stencel, R. E., Backman, D. E., and Thakur, N.: 1999, ApJ 520, 215. Grady, C. A., Sitko, M. L., Bjorkman, K. S., Perez, M. R., Lynch, D. K., Russell, R. W., et al.: 1997, ApJ 483, 449. Greaves, J. S. and Wyatt, M. C.: 2003, MNRAS 345, 1212. Habing, H. J., Dominik, C., Jourdain de Muizon, M., Kessler, M. F., Laureijs, R. J., Leech, K., et al.: 1999, Nature 401, 456. Habing, H. J., Dominik, C., Jourdain de Muizon, M., Laureijs, R. L., Kessler, M. F., Leech, K., et al.: 2001, A&A 365, 545. Heinrichsen, I., Walker, H. J., and Klaas, U.: 1998, MNRAS 293, L78. Heinrichsen, I., Walker, H. J., Klaas, U., Sylvester, R. J., and Lemke, D.: 1999, MNRAS 304, 589. Henry, T. J., Soderblom, D. R., Donahue, R. A., and Baliunas, S. L.: 1996, AJ 111, 439. Johnson, H. M. and Wright, C. D.: 1983, ApJS 53, 643. Jourdain de Muizon, M., Laureijs, R. J., Dominik, C., Habing, H. J., Metcalfe, L., Siebenmorgen, R., et al.: 1999, A&A 350, 875. Jourdain de Muizon, M., Laureijs, R. J., Habing, H. J., Leech, K., Kessler, M. F., Metcalfe, L., et al.: 2001, Earth Moon Planets 85–86, 201. Jourdain de Muizon, M., Laureijs, R. J., and Augereau, J. C.: 2005, A&A in preparation. Kenyon, S. J. and Bromley, B. C.: 2002a, AJ 123, 1757. Kenyon, S. J. and Bromley, B. C.: 2002b, ApJ 577, L35. Kessler, M. F., Steinz, J. A., Anderegg, M. E., Clavel, J., Drechsel, G., Estaria, P., et al.: 1996, A&A 315, L27. Lachaume, R., Dominik, C., Lanz, T., and Habing, H. J.: 1999, A&A 348, 897. Lagrange, A.-M., Backman, D., and Artymowicz, P.: 2000, in V. Mannings, A. P. Boss, and S. S. Russell (eds.), Protostars and Planets IV, Tucson: University of Arizona Press, p. 639. Laureijs, R. J., Jourdain de Muizon, M., Leech, K., Siebenmorgen, R., Dominik, C., Habing, H. J., et al.: 2002, A&A 387, 285. ´ Lemke, D., Klaas, U., Abolins, J., Abrah´ am, P., et al.: 1996, A&A 315, L64. Li, A. and Greenberg, J. M.: 1997, A&A 323, 566. Liseau, R.: 1999, A&A 348, 133. Malfait, K., Waelkens, C., Waters, L. B. F. M., Vandenbussche, B., Huygen, E., and de Graauw, M. S.: 1998, A&A 332, L25. Meeus, G., Waters, L. B. F. M., Bouwman, J., van den Ancker, M. E., Waelkens, C., and Malfait, K.: 2001, A&A 365, 476. Morbidelli, A., Petit, J.-M., Gladman, B., and Chambers, J.: 2001, Meteorit. Planet. Sci. 36, 371. Natta, A., Meyer, M. R., Beckwith, S. V. W.: 2000, ApJ 534, 838.

214

M. JOURDAIN DE MUIZON

Pasquini, L.: 1992, A&A 266, 347. Perryman, M. A. C., Lindegren, L., Kovalevsky, J., Hoeg, E., Bastian, U., Bernacca, P. L., et al.: 1997, A&A 323, 49. Perryman, M. A. C., ESA.: 1997, The Hipparcos and Tycho Catalogues. Astrometric and Photometric Star Catalogues Derived from the ESA Hipparcos Space Astrometry Mission, ESA SP Series Vol No. 1200, ISBN: 9290923997 (set), ESA Publications Division, Noordwijk, Netherlands. Plets, H. and Vynckier, C.: 1999, A&A 343, 496. Robberto, M., Meyer, M. R., Natta, A., and Beckwith, S. V. W.: 1999, in Cox, P., and Kessler, M. F. (eds.), The Universe as seen by ISO, ESA SP-427, p. 195. Shoemaker, E. and Shoemaker, C.: 1999, in Beatty, J. K., Petersen, C. and Chaikin, A. (eds.), The New Solar System 4th edn, Sky Publishing Corporation and Cambridge University Press, p. 69. Spangler, C., Sargent, A. I., Silverstone, M. D., Becklin, E. E., and Zuckerman, B.: 2001, ApJ 555, 932. Thi, W. F., van Dishoeck, E. F., Blake, G. A., van Zadelhoff, G. J., and Hogerheijde, M. R.: 1999, ApJ 521, L63. Thi, W. F., Blake, G. A., van Dishoeck, E. F., van Zadelhoff, G. J., Horn, J. M. M., Becklin, E. E., et al.: 2001a, Nature 409, 60. Thi, W. F., van Dishoeck, E. F., Blake, G. A., van Zadelhoff, G. J., Horn, J. M. M., Becklin, E. E., et al.: 2001b, ApJ 561, 1074. van den Ancker, M. E., Meeus, G., Cami, J., Waters, L. B. F. M., and Waelkens, C.: 2001, A&A 369, L17. Walker, H. J. and Heinrichsen, I.: 2000, Icarus 143, 147. Waters, L. B. F. M. and Waelkens, C.: 1998, ARA&A 36, 233. Zuckerman, B. and Webb, R. A.: 2000, ApJ 535, 959. Zuckerman, B.: 2001, ARA&A 39, 549.

LATE STAGES OF STELLAR EVOLUTION JORIS A. D. L. BLOMMAERT1,∗ , JAN CAMI2 , RYSZARD SZCZERBA3 and MICHAEL J. BARLOW4 1 Instituut

voor Sterrenkunde, K.U. Leuven, Celestijnenlaan 200B, B-3001 Leuven, Belgium Ames Research Center, MS 245-6, Moffett Field, CA 94035, U.S.A. 3 N. Copernicus Astronomical Center, Rabia´ nska 8, 87-100 Toru´n, Poland 4 Department of Physics & Astronomy, University College London, Gower Street, London, WC1E 6BT, U.K. (∗ Author for correspondence: E-mail: [email protected]) 2 NASA

(Received 18 October 2004; Accepted in final form 2 November 2004)

Abstract. A large fraction of ISO observing time was used to study the late stages of stellar evolution. Many molecular and solid state features, including crystalline silicates and the rotational lines of water vapour, were detected for the first time in the spectra of (post-)Asymptotic Giant Branch (AGB) stars. Their analysis has greatly improved our knowledge of stellar atmospheres and circumstellar environments. A surprising number of objects, particularly young planetary nebulae with Wolf-Rayet (WR) central stars, were found to exhibit emission features in their ISO spectra that are characteristic of both oxygen-rich and carbon-rich dust species, while far-IR observations of the PDR around NGC 7027 led to the first detections of the rotational line spectra of CH and CH+ . Keywords: Infrared: stars, Stars: AGB and post-AGB, Stars: atmospheres, Stars: late-type, Stars: mass-loss, Stars: symbiotic, Stars: carbon, Stars: supergiants, Stars: Wolf-Rayet, Planetary Nebulae, Novae, dust, molecules

1. Introduction ISO has been tremendously important in the study of the final stages of stellar evolution. A substantial fraction of ISO observing time was used to observe different classes of evolved stars. IRAS had already shown the strong potential to discover many evolved stars with circumstellar shells in the infrared wavelength range. ISO provided the opportunity to extend this survey with higher sensitivity and spatial resolution, allowing many more sources to be detected in the inner regions of our Galaxy and in the Magellanic Clouds. But especially ISO’s spectroscopic capabilities in the 2–200 μm wavelength range allowed a detailed study of individual objects through observations of gas phase molecular bands, ionic forbidden lines and solid state bands. These studies are important not only to better understand the Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands, and the United Kingdom), and with the participation of ISAS and NASA. Space Science Reviews (2005) 119: 215–243 DOI: 10.1007/s11214-005-8057-y

 C

Springer 2005

216

J. A. D. L. BLOMMAERT ET AL.

processes that dominate the final stages of stellar life but also because the interstellar medium is enriched in heavy elements from the mass lost by these stars during the last phases of their evolution. In the following sections, we will present highlights from the ISO observations of different classes of evolved stars: Asymptotic Giant Branch (AGB) stars, post-AGB objects, planetary nebulae, novae, symbiotic stars and Wolf-Rayet (WR) stars.

2. The Asymptotic Giant Branch Stars When low- and intermediate-mass stars have exhausted hydrogen in their cores, they ascend the HR diagram to become cool giants, first on the Red Giant Branch (RGB) and later on the AGB. These phases are characterized by substantial mass loss; in the AGB phase, the mass loss effectively drives stellar evolution. The low effective stellar temperatures as well as the copious amounts of circumstellar material that form as a consequence of this mass loss make these objects ideal targets in the infrared. Asymptotic Giant Branch stars have been extensively studied with the various instruments onboard ISO, and have greatly enhanced the understanding of the composition, dynamics and interaction of the stellar atmosphere and the surrounding circumstellar material. In particular, SWS and LWS have greatly enhanced our knowledge of the composition of the molecular envelope and the sometimes thick dust shells surrounding these stars, while the sensitivity of ISOCAM greatly extended our knowledge of the fundamental properties of tens of thousands giant stars.

2.1. MOLECULES IN THE STELLAR ATMOSPHERES . . . Asymptotic Giant Branch stars are generally divided into oxygen-rich and carbonrich stars. All stars start of with a photospheric C/O ratio which is lower than unity; during the evolution on the AGB however, dredge-ups bring freshly processed carbon to the surface, and gradually enrich the atmosphere in carbon. Depending (mainly) on the initial mass, some stars eventually experience a carbon preponderance in their atmosphere, and then they are called carbon stars. Asymptotic Giant Branch stars have very extended and relatively cool (typically 3000 K) atmospheres, allowing the formation of molecules in the stellar atmospheres. As the stellar envelope material cools down, CO is about the first molecule to be formed. This process is so efficient (and CO so stable) that for O-rich stars virtually all carbon is locked up in CO, leaving only oxygen for further chemistry. For carbon stars, all oxygen is locked up in CO and the chemistry is carbon driven. The difference in spectral appearance is enormous, both in the molecular and dust contents (see Tables I and II).

LATE STAGES OF STELLAR EVOLUTION

217

TABLE I Overview of molecular bands and dust features reported in ISO observations of O-rich AGB stars λ (μm)

Component Molecular features H2 O OH CO SiO CO2

SO2

2.3–197 2.80–4.00 Fundamental 4.6 First overtone 2.3 Fundamental 8.0 First overtone 4.0 ν3 Asymmetric stretch 4.2 ν2 Bending 14.96 Various combination bands 12–18 ν3 Asymmetric stretch 7.55

Dust features Magnesiow¨ustite Mg(1−x) Fex O Alumina Al2 O3 Spinel MgAl2 O4 Crystalline water-ice Metallic Fe

19.5 11.0 13.0 43,60 2–10

References

[1–20, 29] [2, 4, 13–18] [2, 4, 14–19] [2, 4, 14–19] [5, 14–19] [2, 4, 13–19] [2, 4, 6, 14, 15, 17, 18] [3, 6, 9, 14, 15, 17, 18, 24, 28] [3, 6, 9, 14, 15, 17, 18, 24] [5, 15, 17, 18]

[15, 21, 24] [15, 21] [15, 22–24] [25, 26] [27]

For an overview of the silicate dust we refer to (Molster and Kemper, this volume). [1] Neufeld et al. (1996); [2] Tsuji et al. (1997); [3] Justtanont et al. (1998); [4] Yamamura et al. (1999a); [5] Yamamura et al. (1999b); [6] Cami et al. (2000); [7] Tsuji (2000); [8] Zubko and Elitzur (2000); [9] Markwick and Millar (2000); [10] Tsuji (2001); [11] Jørgensen et al. (2001); [12] Jones et al. (2002); [13] Matsuura et al. (2002a); [14] Matsuura et al. (2002b); [15] Cami (2002); [16] Decin et al. (2003); [17] Van Malderen (2003); [18] Justtanont et al. (2004); [19] Van Malderen et al. (2004); [20] Truong-Bach et al. (1999); [21] Posch et al. (2002); [22] Posch et al. (1999); [23] Fabian et al. (2001); [24] Sloan et al. (2003); [25] Sylvester et al. (1999); [26] Dijkstra et al. (2003a); [27] Kemper et al. (2002a); [28] Ryde et al. (1999); [29] Barlow et al. (1996).

The presence of molecules can easily be inferred through their ro-vibrational bands in the infrared. Such molecular bands have been used – amongst others – with great success to extend the MK classification to the infrared (Vandenbussche et al., 2002) and to improve the calibration of SWS (see Decin et al., 2003) using detailed models for A0–M2 stars; in turn, the SWS observations yielded many improvements in the models. Different molecules turn out to be sensitive to different stellar parameters, thereby, allowing to determine these stellar parameters from the ISO observations. The ISO observations of molecular bands have revealed the importance of fundamentally new ideas about modelling of such stars; one example is the development of fully consistent hydrodynamical codes, which include a frequency dependent radiative transfer (see e.g. H¨ofner, 1999; Gautschy-Loidl et al., 2004).

218

J. A. D. L. BLOMMAERT ET AL.

TABLE II Overview of molecular bands and dust features reported in ISO observations of C-rich AGB stars Component Molecular features CO CH CS SiS HCN

C3 C2 H2

Dust features SiC MgS

Fundamental Fundamental 1st overtone Fundamental First overtone Various bands 2ν2 ν2 2ν20 − ν21 ν3 Various bands ν4 + ν 5 ν5

λ (μm)

References

4.6 3.3–4.1 3.9–4.2 7–8 6.6–7 2.58–3.86 7.0 14.04 14.30 5.2 3.05 3.8 7–8 13.7

[1, 2, 4, 8] [4, 8] [1, 4, 8] [1, 4, 5, 8] [4] [1, 2, 4, 7, 8] [1, 4, 8] [1, 5, 6, 8] [5, 6, 8] [1–3, 8] [1, 2, 8] [1, 2, 8] [1, 2, 5, 8] [1, 2, 6, 8]

11.3 26–35

[2] [2, 9]

[1] Jørgensen et al. (2000); [2] Yang et al. (2004); [3] Hron et al. (1999); [4] Aoki et al. (1998); [5] Aoki et al. (1999); [6] Yamamura et al. (1999c); [7] Harris et al. (2003); [8] Gautschy-Loidl et al. (2004); [9] Hony et al. (2002a).

2.2. . . . AND BEYOND: THE “EXTENDED” ATMOSPHERE OF AGB STARS However, early in the ISO lifetime, it became clear that the molecular absorption seen in the SWS spectra of several O-rich AGB stars is quite different than was expected from detailed hydrostatic model calculations. Tsuji et al. (1997) noticed that CO and SiO absorption in early M giants was weaker than expected, while water vapour and CO2 bands were much stronger than predicted; in fact, the latter molecule was not predicted to be present in the stellar atmosphere at all. They found that the molecular features due to CO, CO2 and SiO correspond to excitation temperatures in the range 750–1250 K, indicating that they originate outside the stellar photosphere. This led to the terminology “extended atmosphere”, “warm molecular layer” or simply MOLsphere to denote that region close to the stellar photosphere (typically up to a few R ) where this excess molecular absorption or emission originates from and corresponds to a quasi-static molecular zone that was already hypothesized to exist based on ground-based data.

LATE STAGES OF STELLAR EVOLUTION

219

Soon, a wealth of molecular bands from this extended atmosphere were discovered and/or analysed (see Table I). These molecular bands are now used as probes to map the physical conditions (such as temperature and density) in the direct surroundings of the AGB stars (e.g. Cami et al., 2000; Justtanont et al., 2004). Of particular importance is water vapour, which is detected in both SWS and LWS spectra of O-rich AGB stars (and many others; see Cernicharo and Crovisier, this volume) through literally millions of spectral lines, and is the source for large differences in spectral appearance between Semi-Regulars and Miras. For SemiRegulars, water vapour is just a component in the spectra up to 10 μm, and the overall continuum shape is still determined by the stellar photosphere. In Mira variables on the other hand, the water vapour layer is dense, and due to the large opacity over a large wavelength range, this layer determines the shape of the IR continuum up to at least 10 μm in these objects. Analyses based on SWS observations suggested that this semiopaque layer typically extends to ∼2R ; this turns out to be in good agreement with interferometric observations in the infrared (Cami, 2003). Also, the pure rotational H2 O lines were first detected in the SWS (Neufeld et al., 1996) and LWS (Barlow et al., 1996) spectra of AGB stars. Molecular bands are also detected in the spectra of carbon stars. Absorption features due to C2 , HCN (and HNC), C2 H2 and C3 are commonly detected (see Table II); moreover, some (likely molecular) features detected in the SWS spectra are still unidentified (Yang et al., 2004). While the excitation temperatures of these molecular bands are in some cases comparable to the molecular bands in the O-rich stars, there is still debate about the exact origin of these molecular bands (see e.g. (Jørgensen et al., 2000)). It is clear, however, that for the carbon stars also, the ISO observations have greatly expanded the possibilities to study the stellar and (possibly) circumstellar envelope. 2.3. DUST The importance of ISO observations for the study of circumstellar dust can hardly be overestimated. The discovery of crystalline silicates in a wide variety of objects (see Molster and Kemper, this volume) and its consequences for the study of circumstellar and interstellar dust opened up the field of astromineralogy. Observations with SWS and LWS have resulted in the detection and identification of many more dust components in the circumstellar dust shells of AGB stars, and shed new light on the pathways to dust formation. Moreover, ISO observations have allowed to directly link the molecular features to the dust content in these stars (see e.g. Cami, 2002; Sloan et al., 2003). In the spectra of O-rich stars with low mass-loss rates, the typical silicate bands commonly found in Miras are often at most a minor component in the dust spectra. In these stars, prominent features in the ISO data have provided convincing evidence for a prominent role played by simple oxides such as Al2 O3 (11 μm) or MgFeO (19.5 μm). The famous 13 μm feature (already detected in IRAS-LRS spectra; see

220

J. A. D. L. BLOMMAERT ET AL.

(Vardya et al., 1986) is attributed to spinel (MgAl2 O4 , Posch et al., 1999; Fabian et al., 2001), although there is still a lively debate about this identification. For instance, Sloan et al. (2003) find no correlation between the 13 μm feature and an observed band at 32 μm which was also presumably due to spinel. For stars with higher mass-loss rates, silicates (both amorphous and crystalline) are the dominant dust component (see Molster and Kemper, this volume). Also the presence of crystalline water-ice and metallic Fe has been demonstrated in these stars. Many more components (especially at longer wavelengths) can be observed in the more evolved stars (post-AGB and PNe, see later sections). For the carbon stars, the situation is quite different. The bulk of the dust is in the form of amorphous carbon, which determines the energy distribution but has no resonances in IR. Most energy distributions of the carbon stars seem to follow a sequence of decreasing TNIR which is interpreted as a cooling and thickening of the circumstellar dust shell. The infrared spectra of carbon stars furthermore show only a small number of solid state bands. Commonly detected dust features include the 11.3 μm feature due to SiC (Gilra, 1973) and the “30 μm feature”, which is believed to be due to MgS (see e.g. Hony et al., 2002a). A particular class of objects are the “mixed chemistry” objects. These are stars that show the presence of typically O-rich and C-rich dust at the same time. In all known cases, the O-rich material is cool and often highly crystalline, while the star itself and/or a warm dust component are typically rich in C. While such objects might represent stars that have only recently become a carbon star, it turns out that most of the well-studied objects in this class belong to a binary system. In such cases, the O-rich material is stored in a disk around the companion star (see e.g. Yamamura et al. (2000) or in a circumbinary disk (such as in the Red Rectangle, see Waters et al. (1998a)). 2.4. STUDIES OF AGB STARS IN DIFFERENT ENVIRONMENTS Mid-infrared studies before ISO were mostly limited to relatively nearby objects; this was especially true for spectroscopical studies. ISO provided the sensitivity and the spatial resolution to investigate AGB stars in the inner regions of the Milky Way galaxy and in the Magellanic Clouds. Studies in these regions offer several advantages, first it allows to study different populations (different metallicities) and second, often the distance is known and thus the luminosities. 2.4.1. The Inner Milky Way The second largest program performed with the ISO satellite, called ISOGAL, is a mid-infrared survey along the galactic plane, performed with ISOCAM (Omont et al., 2003). About 16 square degrees were observed mostly at two wavelengths: 7 and 15 μm. In comparison with IRAS, this survey provides data that are 2 orders of magnitude more sensitive and have 1 order of magnitude higher spatial

LATE STAGES OF STELLAR EVOLUTION

221

Figure 1. ISOGAL [15] vs. (Ks − [15])0 diagram of galactic bulge sources, taken from Ojha et al. (2003). The different symbols are used to indicate sources for which different sorts of associations or data exist. The curves represent sequences of luminosities ranging from 500 to 20,000L  , with increasing mass-loss rates. The approximate scales of mass-loss rates are displayed at the top. For further details on how these were derived (see Ojha et al., 2003).

resolution. A systematic cross-identification with near-infrared (I, J, Ks ) DENIS (Epchtein et al., 1997) sources was performed for the ISOGAL detected sources. This has led to a five-wavelength ISOGAL catalogue, which can be obtained through VizieR. Several papers describe results on AGB stars obtained on fields towards the galactic bulge where the extinction is less strong than along the plane. Figure 1 shows a [15] versus (Ks − [15])0 diagram of bulge ISOGAL sources, taken from Ojha et al. (2003). A clear linear sequence of increasing (Ks − [15])0 colour for brighter 15 μm fluxes can be seen, as was also originally presented in the earlier Omont et al. (1999) paper for the [15] versus [7]–[15] diagram. The increasing red colour of (Ks − [15])0 as function of [15], can be explained by an increasing mass-loss rate. The exact amount of the mass-loss rate is model dependent, but is in the order of 10−9 to a few 10−7 M per year (Ojha et al., 2003) for the sources in the sequence and higher for the redder sources ((Ks − [15])0 > 2.5) which are identified as Mira variables (see the following paragraph). An important point to mention is that the relatively low mass-loss rates observed by the 15 μm excess cannot be observed through the near-infrared colours. Omont et al. (1999) and Glass et al. (1999) demonstrated that the observed sources are red giants above the RGB limit. A comparison by Glass et al. (1999),

222

J. A. D. L. BLOMMAERT ET AL.

and Glass and Schultheis (2002) of ISOGAL data with a spectroscopic survey of Baade’s window NGC 6522 shows that M giants as early as M2 are detected by ISOGAL, but that it is complete from M5 onwards. A further investigation of the circumstellar dust of the mass losing AGB stars comes from ISOCAM Circular Variable Filters measurements. About 20 sources were observed in three bulge fields (5–17 μm). As was expected from the colours, several show clear evidence of a mid-IR excess related to the circumstellar dust (Blommaert et al., 2000, manuscript in preparation). Most sources do not show the typical relatively peaked silicate 9.7 μm feature, but rather a broad feature with a maximum peaking at longer (∼12 μm) wavelengths. Such features are known from previous work on IRAS data (Sloan and Price, 1995, and references therein). They have been associated with amorphous aluminium oxide grains (Onaka et al., 1989; Speck et al., 2000a). Such features are mostly seen in low mass-loss rate sources, confirming the interpretation that ISOGAL is mainly detecting the onset of mass loss on the AGB. Another important characteristic of AGB stars is their variability. Typical examples are Mira and Semi-Regular variables, with periods in the range of hundred to a few hundred days. Mass loss and variability are related in the sense that the longer period and larger amplitude variable stars also show higher mass-loss rates with the extreme examples being the OH/IR stars, which have the longest periods (up to a thousand days) and the highest mass-loss rates (∼10−5 M per year). Alard et al. (2001) and later also Glass and Schultheis (2002) combined the ISOGAL data with data coming from the gravitational lensing experiment MACHO for the Baade windows. In comparison to earlier searches for variable stars (e.g. Lloyd Evans, 1976), the new surveys find variability down to much smaller amplitudes (0.5 mag compared to better than 0.1 mag). Alard et al. (2001) conclude that almost all sources detected at 7 and 15 μm, and, thus, above the RGB limit are variable. A period of 70 days or longer is a necessary but not a sufficient condition for mass loss to occur (Figure 2). ISOGAL also allows to study the AGB stars as tracers for the underlying stellar population. First results are given in van Loon et al. (2003) for the bulge. The ISOGAL survey was also used to find the most extreme mass-loss rate AGB stars near the Galactic Centre. Ortiz et al. (2002) found infrared counterparts for all the objects detected in the OH (1612 MHz) maser surveys of the Galactic Centre. Blommaert (2003) observed the Spectral Energy Distributions of a sample of 26 OH/IR stars in the wavelength range of 2–60 μm. The main results from these studies are that the peak of the bolometric magnitude distribution falls at −5 and that only a relatively small number of stars have high luminosities (and thus less than 1 Gyr old). The OH/IR stars fall predominantly below the extension of the Mira Period–Luminosity relation. Through the modelling of the SEDs, individual total mass-loss rates were determined, which range from 10−6 to 10−4 M per year.

LATE STAGES OF STELLAR EVOLUTION

223

Figure 2. A 15 μm excess (indicative of mass loss) vs. log(Period (days)), taken from (Alard et al., 2001).

2.4.2. The Magellanic Clouds Studying the AGB stars in the Magellanic Clouds has the advantage that the distances and thus the luminosities are known. ISO photometry and spectroscopy of a sample of 57 bright AGB stars and Red SuperGiants in the LMC are presented by Trams et al. (1999b). The sources were selected on basis of IRAS colours indicative of high mass-loss rates. From ISO colour–colour diagrams and the spectra, the spectral type of the dust was determined. About half the sample are classified as carbon stars. No M stars with carbon-rich dust were found, suggesting that Hot Bottom Burning (HBB) cannot efficiently turn back carbon stars into oxygen-rich ones. Mass-loss rates and luminosities for this sample have been determined by van Loon et al. (1999), using a radiative transfer code. The dust-enshrouded carbon stars are generally brighter than the optically bright carbon stars but less bright than the high mass-loss M-type ones. The interpretation given is that only stars between 2 and 4M become carbon stars, whereas more massive stars remain oxygen rich because of HBB. The existence of a few luminous carbon stars may be explained by HBB switching off near the very tip of the AGB and the occurrence of a final thermal pulse thereafter. Connected to this is the first detection of silicate dust around a carbon star (Trams et al., 1999a). In contrast to galactic carbon stars with silicate dust, IRAS04496-6958 is very luminous (Mbol = −6.8). Possibly, the silicate dust is a trace of its oxygen-rich past and it became carbon rich after its last thermal pulse. Most of the supergiants have lower mass-loss rates than the mass consumption rate by nuclear burning. Only two have a much higher mass-loss rate, suggesting that the RSGs lose most of their mantles in very short phases. Also AGB stars have episodes of intensified mass-loss rates, but the mass-loss rate is always higher than the nuclear burning rate.

224

J. A. D. L. BLOMMAERT ET AL.

Blommaert et al. (1999) studied the IRAS sources towards the SMC and find similar results as for the LMC study. There are, however, no carbon stars with luminosities near the AGB limit. All stars suffer extremely high mass-loss rates ˙ > 10−5 M per year), even stars with relatively low masses (∼1M ). (M The earlier studies are based on sources that were already known from the IRAS survey and are thus restricted to the brightest AGB stars. ISOCAM was also used to detect new and lower luminosity AGB stars. Tanab´e et al. (1998) performed a survey of MC clusters and detected very red stars at Mbol = −4.5. High mass-loss rates can thus also be obtained by lower luminosity stars with likely lower initial masses. Another survey with ISOCAM was performed at 4.5, 6.7 and 12 μm on several regions of the Large and Small Magellanic Clouds (Loup et al., 1999). First results were discussed in Cioni et al. (2003), who combined the ISO data with 2MASS and DENIS near-infrared data and light curves extracted from the MACHO database. They confirmed the AGB or supergiant nature of the detected point sources and classified the AGB stars in Mira or Semi-Regular pulsators. 3. Post-AGB Stars The AGB phase of stellar evolution ends in an episode of extremely high mass loss (so-called superwind), in which the mass-loss rate exceeds 10−5 M per year. The high mass loss during the end of AGB evolution creates a circumstellar envelope, which may completely obscure the star. The superwind rapidly strips the star of virtually its entire envelope, thus, effectively terminating the AGB phase. The central star subsequently evolves to higher temperatures on a timescale that depends on its mass and, at the same time, the ejected envelope expands and cools. The star is now in the post-AGB phase and depending on the rate of its evolution, the star may start to photoionize the ejected material, forming a planetary nebula (PN). Post-AGB objects hold the key to understanding how largely spherical AGB circumstellar shells transform into diverse aspherical morphologies (see e.g. Ueta et al., 2000, and references therein). The post-AGB objects that will become planetary nebulae in the near future are called proto-planetary nebulae (PPNe). Since masses of the central stars and, therefore, timescales of their evolution are very uncertain, the term proto-PN usually expresses only our belief that a PN will be created. In this review, however, both terms: post-AGB object and proto-PN will be used equivalently unless there are strong arguments suggesting that a PN will not be formed. For a more detailed discussion of the post-AGB phase of stellar evolution the reader is referred to the Toru´n Conference Proceedings: Post-AGB objects as a phase of stellar evolution (edited by Szczerba and G´orny, 2001) and to the recent reviews by van Winckel (2003) and by Waelkens and Waters (2004). Just like in the previous sections, ISO results related to post-AGB objects with O-rich and C-rich circumstellar shells are discussed separately. Special importance will be given to the signatures of mixed chemistry.

LATE STAGES OF STELLAR EVOLUTION

225

3.1. POST-AGB OBJECTS WITH O-RICH CIRCUMSTELLAR SHELLS As far as dust is concerned, the most important ISO discovery seems to be the detection of crystalline silicates from both young and old stars. Lack of crystalline silicate features in the interstellar medium probably means that after their formation during stars’ death they do not survive until new stars are born. In a series of papers (Molster et al., 2002a,b,c) crystalline silicate features are investigated in the ISO spectra of 17 O-rich circumstellar dust shells surrounding evolved stars. The considered sample contains six post-AGB objects and HD 179821, for which it is yet a matter of debate whether this object is a massive supergiant or a low-mass object in its post-AGB phase of evolution. Apart from the broad 10 and 18 μm bands that are due to amorphous silicates, Molster et al. found at least 49 narrow bands between 10 and 60 μm (divided into seven complexes), which can be attributed to crystalline silicates (olivines and pyroxenes). The richness of the crystalline silicate features in the ISO spectra contains important information related to the conditions in which the grains were formed and processed. It allows detailed studies of the mineralogy of these dust shells. Note, however, that abundances of crystalline silicates are rather low and do not exceed about 10–15% of the total amount of silicates (see Molster, 2000, and references therein). Molster et al. (2002a) found that the crystalline silicate band strengths correlate with the geometry of circumstellar environment: disk (strong) or outflow sources (weak crystalline silicate bands). It has been suggested that in disk sources low temperature crystallization may take place (Molster et al., 1999), while in outflow sources high temperature crystallization in layers that are close to the star is more probable. Among six analysed post-AGB stars, four possibly have disk and two are classified as outflow sources. It is worth to note that two of the disk sources (Red Rectangle and Roberts 22) show also carbon-rich dust as evidenced by the polycyclic aromatic hydrocarbon (PAH) features. The PAHs are predominantly present in the scattering lobes, while the crystalline silicates are expected to be present in the disks (Waters et al., 1998a). A full 2–200 μm ISO spectrum was used to constrain the dust properties in a post-AGB star HD 161796 (Hoogzaad et al., 2002). A good fit to the spectral energy distribution was achieved using four co-spatial but distinct dust components: amorphous silicates (63%), forsterite (4%), enstatite (6%) and crystalline water-ice (27%). The derived temperature of water-ice suggest that the ice must be formed as a mantle on top of an amorphous silicate core, which requires high mass-loss rates, exceeding 5 × 10−5 M per year. The derived mass-loss rate is high enough to allow for this process to be efficient. An interesting example of an OH/IR post-AGB star, still obscured by the ejected material, is IRAS 16342–3814 (OH 344.1+5.8, also called “the water fountain nebula”). This source is classified as a proto-PN (Sahai et al., 1999) and is the only OH/IR star known to have crystalline silicate absorption features seen up to almost 45 μm in its SWS spectrum (Dijkstra et al., 2003b). This suggests that

226

J. A. D. L. BLOMMAERT ET AL.

IRAS 16342−3814 recently had an extremely high mass-loss rate and is perhaps the youngest PPN observed until now. Besides the crystalline silicates (forsterite, diopside and possibly clino-enstatite) identified in the SWS spectrum, crystalline water-ice is also detected. In the LWS spectrum an unidentified feature at 48 μm is seen. A deep and wide water-ice band at 3.05 μm is seen also in another young OH/IR PPN: IRAS 22036 + 5306 (Sahai et al., 2003), together with wide absorption by silicates around 10 and a 60 μm feature, perhaps due to crystalline H2 O-ice (the absence of a corresponding 43 μm H2 O-ice feature may be due to a decline in its strength at low temperatures; Dijkstra et al. (2003a). The freeze-out of water onto dust grains indicated by the ice features requires high densities, which are likely to be characteristic of the dusty disk seen in the Hubble Space Telescope image. In addition, in the SWS spectrum of this object, an unidentified feature at 3.83 μm (tentatively attributed to the H2 S) is detected. A sample composed of O-rich PPNe has been observed with the ISO spectrometers to search for atomic fine structure (FS) lines (Castro-Carrizo et al., 2001). The low-excitation transitions of [O I], [C II], [N II], [Si I], [S I], [Fe I] and [Fe II] were observed. Taking into account the sample of C-rich PPNe presented by Fong et al. (2001) (see Section 3.2), they concluded that PPNe emit in these atomic transitions only when the central star is hotter than about 10 000 K. This result suggests that such lines predominantly arise from photo-dissociation regions (PDRs), and not from the shocked regions.

3.2. POST-AGB OBJECTS WITH C-RICH CIRCUMSTELLAR SHELLS ISO SWS observations of C-rich PPNe have been discussed by Hrivnak et al. (2000) with emphasis given to wavelengths longer than about 20 μm. At these wavelengths, prominent dust features at 21 and around 30 μm are seen. The 21 μm band has been discovered by Kwok et al. (1989) in the IRAS low-resolution spectra of some C-rich PPNe, while the 30 μm band in this class of objects has been detected in the Kuiper Airborne Observatory spectra by Omont et al. (1995). The 21 μm feature is not seen in the spectra of either the precursors to PPNe or the successors of PPNe. Note, however, that Hony et al. (2001b) reported tentative detection of this feature in two planetary nebulae with [WR] central stars, while Pei and Volk (2003) reported its detection in PN IC 418, and Volk et al. (2000) suggested possible presence of this band in two extreme carbon stars. On the other hand, the 30 μm band has been detected first in some C-rich AGB stars and in C-rich planetary nebulae (see Omont et al., 1995, and references therein). Thanks to the ISO observations, Volk et al. (1999) were able to obtain an intrinsic profile of the 21 μm feature, in fact centered at 20.1 μm. This feature has been attributed to various molecular and solid-state species, none of which satisfy all the constraints, although titanium carbide, PAHs and even silicon carbide (see

LATE STAGES OF STELLAR EVOLUTION

227

discussion in Speck and Hofmeister, 2004, and references therein) seem to be the most favoured candidates. On the other hand, in spite of some problems with the SWS data above 26 μm, it seems that the 30 μm feature is not so smooth as the 21 μm one and, instead, shows a kind of substructure (Szczerba et al., 1999). In fact, analysis of the ISO data for other post-AGB sources apparently resolves the 30 μm feature into two components at about 26 and 33 μm (Volk et al., 2002). The derived intrinsic profiles for these features opens a possibility to search for their carrier(s). Recently, Hony et al. (2002a) analysed ISO spectra of C-rich sources including low mass carbon stars, extreme carbon stars, post-AGB objects and planetary nebulae. They modelled the whole range of the extracted “30” μm band profiles by using magnesium sulphide. They argued that in some sources a residual emission at approximately 26 μm can also be fitted using MgS grains but with a different grain shape distribution. The further strengthening of the MgS identification as a carrier of the “30” μm feature comes from detailed modelling of HD 56126, one of the post-AGB “21 μm object” (see Hony et al., 2003). The ISO spectra of C-rich sources show many dust features including Unidentified Infrared Bands (UIRs), which are attributed very often to PAHs (see a comprehensive review by Tielens et al., 1999; Hony et al., 2001a). Besides the well known UIR bands at 3.3, 6.2, 7.7, 8.6 and 11.3 μm, ISO has revealed a wealth variety of weaker features, satellite bands and sub-features. While the final identification of the carrier(s) responsible for these features is still under discussion, there is an agreement that the observed UIR bands are very characteristic for aromatic structures (the C atoms are arranged in planar hexagons – like in graphite – with attached hydrogen atoms at the edges). A detailed analysis of SWS spectra for different kind of objects (Peeters et al., 2002) demonstrated that the UIR emission features in the region 6–9 μm clearly show profile variations. The observed variations in the characteristics of the UIR emission bands are linked to the local physical conditions, but they do not reflect – at least at first glance – a clear link between PAH spectrum and the post-AGB source evolutionary state (see Peeters et al., 2002; Peeters et al., this volume, for details). ISO observations of post-AGB objects demonstrated that there is an evolution of carbonaceous material from aliphatic structures (the C atoms are arranged in a tetrahedral network – like in diamond – with attached H atoms at the edges) to the aromatic structures during the evolution from the proto-planetary to the planetary nebulae phase (Kwok et al., 1999, 2001). AFGL 618 and AFGL 2688 were the first stars (already 30 years ago) to be considered to be in transition from AGB to PN, and they are now the best investigated sources among PPNe. When ISO results are discussed for C-rich PPNe, very often these two objects are given as an example and in this review the situation will be similar. Let us start with polarimetric imaging of AFGL 2688 at 4.5 μm with ISOCAM which allowed to conclude that the polarization of starlight induced by dust grains is almost independent of wavelength between 2 and 4.5 μm (Kastner et al., 2000). This finding indicates that scattering dominates over thermal emission at these wavelengths, and that the dust grains have characteristic radii less than 1 μm.

228

J. A. D. L. BLOMMAERT ET AL.

ISOPHOT linear scans of AFGL 2688 and AFGL 618 demonstrated that both these objects have extremely extended dust shells (see Speck et al., 2000b. Assuming constant expansion velocities, ages for these dust shells are of the order of 105 years. The infrared intensities show “periodic” enhancements, timescales for which (few times 104 years) coincide with thermal pulses on the AGB. ISO spectroscopy allowed also to detect molecular bands in spectra of some C-rich post-AGB objects. For example, Cernicharo et al. (2002) reported the detection of a molecular band at 57.5 μm in PPNe AFGL 618 and AFGL 2688 that has been tentatively assigned to the ν5 bending mode of C4 . Polyynes such as C2 H2 , C4 H2 and C6 H2 , and single aromatic species such as benzene, have been detected in AFGL 618 (Cernicharo et al., 2001a,b). These discoveries support the idea that more complex C-rich molecules could be formed in space. However, the mechanisms allowing the growth of carbon-rich molecules are still poorly known, and the full set of molecules that could participate in the chemical reactions leading to the formation of large complex carbon-rich species has yet to be identified. Analysis of the LWS spectra for the two aforementioned PPNe allowed to detect rotational lines of 12 CO (J = 14–13 to J = 41–40) and lines of 13 CO (J = 14–13 to J = 19–18), as well as HCN and HNC (Justtanont et al., 2000; Herpin et al., 2002). In the early stages, represented by AFGL 2688, the longwave emission is dominated by CO lines. In the more advanced stage (AFGL 618), very fast outflows are present, which, together with the strong UV radiation field from the central star, dissociate CO. The released atomic oxygen allows formation of new O-bearing species, such as H2 O and OH, in a C-rich environment. In case of AFGL 618, several lines of OH and H2 O are detected (Herpin and Cernicharo, 2000; Herpin et al., 2002). In CRL 618 the abundance of HNC is enhanced with respect to HCN as a result of chemical processes occurring in the PDR. HR 4049 was first suggested to be an object in the transition phase from the AGB to PN stage by Lamers et al. (1986) who discovered both a large IR excess and severe UV deficiency, indicating the presence of circumstellar dust. Presently, the source is considered to be the prototype of a group of post-AGB stars in a binary system with extremely metal-depleted atmospheres (van Winckel et al., 1995). The ISO/SWS spectrum of this object shows the signatures of C-rich dust (PAHs; Beintema et al., 1996) and the presence of O-rich gas (isotopomers of CO2 ; Cami and Yamamura, 2001). Analysis of the ISO/SWS spectrum together with other literature data allowed to propose that a very optically thick circumbinary disk could be response for the observed IR emission from HR 4049 (Dominik et al., 2003). A sample composed of C-rich PPNe has been observed with the ISO spectrometers to search for atomic FS lines (Fong et al., 2001). The low-excitation transitions of [O I], [C II], [Si I], [Si II], [S I], [Fe I], [Fe II], [Ne II] and [N II] were observed. However, only few lines in few PPNe were detected.

LATE STAGES OF STELLAR EVOLUTION

229

3.3. OTHER POST-AGB OBJECTS There are two classes of objects that, on the basis of their photospheric abundances, pulsation properties and circumstellar material have been classified as post-AGB stars. They are RV Tau stars and R CrB stars. Both groups of objects are known to have circumstellar dust, giving support to their evolved nature. There are few RV Tau stars observed with ISO. The typical SWS spectrum shows emission bands of amorphous silicates at 10 and 18 μm (see e.g. Szczerba et al., 2003). However, as demonstrated by van Winckel et al. (1998) in case of AC Her, the unusual broad feature at 8–12 μm is due to a high crystallization fraction of the circumstellar silicates. The authors argued that the circumstellar material is trapped in a longlived disk in the system similar to what is observed in the case of the Red Rectangle. Matsuura et al. (2002b) analysed SWS spectra of the RV Tau star, R Scuti and demonstrated that those spectra are dominated by H2 O emission bands with CO, SiO and CO2 bands also present. They argued, however, that R Sct may be a thermally pulsing AGB star, observed in a helium burning phase. Lambert et al. (2001) discussed spectra of some R Coronae Borealis stars. They reported detection of the sharp emission features, which coincides with those of UIRs, only in one of the analysed objects, V 854 Cen. The features coincide with those from laboratory samples of hydrogenated amorphous carbon. Since V 854 Cen, is of order of 1000 more abundant in hydrogen than other typical R CrB stars, the emission features are probably from a carrier containing hydrogen. The extreme H deficiency of the R CrB stars suggests that during their evolution some mechanism removed the entire H-rich stellar envelope. This may be related to a thermal pulse after the star left the AGB. A similar scenario probably applies to “Sakurai’s object”, a star that has experienced dramatic changes during the last decade. The ISO data demonstrate the presence of hot circumstellar dust around this star (Eyres et al., 1998a; Kerber et al., 1999). There is a group of less massive post-AGB objects with central stars of B spectral type (so-called hot post-AGB stars) that have small or even do not have infrared excess at all. One should be aware, however, that a group of Herbig Ae/Be stars (young intermediate-mass pre-main-sequence stars) can be easily confused with such post-AGB candidates. The ISO observations allowed to detect C-rich and Orich dust features in SWS spectra of some hot post-AGB candidates (Gauba and Parthasarathy, 2004), thus indicating the dominant chemistry (C- or O-based) in the circumstellar dust shells and their central stars. Let us finish this part on post-AGB stars with an attempt to classify SWS01 spectra for 61 PPNe by Szczerba et al. (2003). The main dust and/or molecular features were taken into account to propose a division of all spectra into seven classes (four for C-rich and three for O-rich sources). On the basis of their classification, they discussed the connection between post-AGB objects and planetary nebulae with emphasis on possible precursors of PNe with [WR] central stars.

230

J. A. D. L. BLOMMAERT ET AL.

4. Planetary Nebulae and Pre-Planetary Nebulae We have seen earlier in this paper that the terms ‘proto-planetary nebulae’ and ‘post-AGB objects’ are often used interchangeably to denote objects that have left the AGB but have not yet developed significant photo-ionised regions. We include ‘pre-planetary nebulae’ in this group too; the term is sometimes taken to mean that signs of the early development of a circumstellar photoionized region are present for objects in this category. 4.1. PN AND PRE-PN WITH DUAL DUST CHEMISTRIES ISO targeted many O-rich and C-rich objects for spectroscopy using the SWS and LWS instruments; the enormous wavelength coverage (2.4–197 μm) that these two instruments delivered, together with their improved sensitivity relative to NASA’s pioneering Kuiper Airborne Observatory, led to the discovery of many new dust spectral features. One of the big surprises was the relatively high incidence of objects that simultaneously exhibit dust features from both O-rich and C-rich carriers. Ever since its discovery, the Red Rectangle, AFGL 915, has been considered to exhibit the archetypal 3–13 μm UIR-band spectrum (usually attributed to aromatic carbon species, such as PAHs), yet the SWS spectrum of this object shows many crystalline silicate emission features longwards of 20 μm (Waters et al., 1998a). Waters et al. (1998b) found that two planetary nebulae with strong UIR-band spectra, BD+30◦ 3639 and He 2–113, also exhibited a similar array of crystalline silicate emission features longwards of 20 μm. These two nebulae both have carbon-rich nebulae and cool late-type [WCL] WR central stars, as does CPD–56◦ 8032, which Barlow (1997) and Cohen et al. (1999) found to exhibit emission features due to crystalline water-ice, as well as crystalline silicates, in its ISO LWS and SWS spectrum. Cohen et al. (2002) presented the SWS and LWS spectra of more [WCL] PNe showing both strong UIR bands and strong crystalline silicate features, and dubbed them as having ‘dual dust chemistries’. A number of interpretations of the dual dust chemistry phenomenon have been considered by the aforementioned authors, including (a) a recent thermal pulse that has brought up and ejected C-rich material into an O-rich nebula (the phenomenon appears too frequent, however, to be consistent with such dredge-up events, which ought to be encountered only rarely); or (b) the presence of comet clouds at large radii, which have been disrupted by the impact of expanding C-rich nebulae, liberating O-rich silicate and ice particles. However, the most likely explanation now seems to be (c) the presence of pre-existing dust disks around binary systems in these objects, which contain O-rich particles that have been captured and stored from previous O-rich AGB evolutionary stages. Self-shielding disks would enable crystalline silicate and ice particles to remain relatively cool, while C-rich particles synthesised during more recent evolutionary events would be channelled towards polar directions. Direct evidence for the presence of optically thick dust disks around dual dust chemistry

LATE STAGES OF STELLAR EVOLUTION

231

objects has come from the speckle interferometric detection by Osterbart et al. (1997) of an obscuring disk or torus in the Red Rectangle, and the detection of a similar obscuring disk or torus in CPD−56◦ 8032 by De Marco et al. (2002), from HST STIS optical and UV long-slit spectroscopy. As discussed in Section 3.2, a strong 21 μm emission feature is often detected around carbon-rich post-AGB objects. Its non-detection in the spectra of AGB stars and PNe had led to suspicions that the carrier was a transitory species. Hony et al. (2001) have, however, claimed the detection of a 21 μm emission feature in the SWS spectra of two PNe having H-deficient WR central stars, NGC 40 and NGC 6369. Hony et al. (2002b) also reported the detection of a 23 μm emission feature in the SWS spectra of the PNe M2-43 and K3-17, which they attributed to iron sulphide, with M2-43 also showing evidence for weaker FeS features at 34, 38 and 44 μm. M2-43 has a H-deficient WR central star; the nature of the central star in K3-17 is not currently known. Harrington et al. (1998) reported the detection of an unusual 6.4 μm emission feature in the SWS spectra of the PNe Abell 78 and IRAS 151545258 (PM1-89), both of which have H-deficient WC central stars. They identified the feature as an aromatic C C stretch feature produced in carbonaceous grains with little or no hydrogen. It is striking that most of the discoveries discussed earlier have been associated with PNe that have H-deficient WC, WR central stars, a group that accounts for no more than 15% of all PNe. NGC 6302 is a much more evolved, high-excitation PN than the dual dust chemistry [WCL] PNe, with a highly bipolar geometry. Figure 3 shows its continuumsubtracted 2.4–197 μm spectrum (Molster et al., 2001). As well as showing a UIR-band spectrum in the 3–13 μm region (attributed to C-rich particles), longwards of 17 μm the spectrum is seen to be dominated by strong emission features, identified with crystalline silicates and crystalline water-ice (the latter at 44 and 62 μm). There is a very broad feature peaking at about 90 μm in Figure 3, which was not identified by Molster et al., although they suggested that it might be due to hydrous silicates. Kemper et al. (2002b) identified this feature as due to a 92.6 μm band of calcite, CaCO3 . They also suggested that dolomite, CaMg(CO3 )2 , contributes to the observed 62 μm feature. The interest in the calcite identification is that this material normally requires the presence of liquid water for its formation. Water-ice features are identified in the spectrum of NGC 6302, but a non-aqueous formation route would seem to be required for the suggested calcite component in NGC 6302. The relatively narrow emission feature at 69 μm in Figure 3 has been identified with the crystalline silicate material forsterite, Mg2 SiO4 . Bowey et al. (2001) have shown that the peak wavelength of this feature is a sensitive function of temperature for laboratory particle samples, shifting by nearly 1 μm as grains are cooled from 300 to 4 K. Bowey et al. (2002) used these results to derive dust temperatures of 30–140 K from the observed peak wavelengths of the 69 μm feature in the ISO LWS spectra of a range of PNe and post-AGB objects.

232

J. A. D. L. BLOMMAERT ET AL.

Figure 3. The continuum-subtracted spectrum of NGC 6302 from 2.4 to 197 μm. As well as many ionic emission lines, the 3–13 μm region is characterised by the presence of C-rich emission features, while from 17 to 80 μm the spectrum is dominated by emission bands attributable to crystalline silicates and crystalline water-ice (from Molster et al., 2001).

4.2. THE NEUTRAL ZONES AROUND PN Many planetary nebulae have surrounding neutral atomic and molecular zones that emit strongly in the ISO spectral range. Their PDRs are sites of active chemistry, allowing a number of transient species to be produced and detected, in addition to the well known IR lines of CO and H2 . In addition to the 11 rotational lines of CO, Liu et al. (1996) reported the detection of the 119.3 μm fundamental rotational line of OH from the PDR around NGC 7027, in a high signal to noise LWS spectrum. They also detected an emission line at 179.6 μm, which they identified with the fundamental rotational line of o-H2 O. However, a re-analysis of the data by Cernicharo et al. (1997) led to the re-allocation of this line to the J = 2–1 rotational transition of CH+ , supported by the detection in the same LWS spectrum of the 3–2, 4–3, 5–4 and 6–5 transitions of CH+ , at 119.90, 90.03, 72.140 and 60.247 μm, the first detection of the rotational spectrum of CH+ . Liu et al. (1997) reported the detection in the NGC 7027 LWS spectrum of emission lines at 149.18 and 180.7 μm which matched the wavelengths of CH rotational lines and represented the first detection

LATE STAGES OF STELLAR EVOLUTION

233

of the far-IR lines of this radical. Detailed models for the PDR around NGC 7027 have been constructed by Yan et al. (1999) and by Hasegawa et al. (2000), and have provided good matches to the observed LWS rotational line fluxes from CO, OH, CH and CH+ . Liu et al. (2001) determined the temperatures and densities of the PDRs around 24 PNe, from the observed line intensity ratios of [C II] 158 μm and [O I] 63 and 146 μm, with the deduced temperatures typically falling between 200 and 500 K and the densities ranging from 104 to 3×105 cm−3 . With a temperature of 1600 K, NGC 7027 had one of the warmest, as well as one of the densest, PDRs studied. NGC 7027 is an example of a young, high-density and incompletely ionised PN. Turning to the other extreme of PN evolution, the old, low-density PN NGC 7293 (the Helix Nebula) was imaged by Cox et al. (1998) in the ISOCAM 6.9 and 15 μm filters. Their complementary ISOCAM CVF spectra revealed that the emission in the 6.9 μm image was completely dominated by the S(5) line of H2 and that no dust continuum or PAH UIR-band emission was detectable. The H2 line intensity distribution could be fitted by a rotational temperature of 900 K and column densities of ∼3×1018 cm−2 , while the total luminosity in H2 lines amounted to 6% of the central star luminosity, much higher than predicted for PDRs. The H2 emission appears to originate from the many neutral globules in the Helix Nebula that have been imaged at optical wavelengths and in mm-wave CO lines. Persi et al. (1999) have presented ISOCAM images and CVF spectrophotometry of a further six PNe, ranging from high-density to low-density objects.

4.3. THE IONISED REGIONS OF PN Liu et al. (2001) presented ISO LWS [O III] 52 and 88 μm, [N III] 57 μm and [N II] 122 μm fluxes for a sample of 51 PNe. The electron densities derived from the [O III] 52/88 μm flux ratios were found to be systematically lower than those obtained for the same nebulae from higher critical density optical diagnostic line ratios, consistent with the presence of significant density variations within the nebulae, with the [O III] 52 and 88 μm lines (critical densities of 3500 and 1500 cm−3 , respectively) being quenched in the higher density nebular zones. However, Tsamis et al. (2004) have shown that for those nebulae for which the IR and optical density diagnostic ratios all indicate electron densities below the critical densities of the [O III] IR lines, the electron temperatures derived from the [O III] λ5007/λ4363 ratio and from the λ5007/(52 μm + 88 μm) ratio agree well with each other, ruling out the presence of significant temperature fluctuations in these nebulae, since the IR lines of [O III] have much lower excitation energies than do the optical lines such as λ4363. The SWS spectral region contains many ionic FS lines and a number of papers have been published presenting FS line fluxes and nebular abundance analyses for 18 different PNe (Beintema and Pottasch, 1999; Pottasch and Beintema, 1999; Pottasch

234

J. A. D. L. BLOMMAERT ET AL.

et al., 2000, 2001, 2002, 2003a,b, 2004; Bernard-Salas et al., 2001, 2002, 2003; Ercolano et al., 2003; Surendiranath et al., 2004). These analyses take advantage of the low excitation energies of IR FS lines and thus the relative insensitivity of the derived ionic abundances to the adopted nebular electron temperature. If the particular FS line has a critical density below a few ×104 cm−3 , i.e. within the typical electron density range for PNe, then the derived ionic abundance is quite sensitive to the adopted electron density, but many FS lines in the SWS domain have critical densities above this range and so make good abundance diagnostics. Marigo et al. (2003) have compared the abundances derived in many of the earlierreferenced papers to synthetic evolutionary models for the AGB phase, confirming the presence of ‘two groups of PNe, one indicating the occurrence of only the third dredge-up during the TP-AGB phase, and the other showing also the chemical signature of hot-bottom burning’. van Hoof et al. (2000) have used ISO SWS measurements of [Ne V] infrared FS lines to re-assess the accuracy of published collision strengths for the Ne4+ ion, concluding that the R-matrix calculations of Lennon and Burke (1994), which had appeared to conflict with some earlier observations, were accurate to at least 30%. Rubin et al. (2002) discussed the observed ratios of a number of density sensitive FS line ratios falling in the SWS domain and showed that several of the measured ratios for PNe were outside the range predicted using current atomic data (including the [Ne V] infrared FS lines discussed by van Hoof et al.), implying the need for improved atomic collision strengths. Feuchtgruber et al. (1997) used ISO SWS grating and Fabry–P´erot observations of several PNe to measure improved rest wavelengths for 29 infrared FS lines, while Feuchtgruber et al. (2001) provided similar data for a further six FS lines.

5. Observations of Novae Many of the same ionic FS lines observed in planetary nebulae can also be observed in the spectra of novae, and ISO spectroscopy of these lines has been able to reveal a great deal of information about the ionised gas component of nova shells. Salama et al. (1996) reported SWS and LWS observations of Nova V1974 Cyg 1992, obtained 1494 days after the outburst. From measured FS line flux ratios, they estimated electron temperatures in the ejecta and derived an enhanced Ne/O ratio of ∼4. Salama et al. (1997) presented further ISO SWS observations of Nova V1974 Cyg 1992, including detailed line profiles of the [Ne V] 14.32 and 24.32 μm lines at four different epochs. The density sensitive [Ne V] line ratio was found to decrease with time t as t −3 , in accord with ejecta expansion expectations. Salama et al. (1997) also presented ISO spectra of Nova HR Del 1967 and Nova V705 Cas 1993. Further ISO observations of the latter nova, obtained between 950 and 1455 days after outburst, were presented by Salama et al. (1999). The [O IV] line at 25.89 μm was found to be strongly in emission. Upper limits to the fluxes of neon

LATE STAGES OF STELLAR EVOLUTION

235

FS lines led to an upper limit of 0.5 solar for the Ne/O number ratio in the nova shell. As well as obtaining observations of older novae, a nova Target of Opportunity programme was carried out during the 1995–1998 duration of the ISO mission, which was combined with co-ordinated complementary ground-based optical and near-IR spectroscopic observations. The ToO programme was unlucky in that no bright dust-forming novae occurred during the mission. However, the novae that were observed produced a rich harvest of information on second and third row elemental abundance distributions with which to compare with explosive nucleosynthesis predictions. Lyke et al. (2001) presented ISO, IUE and ground-based data on Nova V1425 Aql 1995 and combined these with a detailed photoionization model for the evolving nova shell. The mass of the shell was derived to be 2.5–4.3×10−5 M , with the modelling indicating that C and O were enhanced by a factor of 9 with respect to solar, while N was enhanced by a factor of 100 and Ne was only slightly enhanced. The presence of a CO white dwarf in the system was confirmed and a distance estimate of 3.0 ± 0.4 kpc to the nova was obtained. From their SWS IR and AAT optical and near-IR spectra of the heavily reddened Nova CP Cru 1996, Lyke et al. (2003) derived a distance of 2.6 ± 0.5 kpc to the system. Relative to solar, they derived abundance enhancements for N, O and Ne of 75, 17 and 27, with Mg showing an approximately solar abundance. The enhanced neon abundance but relatively low Mg abundance led them to interpret this object as a ‘missing link’ between CO novae and ONeMg novae. Nova V723 Cas 1995 was observed at six different epochs by ISO, from 217 to 805 days after the outburst, showing prominent highly ionised coronal FS line emission (Evans et al., 2003). From the flux ratios of hydrogen recombination lines having similar wavelengths, they derived reddening insensitive electron temperature and density estimates, finding Ne (cm−3 ) ∼ 2 × 108 [250/t (days)], where t was the time since outburst. For a distance of 4 kpc, they estimated an ejected shell mass of between 2.6×10−5 and 4.3×10−4 M . No lines of neon were detected, excluding V723 Cas from the class of ONeMg novae and implying that it has a CO white dwarf core.

6. Symbiotic Stars Symbiotic stars are long-period binaries that generically contain a cool red giant star and a hot, evolved object that produces ionisation in the circumstellar outflow. The hot component of the vast majority of symbiotic systems is typically a hot white dwarf orbiting close enough to the red giant that it can accrete material from its wind. There are two distinct classes of symbiotic systems: the S-type (stellar) with normal red giants and orbital periods of about 1–15 years, and D-type (dusty) with Mira primaries usually surrounded by a dust shell, and orbital periods longer than about 10 years (Nussbaumer and Stencel, 1988). Symbiotic stars are interacting

236

J. A. D. L. BLOMMAERT ET AL.

binaries with the longest orbital periods, and their study is essential to understand the evolution and interaction of binary systems. IRAS measurements of the brighter symbiotic stars (most symbiotics were only detected with IRAS as mili-Jy sources), indicated the presence of dust shells with characteristic temperatures of 250–800 K and dimensions of tens of AU (Anandarao et al., 1988). These warm shells are surrounded by more extended ones that were detected recently in sub-mm wavelengths ((Mikolajewska et al., 2002), and references therein). Some symbiotic systems were observed with ISO and the results from the published observations are discussed later. The majority of symbiotics were however too faint for ISO and are in fact ideally suited for the Spitzer Space Telescope. Preliminary results from ISO observations of symbiotic stars were presented by Eyres et al. (1998b). ISO SWS observations of the symbiotic novae (or very slow novae) RR Tel and V1016 Cyg reveal prominent, broad 10 and 18 μm silicate dust features. In both cases, the 10 μm feature is well fitted by the crystalline silicate feature found in novae Her 91. There is some observational evidence that the silicate feature at 10 μm show time-dependent variations. In addition, SWS spectra show a number of narrow emission lines, which can be used to constrain abundances. ISOPHOT observations were made for the S-type symbiotics AG Dra and AG Peg, providing multi-filter photometry and spectrophotometry. The observed emission is dominated by the giant-component continuum. There are some additional features: a broad plateau between 2.47 and 3.0 μm; a broad feature around 3.2 μm (AG Dra) and at 3.8 μm (AG Peg); a clear excess between 10 and 15 μm in both cases. Note, however, that some of these features could be of instrumental origin. Schild et al. (1998) discussed spectral features in SWS and LWS spectra of CH Cyg. Much of the ISO wavelength range is dominated by the emission of silicate dust with almost no nebular emission lines (at the time of observations the symbiotic activity of CH Cygni was small). In addition, they find strong OH and weak H2 O emission between 60 and 130 μm. Note that OH molecules are absent in spectra of single giants of similar spectral type. Apart from the well known photospheric absorptions of CO (the band heads of 12 CO and 13 CO are well discernible), OH and SiO, they also tentatively detected traces of HCl and relatively weak PAH features at 6.3 and 11.3 μm. The authors suggested that carbon-rich material may be ejected from the outbursting companion. Schild et al. (2001) presented and discussed SWS and LWS observations of the symbiotic star HM Sge. The emission from this star is dominated by silicate dust with a number of nebular emission lines superimposed on the dust continuum. In the short wavelengths, there are no molecular absorption features that suggest the presence of warm dust. Analysis of the overall spectral energy distribution of HM Sge allowed to suggest that in addition to a Mira dust shell there is a second dust component associated with the hot source. This second shell is located at the interface between the Mira dust shell and the dust-free region carved out by the hot companion. The dust envelope associated with the Mira is very extended and

LATE STAGES OF STELLAR EVOLUTION

237

optically thin, whereas the second dust component is geometrically rather thin but dust rich. The general agreement of the presented model with ISO observations suggest that radiation (not collisions) is the main dust heating mechanism. There are still ISO spectra available (but probably of lower quality than the one discussed earlier) for some symbiotic stars which show many nebular emission lines and can be used together with optical data to better constrain chemical abundances in ionised parts of these systems.

7. Wolf-Rayet Stars Wolf-Rayet stars represent one of the final stages of stellar evolution for massive stars. In general, they are thought to follow an evolutionary sequence divided into three phases: WN, WC and WO, which correspond to dominant emission lines in their spectra (see van der Hucht, 2001 for a review). These emission lines are formed in the stars’ hot winds (typical mass-loss rates are between 10−6 and 10−4 M per year) and have velocities in the range between 1000 and 2500 km/s. The WR phase is predicted to be brief, lasting only a few 105 years before ending in Type II supernova explosions. Despite their harsh environment, some WC-type WR stars are known to be surrounded by circumstellar dust (e.g. Williams, 1995). Therefore, ISO spectroscopic observations were suitable for both: to study chemical composition of ejected and ionised gas, as well as dust spectral features. Preliminary results from ISO SWS observations of seven WR stars were presented by van der Hucht et al. (1996). They discussed Ne and He abundances in WR 11 on the basis of SWS 04 spectra, and dust features in WC 8–9 stars: WR 48a, WR 98a, WR 104, WR 112 and WR 118 from medium resolution SWS 02 observations. Their target stars, except for WR 11, are heavily reddened and SWS spectra show absorptions by silicates and hydrocarbons, possibly of interstellar origin. Still preliminary results were discussed in more detail during “ISO’s view on stellar evolution” conference held in Noordwijkerhout in July, 1997 (Morris et al., 1998; Willis et al., 1998; Williams et al., 1998). One of the problems considered there was the neon abundance. Stellar evolution models predict an enormous enhancement in surface Ne abundances in WC stars, making it the fourth most abundant behind He, C and O in the WC stars. Before the launch of ISO, observations suggested that Ne/H is enhanced by a factor of 2 rather than >10 as models predicted (Barlow et al., 1988). This conflict was a major puzzle in WR research. SWS observations of several galactic WR stars have been obtained to provide tests of this controversy. Analysis of the SWS spectra showed that the neon abundance is rather close to that predicted by stellar evolution models in the cases of WR 146 and γ 2 Vel (WR 11), but rather lower than predicted in the case of WR 135. Surprisingly, an overabundance of Ne was found in a WN8 star, WR 147, and rotational mixing has been proposed as a plausible explanation. The WN8(h) + B0.5 V binary system WR 147 has been further analysed by Morris et al. (2000). Using SWS data,

238

J. A. D. L. BLOMMAERT ET AL.

they were able to put constraints on the surface abundances of Ca, S and Ne which were in a good agreement with the solar values of these elements for the H-depleted environment of WR 147. A detailed analysis of the Ne abundance problem in WC stars, based on the SWS and ground-based data, was presented further by Willis et al. (1997) for WR 146 and improved by Dessart et al. (2000), who considered also other WC stars: WR 11, WR 90 and WR 135. They estimated that Ne/He = 3–4×10−3 in the case of WR 11 and put an upper limit of the same order on the three remaining WC stars. Neon is highly enriched, but the observed Ne/He ratios are still a factor of 2 lower than the predictions of current evolutionary models for massive stars (see e.g. Maeder, 1991; Meynet, 1999). They also suggested that an imprecise mass loss and distance were responsible for the much greater discrepancy in neon content identified by Barlow et al. (1988). Note that Smith and Houck (2001) have obtained groundbased mid-infrared (8–13 μm) spectra for a sample of galactic WR stars which, except for an unresolved discrepancy for WR 146, show agreement with the SWS observations. It is worth mentioning also the possibility to determine terminal speeds of winds in three single WN stars using [Ca IV] 3.207 μm line widths from SWS data for these stars (Ignace et al., 2001). A detailed analysis of these line profiles allowed them also to conclude that the effect of turbulence in the winds of single WN stars is significant. They also showed that the line profile shapes in WR binaries appear asymmetric. In addition, using SWS data, Ignace et al. (2003) were able to put constraints on the velocity law and clumpiness of the wind from the WN-type star WR 136. As far as dust in H-poor WC stars is concerned, there are two principal questions: how dust is formed near such hot stars and the apparent failure to observe in WR stars any of the expected precursors (such as C-chain molecules) of the C-rich dust. SWS observations of circumstellar dust around WR stars reveal a variety of spectral energy distributions (SEDs). The mid-infrared SEDs of circumstellar shells around late-WC stars are approximately Planckian, with strong interstellar absorption features (amorphous silicates, CO2 -ice at 4.27 μm, and the 6.2 μm feature to which molecules with the C double bond may contribute). The results of SED modelling by Williams et al. (1998) allowed them to examine the contributions made by different dust grain types to the observed emission, as well as the relative contributions made by circumstellar and interstellar components to the observed reddening. The authors demonstrated that AC-type amorphous carbon (Colangeli et al., 1995) gave fairly good fits to the observed SEDs. From pre-ISO observations with the KAO of two dusty WC9 stars, Cohen et al. (1989) had reported the presence of the 7.7 μm but the absence of the 6.2 μm unidentified infrared features (frequently attributed to C C stretch modes of PAHs), while Chiar et al. (2002) reported the detection of features at ∼6.4 and 7.9 μm in the ISO SWS spectrum of the dusty WC8 WR star WR 48a. Since the atmospheres of these stars are H poor, the detected features are believed to be due to large

LATE STAGES OF STELLAR EVOLUTION

239

hydrogen-poor carbonaceous molecules or amorphous carbon dust grains and not PAHs. Interestingly, the features in the spectrum of WR 48a resemble the features observed in the spectra of H-deficient planetary nebulae. These similarities point towards a similar origin for the dust in these H-deficient environments and highlight the apparent sensitivity of the bands to local physical conditions.

References Alard, C., Blommaert, J. A. D. L., Cesarsky, C., et al. (ISOGAL and MACHO Collaborations), 2001, ApJ 552, 289. Anandarao, B. G., Taylor, A. R., and Pottasch, S. R.: 1988, A&A 203, 361. Aoki, W., Tsuji, T., and Ohnaka, K.: 1998, A&A 340, 222. Aoki, W., Tsuji, T., and Ohnaka, K.: 1999, A&A 350, 945. Barlow, M. J., Nguyen-Q-Rieu, Truong-Bach, Cernicharo, J., Gonzalez-Alfonso, E., et al.: 1996, A&A 315, L241. Barlow, M. J., Roche, P., and Aitken, D. K.: 1988, MNRAS 232, 821. Barlow, M. J.: 1997, ApSS 251, 15. Beintema, D. A., van den Ancker, M. E., Molster, F. J., et al.: 1996, A&A 315, L369. Beintema, D. A. and Pottasch, S. R.: 1999, A&A 347, 942. Bernard Salas, J., Pottasch, S. R., Beintema, D. A., and Wesselius, P. R.: 2001, A&A 367, 949. Bernard Salas, J., Pottasch, S. R., Feibelman, W. A., and Wesselius, P. R.: 2002, A&A 387, 301. Bernard Salas, J., Pottasch, S. R., Wesselius, P. R., and Feibelman, W. A.: 2003, A&A 406, 165. Blommaert, J. A. D. L., Trams, N. R., Groenewegen, M. A. T., et al.: 1999, in: Cox, P. and Kessler, M. F. (eds.), The Universe as Seen by ISO. ESA-SP 427, p. 273. Blommaert, J. A. D. L., Omont, A., and The ISOGAL Collaboration: 2000, Mem. Soc. Astron. Ital. 71, 623. Blommaert, J. A. D. L.: 2003, ASSL Vol. 283: Mass-Losing Pulsating Stars and their Circumstellar Matter, p. 355. Bowey, J. E., Lee, C., Tucker, C., Hofmeister, A. M., Ade, P. A. R., and Barlow, M. J.: 2001, MNRAS 325, 886. Bowey, J. E., Barlow, M. J., Molster, F. J., et al.: 2002, MNRAS 331, L1. Cami, J. and Yamamura, I.: 2001, A&A 367, L1. Cami, J.: 2002, PhD Thesis, University of Amsterdam. Cami, J.: 2003, in: ASSL Vol. 283: Mass-Losing Pulsating Stars and their Circumstellar Matter, p. 209. Cami, J., Yamamura, I., de Jong, T., Tielens, A. G. G. M., Justtanont, K., and Waters, L. B. F. M.: 2000, A&A 360, 562. Castro-Carrizo, A., Bujarrabal, V., Fong, D., et al.: 2001, A&A 367, 674. Cernicharo, J., Liu, X.-W., Gonzalez-Alfonso, E., et al.: 1997, ApJ 483, L65. Cernicharo, J., Heras, A. M., Pardo, J. R., et al.: 2001a, ApJ 546, L127. Cernicharo, J., Heras, A. M., Tielens, A. G. G. M., et al.: 2001b, ApJ 546, L123. Cernicharo, J., Goicoechea, J. R., and Benilan, Y.: 2002, ApJ 580, L157. Chiar, J. E., Peeters, E., and Tielens, A. G. G. M.: 2002, ApJ 579, L91. Cioni, M.-R. L., Blommaert, J. A. D. L., Groenewegen, M. A. T., et al.: 2003, A&A 406, 51. Cohen, M., Tielens, A. G. G. M., and Bregman, J. D.: 1989, ApJ 344, L13. Cohen, M., Barlow, M. J., Sylvester, R. J., et al.: 1999, ApJ 513, L135. Cohen, M., Barlow, M. J., Liu, X.-W., and Jones, A.: 2002, MNRAS 332, 879. Cox, P., Boulanger, F., Huggins, P. J., et al.: 1998, ApJ 495, L23.

240

J. A. D. L. BLOMMAERT ET AL.

Colangeli, L., Menella, V., Palumbo, P., et al.: 1995, A&AS 113, 561. Decin, L., Vandenbussche, B., Waelkens, C., et al.: 2003, A&A 400, 709. De Marco, O., Barlow, M. J., and Cohen, M.: 2002, ApJ 574, L83. Dessart, Luc, Crowther, P. A., Hillier, D. J., et al.: 2000, MNRAS 315, 407. Dijkstra, C., Dominik, C., Hoogzaad, S. N., et al.: 2003a, A&A 401, 599. Dijkstra, C., Waters, L. B. F. M., Kemper, F., et al.: 2003b, A&A 399, 1037. Dominik, C., Dullemond, C. P., Cami, J., and van Winckel, H.: 2003, A&A 397, 595. Epchtein, N., de Batz, B., Capoani, L., et al.: 1997, Messenger 87, 27. Ercolano, B., Morisset, C., Barlow, M. J., Storey, P. J., and Liu, X.-W.: 2003, MNRAS 340, 1153. Evans, A., Gehrz, R. D., Geballe, T. R., et al.: 2003, AJ 126, 1981. Eyres, S. P. S., Evans, A., Geballe, T. R., Salama, A., and Smalley, B.: 1998a, MNRAS 298, L37. Eyres, S. P. S., Evans, A., Salama, A., et al.: 1998b, Astrophys. Space Sci. 255, 361. Fabian, D., Posch, T., Mutschke, H., Kerschbaum, F., and Dorschner, J.: 2001, A&A 373, 1125. Feuchtgruber, H., Lutz, D., Beintema, D. A., et al.: 1997, ApJ 487, 962. Feuchtgruber, H., Lutz, D., and Beintema, D. A.: 2001, ApJS 136, 221. Fong, D., Meixner, M., Castro-Carrizo, A., et al.: 2001, A&A 367, 674. Gauba, G. and Parthasarathy, M.: 2004, A&A 417, 201. Gautschy-Loidl, R., H¨ofner, S., Jørgensen, U. G., and Hron, J.: 2004, A&A 422, 289. Gilra, D. P.: 1973, in: IAU Symposium 52: Interstellar Dust and Related Topics, p. 517. Glass, I. S., Ganesh, S., Alard, C., et al.: 1999, MNRAS 308, 127. Glass, I. S. and Schultheis, M.: 2002, MNRAS 337, 519. Harrington, J. P., Lame, N. J., Borkowski, K. J., Bregman, J. D., and Tsvetanov, Z. I.: 1998, ApJ 501, L123. Harris, G. J., Pavlenko, Y. V., Jones, H. R. A., and Tennyson, J.: 2003, MNRAS 344, 1107. Hasegawa, T., Volk, K., and Kwok, S.: 2000, ApJ 532, 994. Herpin, F. and Cernicharo, J.: 2000, ApJ 530, L129. Herpin, F., Goicoechea, J. R., Pardo, J.R., and Cernicharo, J.: 2002, ApJ 577, 961. H¨ofner, S.: 1999, A&A 346, L9. Hony, S., Bouwman, J., Keller, L. P., and Waters, L. B. F. M.: 2002b, A&A 393, L103. Hony, S., Van Kerckhoven, C., and Tielens, A. G. G. M., et al.: 2001a, A&A 378, L41. Hony, S., Waters, L. B. F. M., and Tielens, A. G. G. M.: 2001b, A&A 378, L41. Hony, S., Waters, L. B. F. M., and Tielens, A. G. G. M.: 2002a, A&A 390, 533. Hony, S., Tielens, A. G. G. M., Waters, L. B. F. M., and de Koter, A.: 2003, A&A 402, 211. Hoogzaad, S. N., Molster, F. J., Dominik, C., et al.: 2002, A&A 389, 547. Hrivnak, B., Volk, K., and Kwok, S.: 2000, ApJ 535, 275. Hron, J., Loidl, R., H¨ofner, S., Jørgensen, U. G., Aringer, B., Kerschbaum, F., et al.: 1999, in: IAU Symposium 191: Asymptotic Giant Branch Stars, p. 181. Ignace, R., Cassinelli, J. P., Quigley, M., and Babler, B.: 2001, ApJ 558, 771. Ignace, R., Quigley, M., and Cassinelli, J. P.: 2003, ApJ 596, 538. Jones, H. R. A., Pavlenko, Y., Viti, S., and Tennyson, J.: 2002, MNRAS 330, 675. Jørgensen, U. G., Hron, J., and Loidl, R.: 2000, A&A 356, 253. Jørgensen, U. G., Jensen, P., Sørensen, G. O., and Aringer, B.: 2001, A&A 372, 249. Justtanont, K., Barlow, M., Tielens, A. G. G. M., et al.: 2000, A&A 360, 1117. Justtanont, K., de Jong, T., Tielens, A. G. G. M., Feuchtgruber, H., and Waters, L. B. F. M.: 2004, A&A 417, 625. Justtanont, K., Feuchtgruber, H., de Jong, T., Cami, J., Waters, L. B. F. M., Yamamura, I., et al.: 1998, A&A 330, L17. Kastner, J. H., Li, J.-Q., Siebenmorgen, R., and Weintraub, D. A.: 2000, AJ 123, 2658. Kemper, F., de Koter, A., Waters, L. B. F. M., Bouwman, J., and Tielens, A. G. G. M.: 2002a, A&A 384, 585.

LATE STAGES OF STELLAR EVOLUTION

241

Kemper, F., Jager, C., Waters, L. B. F. M., et al.: 2002b, Nature 415, 295. Kerber, F., Blommaert, J. A. D. L., Groenewegen, M. A. T., Kimeswenger, S., Käufl, H. U., and Asplund, M.: 1999, A&A 350, L27. Kwok, S., Volk, K., and Hrivnak, B.: 1989, ApJ 345, L51. Kwok, S., Volk, K., and Hrivnak, B.: 1999, A&A 350, L35. Kwok, S., Volk, S., and Bernath, P.: 2001, ApJ 554, L87. Lambert, D. L., Rao, K. N., Pandey, G., and Ivans, I. I.: 2001, ApJ 555, 925. Lamers, H., Waters, L. B. F. M., Garmany, C., et al.: 1986, A&A 154, L20. Lennon, D. J. and Burke, V. M.: 1994, A&AS 103, 273. Liu, X.-W., Barlow, M. J., Nguyen-Q-Rieu, et al.: 1996, A&A 315, L257. Liu, X.-W., Barlow, M. J., Dalgarno, A., et al.: 1997, MNRAS 290, L71. Liu, X.-W., Barlow, M. J., Cohen, M., et al.: 2001, MNRAS 323, 343. Lloyd Evans, T.: 1976, MNRAS 174, 169. Loup, C., Cioni, M. R., Blommaert, J. A. D. L., et al.: 1999, in: Cox, P. and Kessler, M. F. (eds.), The Universe as Seen by ISO. ESA-SP 427, p. 369. Lyke, J. E., Gehrz, R. D., Woodward, C. E., et al.: 2001, AJ 122, 3305. Lyke, J. E., Koenig, X. P., Barlow, M. J., et al.: 2003, AJ 126, 993. Maeder, A.: 1991, A&A 242, 93; A&A 315, L241. Marigo, P., Bernard-Salas, J., Pottasch, S. R., Tielens, A. G. G. M., and Wesselius, P. R.: 2003, A&A 409, 619. Markwick, A. and Millar, T.: 2000, A&A 359, 1162. Matsuura, M., Yamamura, I., Cami, J., Onaka, T., and Murakami, H.: 2002a, A&A 383, 972. Matsuura, M., Yamamura, I., Zijlstra, A. A., and Bedding, T.R.: 2002b, A&A 387, 1022. Meynet, G.: 1999, in: van der Hucht, K. A., Koenigsberger, G., and Eenens, P. R. J. (eds.), Wolf-Rayet Phenomena in Massive Stars and Starburst Galaxies, IAU Symposium 193, ASP, San Francisco, p. 218. Mikolajewska, J., Ivison, R. J., and Omont, A.: 2002, Adv. Space Res. 30, 2045. Molster, F. J.: 2000, ‘Crystalline Silicates in Circumstellar Dust Shells’, PhD Thesis, University of Amsterdam. Molster, F. J., Lim, T. L., Sylvester, R. J., et al.: 2001, A&A 372, 165. Molster, F. J., Waters, L. B. F. M., Tielens, A. G. G. M., and Barlow, M. J.: 2002a, A&A 382, 184. Molster, F. J., Waters, L. B. F. M., and Tielens, A. G. G. M.: 2002b, A&A 382, 222. Molster, F. J., Waters, L. B. F. M., Tielens, A. G. G. M. et al.: 2002c, A&A 382, 241. Molster, F. J., Yamamura, I., Waters, L. B. F. M., et al.: 1999, Nature 401, 563. Morris, P. W., van der Hucht, K. A., Crowther, P. A., et al.: 2000, A&A 353, 624. Morris, P. W., van der Hucht, K. A., Willis, A. J., and Williams, P. M.: 1998, in: Waters, L. B. F. M., Waelkens, C., van der Hucht, K. A., and Zaal, P. A. (eds.), ISO’s View on Stellar Evolution, Astrophysics and Space Science 255, Kluwer Academic Publishers, Dordrecht/Boston/Leiden, p. 157. Neufeld, D. A., Chen, W., Melnick, G. J., de Graauw, T., Feuchtgruber, H., Haser, L., et al.: 1996, A&A 315, L237. Nussbaumer, H. and Stencel, R.: 1987, in: Exploring the Universe with the IUE Satellite (A88-22626 07-90), D. Reidel Publishing Co., Dordrecht, p. 203. Ojha, D. K., Omont, A, Schuller, F., et al.: 2003, A&A 403, p. 141. Omont, A., Moseley, S. H., Cox, P., et al.: 1995, ApJ 454, 819. Omont, A., Ganesh, S., Alard, C., Blommaert, J. A. D. L., Caillaud, B., et al.: 1999, A&A 348, 755. Omont, A., Gilmore, G. F., Alard, C., et al.: 2003, A&A 403, 975. Onaka, T., de Jong, T., and Willems, F. J.: 1989, A&A 218 169. Ortiz, R., Blommaert, J. A. D. L., Copet, E., et al.: 2002, A&A 388, 279. Osterbart, R., Langer, N., and Weigelt, G.: 1997, A&A 325, 609.

242

J. A. D. L. BLOMMAERT ET AL.

Peeters, E., Hony, S., Van Kerckhoven, C., et al.: 2002, A&A 390, 1089. Pei, A. and Volk, K.: 2003, in: Kwok, S., Dopita, M., and Sutherland, R. (eds.), Planetary Nebulae: Their Evolution and Role in the Universe, Astronomical Society of the Pacific. Persi, P., Cesarsky, D., Marenzi, A. R., et al.: 1999, A&A 351, 201. Posch, T., Kerschbaum, F., Mutschke, H., Dorschner, J., and J¨ager, C.: 2002, A&A 393, L7. Posch, T., Kerschbaum, F., Mutschke, H., Fabian, D., Dorschner, J., and Hron, J.: 1999, A&A 352, 609. Pottasch, S. R. and Beintema, D. A.: 1999, A&A 347, 975. Pottasch, S. R., Beintema, D. A., and Feibelman, W. A.: 2000, A&A 363, 767. Pottasch, S. R., Beintema, D. A., Bernard Salas, J., and Feibelman, W. A.: 2001, A&A 380, 684. Pottasch, S. R., Beintema, D. A., Bernard Salas, J., Koornneef, J., and Feibelman, W. A.: 2002, A&A 393, 285. Pottasch, S. R., Hyung, S., Aller, L. H., et al.: 2003a, A&A 401, 205. Pottasch, S. R., Bernard-Salas, J., Beintema, D. A., and Feibelman, W. A.: 2003b, A&A 409, 599. Pottasch, S. R., Bernard-Salas, J., Beintema, D. A., and Feibelman, W. A.: 2004, A&A 423, 593. Rubin, R. H., Dufour, R. J., Colgan, S. W. J., et al.: 2002, RMexAC 12, 106. Ryde, N., Eriksson, K., and Gustafsson, B.: 1999, A&A 341, 579. Sahai, R., Te Lintel Hekkert, P., Morris, M., et al.: 1999, ApJ 514, L115. Sahai, R., Zijlstra, A., S´anczez Contreras, C., and Morris, M.: 2003, ApJ 586, L81. Salama, A., Evans, A., Eyres, S. P. S., Leech, K., Barr, P., and Kessler, M. F.: 1996, A&A 315, L209. Salama, A., Barr, P., Clavel, J., et al.: 1997, ApSS 255, 227. Salama, A., Eyres, S. P. S., Evans, A., Geballe, T. R., and Rawlings, J. M. C.: 1999, MNRAS 304, L20. Schild, H., Dumm, T., Folini, D., et al.: 1998, in: Kesler, M. and Cox, P. (eds.), The Universe as Seen by ISO, ESA SP-427, p. 397. Schild, H., Eyres, A. P. S., Salama, A., and Evans, A.: 2001, A&A 378, 146. Sloan, G. C. and Price, S. D.: 1995, ApJ 451, 758. Sloan, G. C., Kraemer, K. E., Goebel, J. H., and Price, S. D.: 2003, ApJ 594, 483. Smith, J. D. T. and Houck, J. R.: 2001, ApJ 121, 2115. Speck, A., Barlow, M. J., Sylvester, R. J., and Hofmeister, A. M.: 2000a, A&AS 146, 437. Speck, A., Meixner, M., and Knapp, G. R.: 2000b, ApJ 545, L145. Speck, A. and Hofmeister, A. M.: 2004, ApJ 600, 986. Surendiranath, R., Pottasch, S. R., and Garc´ıa-Lario, P.: 2004, A&A 421, 1051. Sylvester, R. J., Kemper, F., Barlow, M. J., de Jong, T., Waters, L. B. F. M., Tielens, A. G. G. M., et al.: 1999, A&A 352, 587. Szczerba, R. and G´orny, S. K. (eds.): 2001, Post-AGB Objects as a Phase of Stellar Evolution, Kluwer Academic Publishers, Dordrecht/Boston/London. Szczerba, R., Stasi´nska, G., Si´odmiak, N., and G´orny, S. K.: 2003, in: Gry, C., Peschke, S., Matagne, J., Garcia-Lario, P., Lorente, R., and Salama, A. (eds.), Exploiting the ISO Data Archive. Infrared Astronomy in the Internet Age, ESA-SP, Vol. 511, p. 149. Szczerba, R., Henning, Th., Volk, K., et al.: 1999, A&A 345, L39. Tanab´e, T., Nishida, S., Nakada, Y., et al.: 1998, Astrophys. Space Sci. 255, 515. Tielens, A. G. G. M., Hony, S., Van Kerckhoven, C., and Peeters, E.: 1999, in: Cox, P. and Kessler, M. F. (eds.), The Universe as Seen by ISO, ESA-SP, Vol. 427, p. 579. Trams, N. R., van Loon, J. Th., Zijlstra, A. A., et al.: 1999a, A&A 344, L17. Trams, N. R., van Loon, J. Th., Waters, L. B. F. M., et al.: 1999b, A&A 346, 843. Truong-Bach, Sylvester, R. J., Barlow, M. J., Nguyen-Q-Rieu, Lim, T., Liu, X. W., et al.: 1999, A&A 345, 925. Tsamis, Y. G., Barlow, M. J., Liu, X.-W., Storey, P. J., and Danziger, I. J.: 2004, MNRAS 353, 953. Tsuji, T.: 2000, ApJ Lett. 540, L99.

LATE STAGES OF STELLAR EVOLUTION

243

Tsuji, T.: 2001, A&A 376, L1. Tsuji, T., Ohnaka, K., Aoki, W., and Yamamura, I.: 1997, A&A 320, L1. Ueta, T., Meixner, M., and Bobrowsky, M.: 2000, ApJ 528, 861. Vandenbussche, B., Beintema, D., de Graauw, T., et al.: 2002, A&A 390, 1033. van der Hucht, K. A.: 2001, New Astron. Rev. 45, 135. van der Hucht, K. A., Morris, P. W., Williams, P. M., et al.: 1996, A&A 315, L193. van Hoof, P. A. M., Beintema, D. A., Verner, D. A., and Ferland, G. J.: 2000, A&A 354, L41. van Loon, J. Th., Groenewegen, M. A. T., de Koter, A., et al.: 1999, A&A 351, 559. van Loon, J. Th., Gilmore, G. F., Omont, A., Blommaert, J. A. D. L., Glass, I. S., et al.: 2003, MNRAS 338, 857. Van Malderen, R.: 2003, PhD Thesis, K.U. Leuven, Leuven, Belgium. Van Malderen, R., Decin, L., Kester, D., Vandenbussche, B., Waelkens, C., Cami, J., and Shipman, R. F.: 2004, A&A 414, 677. van Winckel, H.: 2003, ARAA 41, 391. van Winckel, H., Waelkens, C., and Waters, L. B. F. M.: 1995, A&A 293, L25. van Winckel, H., Waelkens, C., and Waters, L.B.F.M., et al.: 1998, A&A 336, L17. Vardya, M. S., de Jong, T., and Willems, F. J.: 1986, ApJ Lett. 304, L29. Volk, K., Kwok, S., and Hrivnak, B.: 1999, ApJ 516, L99. Volk, K., Kwok, S., Hrivnak, B., and Szczerba, R.: 2002, ApJ 567, 412. Volk, K., Xiong, G.-Z., and Kwok, S.: 2000, ApJ 530, 408. Waelkens, C. and Waters, L. B. F. M., 2004, in: Habing, H. J. and Olofsson, H. (eds.), Asymptotic Giant Branch Stars, Springer-Verlag, New York. Waters, L. B. F. M., Cami, J., de Jong, T., Molster, F. J., van Loon, J. T., Bouwman, J., et al.: 1998a, Nature 391, 868. Waters, L. B. F. M., Beintema, D. A., Zijlstra, A. A., et al.: 1998b, A&A 331, L61. Williams, P. M.: 1995, in: van der Hucht, P. A. and Williams, P. M. (eds.), Wolf-Rayet Stars: Binaries; Colliding Winds, IAU Symposium 163, Kluwer Academic Publishers, Dordrecht, p. 335. Williams, P. M., van der Hucht, K. A., and Morris, P. W.: 1998, in: Waters, L. B. F. M., Waelkens, C., van der Hucht, K. A., and Zaal, P. A. (eds.), ISO’s View on Stellar Evolution, Astrophys. Space Sci. 255, p. 169. Willis, P. M., Dessart, Luc, Crowther, P. A., et al.: 1997, MNRAS 290, 371. Willis, P. M., Dessart, Luc, Crowther, P. A., et al.: 1998, in: Waters, L. B. F. M., Waelkens, C., van der Hucht, K. A., and Zaal, P. A. (eds.), ISO’s View on Stellar Evolution, Astrophys. Space Sci. 255, p. 167. Yamamura, I., de Jong, T., and Cami, J.: 1999a, A&A 348, L55. Yamamura, I., de Jong, T., Onaka, T., Cami, J., and Waters, L. B. F. M.: 1999b, A&A 341, L9. Yamamura, I., de Jong, T., Waters, L. B. F. M., Cami, J., and Justtanont, K.: 1999c, in: IAU Symposium 191: Asymptotic Giant Branch Stars, p. 267. Yamamura, I., Dominik, C., de Jong, T., Waters, L. B. F. M., and Molster, F. J.: 2000, A&A 363, 629. Yan, M., Federman, S. R., Dalgarno, A., and Bjorkman, J. E.: 1999, ApJ 515, 640. Yang, X., Chen, P., and He, J.: 2004, A&A 414, 1049. Zubko, V. and Elitzur, M.: 2000, ApJ Lett. 544, L137.

INTERSTELLAR MEDIUM

THE COOL INTERSTELLAR MEDIUM ALAIN ABERGEL1,∗ , LAURENT VERSTRAETE1 , CHRISTINE JOBLIN2 , 1,4 ˆ RENE´ LAUREIJS3 and MARC-ANTOINE MIVILLE-DESCHENES 1 IAS,

Universit´e Paris-Sud, Bˆat. 121, 91405 Orsay, France CNRS-Universit´e Paul Sabatier, 9 avenue du Colonel Roche, 31028 Toulouse, France 3 Astrophysics Division, Research and Scientific Support Department of ESA, ESTEC, PO Box 299, 2200 AG Noordwijk, The Netherlands 4 CITA, 50 St-Georges street, Toronto, Ontario, M5S3H8, Canada (∗ Author for correspondence: E-mail: [email protected]) 2 CESR,

(Received 21 September 2004; Accepted in final form 8 November 2004)

Abstract. Infrared spectroscopy and photometry with ISO covering most of the emission range of the interstellar medium has led to important progress in the understanding of the physics and chemistry of the gas, the nature and evolution of the dust grains and also the coupling between the gas and the grains. We review here the ISO results on the cool and low-excitation regions of the interstellar medium, where Tgas  500 K, n H ∼ 100–105 cm−3 and the electron density is a few 10−4 . Keywords: infrared: spectroscopy, photometry; ISM: gas, dust; ISM: PDR, cirrus, cold clouds; ISM: physical processes JEL codes: D24, L60, 047

1. The Cool Interstellar Medium The interstellar medium (ISM) fills the volume of the Galaxy and its evolution is mostly driven by the stellar activity. It is composed primarily of gas and of a small amount (1% in mass) of submicronic dust grains. In spite of their low abundance, dust particles play a key role for the temperature and chemistry of the gas. Thus, the evolution of the interstellar gas through its different phases (from proto-stellar cores to the diffuse medium and vice-versa) and of dust grains are closely related. Although shocks and winds play an important role in the vicinity of stars (Blommaert et al.; Nisini et al.; Peeters et al., this volume), the ISM is primarily heated by stellar radiation. It then cools off by continuum and line emission in the infrared (IR) and submillimetre (submm) range. Observations of the IR emission from the ISM hence provide important clues on the gas state, the nature of the dust particles, the interactions between the gas and the dust as well as the dynamical processes affecting the gas. Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands, and the United Kingdom), and with the participation of ISAS and NASA. Space Science Reviews (2005) 119: 247–271 DOI: 10.1007/s11214-005-8056-z

 C

Springer 2005

248

A. ABERGEL ET AL.

In the course of its lifecycle, the ISM assumes a number of phases described by the temperature Tgas and proton density n H of the gas as well as the electron abundance xe . Most of the ISM mass resides in cool, mostly neutral phases – the cool or low-excitation ISM – where Tgas  500 K, n H ∼ 100–105 cm−3 and xe is a few 10−4 . The cool ISM encompasses photon-dominated regions or PDRs generated by the far-ultraviolet (FUV) stellar radiation field, shielded regions that are cold and molecular (the cold clouds), low-mass star forming clouds, the outskirts of massive star forming regions and the general ISM (the diffuse ISM seen in cirrus clouds). This medium is therefore widespread in the Galaxy (some 10% in volume), and observed to be structured on all scales (a few 0.01 to a few 10 pc, e.g., Falgarone et al., 1998; Miville-Deschˆenes et al., 2003). The cool ISM is the subject of this review while the more highly excited phases of the ISM, the regions around YSOs and late-type stars are discussed in this volume by Peeters et al.; Nisini et al.; Lorenzetti; Blommaert et al., respectively. The ISO mission was particularly well suited for ISM studies because of the complementarity of its four instruments. The CAM and PHOT instruments provided sensitive broad-band IR images at angular resolution of a few arcseconds for ISOCAM to 1 arcmin for ISOPHOT. The former also had a spectro-imaging facility, the CAM-CVF mode operating between 6 and 16 μm with a resolving power R ∼ 40 while the latter was equipped with a grating spectrometer used in the PHOT-S mode (two channels: 2.5–4.9 μm and 5.8–11.6 μm with R ∼ 90 and a 24 × 24 arcsec aperture). On the other hand, the SWS and LWS spectrometers combined relatively higher spectral resolution (R ∼ 200–2000) and full spectral coverage (2.4–45.2 μm for the SWS and 43–197 μm for the LWS). The ISO instruments have detected a great wealth of gas lines and dust spectral features in emission or absorption. High spectral resolution measurements with SWS and LWS have been performed mostly towards moderately to highly excited regions (H II regions, planetary nebulae, PDRs and active galaxies), while less excited regions could only be observed by CAM and PHOT at low spectral resolution or in broad bands. Thus, ISO has provided a complete and unbiased IR spectroscopic census of interstellar matter highlighting the nature and properties of dust, the physical conditions of the gas as well as the major processes (UV stellar irradiation, shock waves) driving the evolution of these species. While the results on the H2 and H2 O molecules (Habart et al.; Cernicharo and Crovisier) and the dust features of silicates and ices (Molster and Kemper; Dartois) are discussed in dedicated chapters of this volume, we focus here on the properties of gas and dust in the cool ISM. We also refer the reader to the recent review by van Dishoeck (2004). 2. The Physics and Chemistry of the Gas Combined observations of gas lines and dust emission proved to be a unique tool to constrain the physics and chemistry of the gas in the cool ISM (Section 2.1) and to quantify the importance of the interactions between gas and dust. These latter

249

THE COOL ISM

are most directly studied in low-excitation PDRs (Section 2.2) while gas dynamical processes are best traced in the less excited, diffuse medium (Section 2.3). 2.1. PROBING INTERSTELLAR C HEMISTRY OF G AS -P HASE M OLECULES

WITH

S PECTROSCOPY

The wavelength range of the ISO spectrometers includes rotational and rovibrational transitions of many light interstellar molecules. The ISO spectroscopy data thus provided an important benchmark for models of chemistry. Furthermore, analysis of ISO spectra with models pointed at the dominant excitation processes and highlighted the structure of the observed regions. A wealth of molecular lines has been found towards the galactic centre. As detailed by Ceccarelli et al. (2002) the Sgr B2 region is a massive molecular cloud containing numerous compact and ultra-compact H II regions and hot cores embedded in a relatively dense and warm envelope (∼100 K), while the entire complex is surrounded by a hot (≥300 K) envelope of diffuse gas (n H ∼ 103 cm−3 ). Sgr B2 is therefore an ideal target to detect molecular absorption. The detection of new absorption features has strongly constrained the abundances and the physical conditions (gas temperature and density) of the absorbing regions. Numerous rotational lines of hydrides (OH, CH, HF, H2 O, H3 O+ , NH3 , NH2 , NH) and hydrocarbons (C3 , C4 or C4 H) have been detected in absorption by LWS (see Tables 4 and 5 of Goicoechea et al., 2004 for the complete list of lines and references). Light hydrides such as OH, CH and H2 O were found at all positions, indicating the widespread presence of these species in molecular clouds. Lines of the H2 O/OH/O0 and NH3 /NH2 /NH sequences were found to trace the importance of photodissociation processes and shock respectively (Goicoechea et al., 2004). The HD molecule has been detected by the LWS (Wright et al., 1999; Caux et al., 2002; Polehampton et al., 2002). A deuterium abundance of D/H ∼0.2–11×10−6 has been derived for molecular clouds in agreement with other determinations in the Solar neighbourhood. The hydrogen fluoride HF has been discovered through its J = 2 − 1 band at 121.7 μm (Neufeld et al., 1997). HF is the dominant reservoir of gas-phase fluorine and, assuming that the absorption arises in the warm envelope, the derived abundance (∼3 × 10−10 ) indicates a strong depletion of fluorine onto dust grains. However, Ceccarelli et al. (2002) mentioned that this absorption could also be due to the hot absorbing layer, in which case the abundance would be an order of magnitude larger. This illustrates that the interpretation of such data is not straightforward, since several components with different physical properties can participate to the detected absorption, as examplified in the work of Vastel et al. (2002) for the C+ and O0 fine-structure lines. Original results have also been obtained on hydrocarbons and carbon chains. Cernicharo et al. (2002) have found a new line at 57.5 μm which may be the first detection of C4 (another possible carrier is C4 H). The required abundance of C4 would make it the most abundant carbon chain in interstellar and circumstellar

250

A. ABERGEL ET AL.

media. Other polar carbon chains such as cyanopolynes (HC2n+1 N with n = 1 − 5) were already detected in the millimeter, but ISO has allowed the detection of pure carbon chains through their low-energy bending modes which fall in spectral regions totally absorbed by the Earth’s atmosphere. On the other hand, ro-vibrational emission lines of HCN and C2 H2 have been found in warm (a few 100 K) molecular cores with abundances ∼10−7 and a likely scenario is that these lines are pumped by warm dust IR emission (Boonman et al., 2003). The methyl radical CH3 has been discovered in absorption toward Sgr A with SWS through its ν2 Q-branch at 16.5 μm and the R(0) line at 16.0 μm (Feuchtgruber et al., 2000). The relatively high abundance of CH3 (∼10−8 ) together with the low abundances of other hydrocarbon molecules (CH, CH4 and C2 H2 ) is inconsistent with the results of published pure gas-phase models of dense clouds. The authors suggest that models including diffuse and translucent clouds or translucent clouds with gas–grain chemistry may resolve this discrepancy. 2.2. THE C OUPLING

BETWEEN

GAS

AND

DUST

IN

L OW-E XCITATION PDRS

Photo Dissociation Regions (PDRs) are ubiquitous in the Galaxy since any interstellar cloud exposed to the ambient radiation field will develop such an interface. More generally, PDRs dominate the IR emission of galaxies and are used to trace their star formation activity (e.g., F¨orster-Schreiber et al., 2004). Early observations of bright, highly excited PDRs (e.g., Harper et al., 1976; Melnick et al., 1979) have led to detailed modelling where the thermal and chemical balance of the gas is fully described. Physical conditions (radiation field intensity and gas density, see below) in PDRs were then obtained from the comparison between model predictions and observations (see the review in Hollenbach and Tielens, 1999). These studies have pointed out the importance of gas–dust interactions and the need to study them further. Among these interactions, the most important are the photoelectric effect on dust which heats the gas and the formation of the H2 molecule on the surface of grains which triggers the chemistry.1 The gas state thus results of a balance between photon-driven processes (photoelectric effect on dust, photodissociation of H2 , . . .) and density dependent reactions (electron recombination, H2 formation on grains, . . .). Hence, the excitation of PDRs is usually described with χ, a scaling factor for the intensity of the radiation field in the FUV,2 and with the local gas proton density n H . Observations – including the ISO mission – have been primarily performed on nearby PDRs which can be spatially resolved. Among these regions those with an edge-on geometry represent ideal targets for the study of gas–grain interactions, dust evolution and chemical stratification as a function of depth (or UV field intensity) within the cloud. 1 Gas–grain

and gain-grain collisions are also important. These processes, however, primarily affect the dust (size distribution, structure and composition) and they are discussed in Section 3.4. 2 For χ = 1 the stellar intensity is 4πν I = 1.2 × 10−6 W/m2 at λ ∼ 100 nm, i.e. the average ν interstellar value in the solar neighbourhood (Mathis et al., 1983).

THE COOL ISM

251

PDRs have been extensively observed with the four instruments of ISO, spanning a wide range of physical conditions. Thanks to the sensitivity of the ISO instruments, the sample of observed PDRs has been extended towards low-excitation regimes (χ  a few 1000 and n H  104 cm−3 ) allowing a detailed study of important microscopical processes. In low-excitation PDRs, the gas thermal budget is amenable to a few dominant processes (Kemper et al., 1999; Habart et al., 2001): the heating is due to the photoelectric effect on dust and the cooling occurs through the fine-structure line emission of C+ (158 μm) and O0 (63 and 145 μm). These lines measured by ISO-LWS have been used to derive the efficiency of the photoelectric heating, assuming thermal balance (heating = cooling). This efficiency noted  represents the fraction of the stellar energy absorbed by a given dust population that goes into gas heating. In practice,  can be determined observationally by correlating the gas cooling line to the dust emission. In such a detailed study towards L1721, an isolated cloud heated by a single B-star (χ ∼ 4 and n H ∼ 3000 cm−3 ), Habart et al. (2001) have shown that the photoelectric heating is dominated by the smallest grains (of radii 1 nm called PAHs, see Section 3) with PAH ∼ 3%. These results have confirmed the theoretical treatment of the gas photoelectric heating (Bakes and Tielens, 1994; Weingartner and Draine, 2001a) used in current PDR models. Towards high-latitude, diffuse clouds (χ ∼ 1 and n H ∼ 100 cm−3 ), Ingalls et al. (2002) have found a statistical correlation between the C+ line and the far-IR emission of big grains (radii 10 nm) as measured with IRAS at 60 and 100 μm which corresponds to an efficiency of ∼4.3%. This latter value, about 1.5 times larger than the theoretical predictions of Weingartner and Draine (2001a), suggests that small grains (PAHs and VSGs, see Section 3.1) undergoing temperature fluctuations contribute significantly to the IR emission. Other studies made use of PDR models treating in detail the gas thermal balance to analyse observed cooling lines. The [C+ ] 158 μm and [O0 ] 63 μm dominant lines were successfully explained with the current treatment of photoelectric heating (Timmermann et al., 1998; Kemper et al., 1999; Liseau et al., 1999; Habart et al., 2001, 2003; Li et al., 2002; Okada et al., 2003; Schneider et al., 2003) confirming independently the photoelectric heating rate used in models. These studies have also shown that the major cooling lines are optically thick, even in low-density PDRs (n H  103 cm−3 ). In particular, the source geometry and line transfer proved to be important for the O0 transitions (Li et al., 2002; Okada et al., 2003). In most cases, the O0 emission line ratio 63 μm/145 μm cannot be readily explained (Liseau et al., 1999) and recent work (Mizutani et al., 2004) has suggested absorption in the [C+ ] 158 μm and [O0 ] 63 μm lines and/or a quenched photoelectric heating as the possible cause of this discrepancy. Finally, comparison of spatial profiles of the C+ and O0 lines to model results has shown that the abundance of PAHs is strongly depressed (by up to a factor 5) in shielded regions (Habart et al., 2001, 2003). A comparable result has been found in the diffuse medium, traced by the emission of small grains as seen by CAM (Section 3.4).

252

A. ABERGEL ET AL.

More indirect analysis of PDRs has also been conducted using the [O0 ] 63 and 145 μm and [C+ ] 158 μm gas cooling lines detected in absorption and/or emission with LWS towards the Galatic centre, in Sgr B2 (Vastel et al., 2002; Goicoechea et al., 2004). These observations have allowed to disentangle the contribution of the three layers (C+ /C+ -C0 /CO) predicted by PDR models. The intensities of the three lines indicate that χ ∼ 103 –104 and n H ∼ 103 –104 cm−3 in the PDRs (Goicoechea et al., 2004). The [C+ ] 158 μm line is seen in absorption, confirming that this main cooling line of the ISM can be optically thick. Moreover, Vastel et al. (2002) show that atomic oxygen is associated with the three layers. Less than 30% of the total O0 column density (∼ 3 × 1019 cm−2 ) comes from the external (C+ ) layers. On the other hand, ∼70% of gaseous oxygen is in the atomic form and not locked up in CO in the molecular clouds lying along the line of sight. This latter result is not explained by current PDR models. This may be related to the low H2 O abundance detected by the SWAS satellite and which is not explained by current chemical networks (Roberts and Herbst, 2002). More observations of O0 , O2 and H2 O in molecular clouds are necessary to solve this issue. For the first time, the pure rotational bands of H2 have been detected with SWS towards many PDRs providing a direct probe of the gas temperature. In lowexcitation PDRs, current PDR models failed to account for this emission suggesting an enhanced H2 formation rate. Observations also suggest that small grains (radii 10 nm) play an important role in the formation mechanism. This issue is discussed in the review on molecular hydrogen by Habart et al. (this volume). 2.3. NON -THERMAL GAS MOTIONS Non-thermal motions can play an important role in exciting and mixing the interstellar gas. Some of the underlying processes are directly related to stars (outflows, winds) and are discussed elsewhere in this volume (Nisini et al.; Peeters et al.). We discuss here processes that pervade most of the ISM, namely, motions resulting from magneto-hydrodynamical (MHD) turbulence and motions driven by the ambient radiation field. MHD turbulence has been studied by Falgarone et al. (1999, 2005) from SWS observations of the pure rotational lines of H2 in emission along a deep sightline in the galactic plane sampling the cold diffuse ISM (i.e., avoiding star forming region and molecular clouds). The detected H2 excitation has shown the presence of a small fraction (a few percents) of warm gas (a few 100 K). This confirms more directly earlier absorption measurements of molecules (CH+ , OH, HCO+ ) that can only form in warm gas through endothermic reactions. As illustrated in Figure 1, collisional excitation by MHD turbulence (in shocks or vortices) can successfully account for the excited H2 . In addition, the required number of turbulent structures can also explain the observed CH+ abundances. More recent studies with FUSE towards stars in the Solar neighbourhood find similar fractions of warm gas (Gry et al., 2002). This suggests that turbulent motions occur with similar rates

THE COOL ISM

253

Figure 1. SWS Intensities (solid squares) of the pure H2 rotational lines in the cold diffuse medium. Ju is the upper rotational number of the transition. Left: The contribution of background PDRs on the line of sight (upper solid line) consists of: (1) 19 diffuse PDRs in the Solar neighbourhood (χ = 1, n H = 30 cm−3 and AV = 0.3, lower solid line), (2) 5 diffuse PDRs in the molecular ring (χ = 10, n H = 100 cm−3 and AV = 2, dashed line) and (3) 7 dense PDRs in the molecular ring (χ = 10, n H = 104 cm−3 and AV = 2, dotted line). The fluorescence contribution of low-density PDRs is too low to account for the fluxes of the excited H2 rotational lines. Center: Residual H2 line intensities once the PDR-type emission (see left panel) has been removed (solid squares). Models of H2 lines emission from a few tens of MHD shocks (dotted lines) at 8 (lower), 10 and 12 km s−1 (upper). Right: Same residuals compared to the emission of about 104 magnetized coherent vortices of rotational velocity 4 km s−1 (dot-dashed) and 3.5 km s−1 (dashed). The dotted line releases the assumption of statistical balance of the H2 level populations. Taken from Falgarone et al. (2005).

throughout the inner parts of the Galaxy. In the near future, the spectrometer of the Spitzer Space Telescope (SST) will look at the H2 excitation towards a wider variety of galactic sightlines as well as in other galaxies. Turbulent motions have also been traced indirectly in cirrus clouds through their impact on the dust size distribution (Section 3.4). In the warmer environment of PDRs, non-thermal motions are dwarfed by thermal processes and their signature is more allusive. For instance, a possible explanation of the low ortho-to-para ratio of H2 observed in the warm layers (a few 100 K) of several PDRs are advection motions due to the propagation of photodissociation fronts (Habart et al., this volume). However, a quantitative estimate of the efficiency of this process awaits a dedicated modelling effort. 3. The Nature and Evolution of Interstellar Dust Grains Observations of IR features in emission or absorption provide important constraints on the nature, composition and properties (size distribution, structure) of the dust

254

A. ABERGEL ET AL.

particles. Work prior to the ISO mission (e.g., Sellgren, 1984; D´esert et al., 1990; Dwek et al., 1997) has shown that interstellar dust comprises a population of small grains (radii a  10 nm) that absorbs about half the energy radiated away by stars and re-emit this energy during temperature fluctuations. As discussed in Section 2.2, small grains play an important role in the physics and chemistry of the interstellar gas. These small grains are believed to be predominantly carbonaceous (e.g., Draine, 2003) and we will discuss their size in terms of the number of carbon atoms they contain, NC . The smallest of them (a  1 nm and NC  a few 100) are the carriers of a family of IR emission bands between 3 and 13 μm characteristic of aromatic rings (Duley and Williams, 1982) and we call these features the aromatic IR bands (AIBs). To explain these bands, a family of large aromatic molecules, the Polycyclic Aromatic Hydrocarbons (PAHs) has been proposed (L´eger and Puget, 1984; Allamandola et al., 1985). As discussed below (Sections 3.1 and 3.2), the PAH model explains the main properties of the AIB observations but still lacks a precise spectroscopic identification with terrestrial analogues. Dust emission at mid-IR wavelengths (e.g., in the IRAS 25 and 60 μm bands) has been ascribed to somewhat larger carbonaceous grains (1  a  10 nm) called Very Small Grains (VSGs) (D´esert et al., 1990). Finally, the far-IR dust emission (λ  30 μm) is ascribed to a mixture of large graphite and silicated grains or big grains (a ∼ a few 10 to a few 100 nm, e.g., Li and Draine, 2001; Weingartner and Draine, 2001b) which emit while being in thermal equilibrium with the radiation field. The ISO instruments have measured the IR emission of dust particles of all sizes in a wide variety of astrophysical environments. The analysis, still ongoing, of this large database has already lead to important steps forward in our understanding of interstellar dust as examplified here and in other reviews of this volume. ISO observations of the mid-IR dust emission spectrum have highlighted the properties of PAHs and VSGs as well as their evolution in the interstellar lifecycle. For the first time, direct evidence of important variations of the abundance of small grains in diffuse and cold clouds has been found in deep photometric ISO data. These latter results point at the importance of dust processing in the interstellar medium (see Section 3.4). 3.1. THE EMISSION S PECTRUM

OF

S MALL GRAINS

AND ITS I MPLICATIONS

One of the prominent results of ISO was to show the ubiquitous presence of the AIBs (at 3.3, 6.2, 7.7, 8.6, 11.3 and 12.7 μm, see Figure 2) in the interstellar medium of the Galaxy (e.g., H II regions: Peeters et al., 2002; planetary nebulae: van Diedenhoven et al., 2004; Uchida et al., 2000; PDRs: Cesarsky et al., 2000b; Verstraete et al., 2001; cirrus and diffuse ISM: Boulanger et al. 1996; Mattila et al., 1996; Chan et al., 2001; Kahanp¨aa¨ et al., 2003), but also in external galaxies (see the reviews by Sauvage et al., and Verma et al., in this volume). Furthermore, the ratio of the 7.7 μm band to the local continuum has now become a tool to measure the starburst activity in galaxies (Lutz et al., 1998).

THE COOL ISM

255

Figure 2. Typical CAM (spectral resolution R ∼ 40, top and middle panels) and SWS (R ∼ 200, bottom panel, from Verstraete et al., 2001) spectra in NGC 2023 (from Abergel et al., 2002). For the CAM spectra, the aromatic lines are fitted with Lorentzian profiles, while the S(1), S(2) and S(3) pure rotational lines of H2 are adjusted with gaussian profiles. The continuum arises further inside the molecular cloud than the AIBs (top panel).

In the cool ISM, away from hot stars (of spectral type earlier than B2), the 6–13 μm AIB spectrum has been found to be strikingly similar considering the large range of excitation conditions spanned by the observations (χ ∼ 1–104 , Boulanger et al., 2000; Uchida et al., 2000; Chan et al., 2001; Verstraete et al., 2001). Conversely, the AIB spectrum towards compact H II regions and circumstellar environments changes significantly as shown in Hony et al. (2001), Peeters et al. (2002) and van Diedenhoven et al. (2004). The authors argued that these variations are due to composition changes in a population of PAHs including substituted and metal complexed species. In fact, newly formed PAHs are expected to be found in the ejecta of evolved stars while in the general interstellar medium (which includes the cool ISM), the population of PAHs has already been processed by the ambient UV radiation field and shocks over long timescales (∼108 yrs). In the following we call the AIBs observed in the interstellar medium, the IS-AIBs (interstellar AIBs). We discuss below the IS-AIBs while the AIBs around stars and H II regions are discussed in Peeters et al. (this volume). For the first time, most AIBs were spectrally resolved by the ISO spectrometers. Individual AIBs have been found to be broad (λ/FWHM ∼30–80), smooth and in some cases clearly asymmetrical (the 3.3, 6.2 and 11.3 μm bands all show a prominent red wing). As shown in Boulanger et al. (1998b) and Verstraete et al. (2001) the AIBs are well represented with Lorentzian profiles possibly indicating

256

A. ABERGEL ET AL.

that the AIB carriers are large molecules. Other methods of decomposition have been applied to quantify the AIB intensities (Peeters et al., 2002; Uchida et al., 2000) illustrating the importance of the AIB definition for the extraction of the underlying continuum. SWS data has clearly shown that the 7.7 μm-feature consists of several components (Joblin et al., 2000; Peeters et al., 2002), including the two main features at ∼7.6 and 7.8 μm first reported by Bregman (1989). New features have been added to the PAH family, like the 16.4 μm band (van Kerckhoven et al., 2000; Moutou et al., 2000). This band has been assigned by the latter authors to PAHs containing pentagonal rings, species which are important in chemical networks of PAH and soot formation (Moutou et al., 2000). The 16.4 μm band is also present in CAM-CVF data but falls at the detection edge of the detectors (Figure 2). Some AIBs have also been detected in absorption (see Section 3.3). More recently an emission band at 4.65 μm has been assigned to deuterated PAHs (Peeters et al., 2004). More features are expected to be found in the ISO archive but also thanks to the Infrared Spectrograph (IRS) now operating on the SST. A recent study on the PDR NGC 7023 reports new features at 6.7, 10.1, 15.8, 17.4 and 19.0 μm (Werner et al., 2004). Features in the far-IR and submm range could also be detected in the future by the Herschel Space Observatory (Joblin et al., 2002). In the IS-AIB spectrum, the positions and widths of the major bands (3.3, 6.2, 8.6 and 11.3 μm) are stable (within a few cm−1 , Verstraete et al., 2001), conversely to spectra towards H II regions and circumstellar environments where significant shifts of the 6–9 μm bands are observed (Peeters et al., 2002). To constrain the nature of PAHs, detailed comparisons between the observed AIBs and the laboratory and theoretical data on small (NC ≤ 50), planar, aromatic molecules have been carried out. Pech et al. (2002) have shown that the profiles of the 6.2 and 11.3 μm bands observed in the planetary nebula IRAS 21282+5050 can be explained with the emission of a mixture a medium-sized PAHs (30 < NC < 80) having similar spectroscopic properties. The band positions and widths arise then naturally from the molecular anharmonicity during temperature fluctuations (see Section 3.2). Thus, the strong variability of the vibrational spectrum of terrestrial PAHs as a function of charge state, hydrogen coverage and chemical structure (e.g., Hudgins et al., 2000; Pauzat et al. 1997; Langhoff 1996) is not reflected in the IS-AIBs of the cool ISM where PAHs are expected to be for instance neutral (χ ∼ 1 to a few 100) or ionized (χ ∼ a few 103 , Dartois and d’Hendecourt 1997). Furthermore, a precise, quantitative assignation of major AIBs is still not achieved: the components of the 7.7 μm and the 8.6 μm bands are not explained by the properties of small terrestrial PAHs (Verstraete et al. 2001; Peeters et al., 2002). Spectroscopic evidence (van Kerckhoven et al., 2000; Hony et al., 2001) and modelling (Schutte et al., 1993; Li and Draine, 2002) has pointed out the importance of large PAHs (NC ≥ 50) for the 6–13 μm AIBs. The vibrational bands of such large systems may be less sensitive to the structure, shape, charge and surface state and provide a better understanding of the observed AIBs. However, recent calculations by Bauschlicher (2002) on the

THE COOL ISM

257

large PAH C96 H24 seem to challenge this picture while laboratory measurements on large systems are still awaited. Another conspicuous aspect of the mid-IR spectrum of the ISM is the presence of broad features (λ/FWHM ∼10) and continua underneath the AIBs. These spectral components are seen in the very excited environments studied by Peeters et al. (2002) such as planetary nebulae but also in the cool ISM (see Figure 2). The intensity of the broad features around ∼7, 10 and 12 μm are best delineated in SWS data where all AIBs are spectrally resolved. Their detailed shape however is poorly constrained (Verstraete et al., 2001). This emission has been discussed in the past and proposed to be associated to the AIB spectrum but not directly correlated with the bands3 (Cohen et al., 1985, 1986, 1989; Roche et al., 1989; Bregman 1989). These broad features and continua may arise from larger species (NC  a few 100) as suggested by several CAM studies of the spatial distribution of this broad-band emission relative to the AIBs in PDRs (Section 3.4). Here also laboratory and/or theoretical spectroscopy of carbon clusters is needed to quantitatively interpret these broad bands and continua. 3.2. EMISSION MECHANISM AND M ODELLING

OF THE

AROMATIC INFRARED BANDS

The high sensitivity and dynamical range of the ISO instruments has provided detection of the AIBs from diffuse to bright regions. In particular, it has been found that the integrated flux of the AIBs globally scales with the intensity of the FUV radiation field as long as χ ≤ 104 (Boulanger et al., 1998a; Uchida et al., 2000). This strongly supports a picture in which small grains are transiently heated to high temperature each time they absorb a UV photon (Puget et al., 1985; Sellgren et al., 1985). It has also been shown that the AIBs can be excited by visible photons (Uchida et al., 1998) and this is well explained by AIB emission models, provided that PAHs are ionized or large (Li and Draine, 2002). Studies of the visible, nearIR absorption cross-sections of PAHs with different sizes and in different charge and hydrogenation states are important since, for instance, in the standard galactic radiation field (Mathis et al., 1983) PAHs absorb about half of their energy in this wavelength range. On the other hand, the high spectral resolution of the SWS has stimulated much work on the AIB profiles and on their formation mechanism. Indeed, this dedicated spectroscopy has provided a unique opportunity to compare to laboratory results. Specifically, detailed studies of the IR emission spectra from small, gas-phase PAHs excited by UV photons or thermally heated have shown that the spectral profile (position and width) of the bands depends on the internal vibrational energy (or temperature) of the molecule (Kim et al., 2001; Cook et al., 1998 and references 3 The

laboratory spectra of individual PAH molecules do not seem to present any continuum (e.g., Joblin et al., 1995).

258

A. ABERGEL ET AL.

therein; Joblin et al., 1995). The observed emission profiles hence result from the sum of the individual lorentzian profiles formed during the cooling of the molecule after UV excitation. The asymmetrical profiles of the 3.3, 6.2 and 11.2 μm bands can thus be explained with an intramolecular broadening due to coupling between vibrational modes in PAHs (Pech et al., 2002). Assuming similar spectroscopic properties for a broad range of PAH size, the stability of the 6–13 μm IS-AIBs is naturally explained by an emission during thermal fluctuations which span a large range of temperatures. In this framework and from detailed profile modelling, one can also infer a photodissociation threshold (in terms of internal energy per carbon atom, a few 0.1 eV/C) beyond which small PAHs are destroyed. The smallest PAHs (NC  40) carrying the 3.3 μm band would then be “selected” by photodissociation leading to similar distribution of internal energy. This would explain why the profile of this AIB is so stable in increasingly excited PDRs (Verstraete et al., 2001; van Diedenhhoven et al., 2004). Finally, the mid-IR spectra observed by the ISO has allowed an empirical definition of the size-dependent IR cross-sections of PAHs and VSGs which have been used in recent quantitative modelling of the dust emission (Li and Draine, 2001). 3.3. MID-I NFRARED E XTINCTION The extinction curve is one of the main tool to study interstellar dust, to correct the photometry of embedded sources and to convert the measured extinctions into column densities. Before ISO, the mid-IR extinction curve was thought to be relatively well understood, and described with a power law Aλ ∼ λ−1.7 from ∼1μm down to a pronounced minimum around 7 μm due to the increasing contribution of the silicate absorption feature at 9.7 μm. It was explained by the absorption of a mixture of graphite and silicate grains (Draine, 1989). The situation is now more controversial, as illustrated in the Figure 4 of the recent review of Draine (2003). Using the measurements of hydrogen recombination lines with the SWS, Lutz et al. (1996) and Lutz (1999) have shown that the IR extinction towards Sgr A does not decrease with increasing wavelengths from 4 to 8 μm, and as a consequence does not present any minimum near 7 μm. On the other hand, Rosenthal et al. (2000) using rovibrational lines of H2 detected by the SWS in Orion have found that the mid-IR extinction decreases with increasing wavelengths and presents a minimum at ∼6.5 μm, as expected. The IR extinction curve depends on the composition of dust grains and to a lesser extent on their size distribution (for λ  2 μm) and may vary from place to place. New observations are therefore necessary to clarify this controversy. The IR extinction curve also presents a number of faint features detected with the SWS towards the galactic centre and other sources with large extinction (Schutte et al., 1998; Chiar et al., 2000; Bregman and Temi 2001). The ubiquitous 3.4 μm band is generally attributed to CH stretching in aliphatic hydrocarbons. The asymmetrical deformation modes of CH at ∼6.9 μm have been detected before ISO

THE COOL ISM

259

with the KAO (Tielens et al., 1996), but the symmetrical deformation mode at 7.2 μm has been found for the first time by Chiar et al. (2000) with SWS data. The strengths of the 3.4, 6.9 and 7.2 μm features appear compatible with the hypothesis that Hydrogenated Aliphatic Carbon (HAC) is the main carrier. Finally, the 3.3 and 6.2 μm bands are detected in absorption towards several lines of sight, and are attributed to PAHs. At this time the absorption of other aromatic bands have not been evidenced in the SWS data. As mentionned by Bregman and Temi (2001), a calibration problem in the SWS data at 11.1 μm may preclude the detection of the 11.2 μm band in absorption. These authors have discovered an absorption band centered at 11.2 μm from ground-based spectra of the molecular cloud surrounding Monoceros R2 IRS 3, and attributed this band to the C H out-of-plane vibrational mode of PAHs. 3.4. DUST E VOLUTION

AND

P ROCESSING

The dense shells around evolved stars are known to form dust grains as demonstrated by observations of the dust IR emission and absorption features in these objects (e.g., Blommaert et al., Molster and Kemper and Peeters et al., in this volume). The timescale of this stardust injection is ∼109 yrs. On the other hand, theoretical studies show that dust grains are efficiently destroyed in supernova shocks (Jones et al., 1994) with a lifetime ∼108 yrs. Efficient mechanisms for grain growth must therefore exist in the other phases of the interstellar medium, most probably in dense, cold clouds where the rates of grain coagulation and accretion of gas phase species are expected to be the largest. High energy gas–grain collisions lead to erosion by vaporization (sputtering), while low energy collisions lead to the reverse process of gas accretion onto dust (Jones et al., 1994). Collisions between large grains (radii 100 nm) at velocities above a few km/s lead to fragmentation into small grains (radii  1 nm) or even grain destruction. By contrast, coagulation occurs through low energy collisions (Jones et al., 1996). UV irradiation can also play a major role by destroying dust grains in high excitation regions (with typically χ > 104 ) and also photo-evaporating condensed species. Evidence for photoprocessing of PAHs and VSGs has been found towards H II regions where the 6–13 μm emission spectrum is dramatically different from the IS-AIBs (Roelfsema et al., 1996; Verstraete et al., 1996). Such processing must leave specific signatures in the dust size distribution and optical properties which in turn affect the gas thermal state and chemistry. As we discuss below, ISO has brought important observational evidences of the evolution of the dust size distribution in the ISM, from diffuse cirrus clouds to dense cores. 3.4.1. PDRs Grain dynamics driven by the radiation field could play a major role in the observed evolution of the mid-infrared spectrum across PDRs. Size segregation in regions anisotropically irradiated by stars may arise from radiation pressure and

260

A. ABERGEL ET AL.

recoil forces (due to the photo-ejection of electrons, the photodesorption of weaklybound surface species and the formation of molecular hydrogen, see Weingartner and Draine, 2001c). This process could strongly modify the size distribution at the illuminated surface of PDRs. Furthermore, the gas in PDRs is suddenly heated and ionized, and experiences a strong pressure gradient. It therefore flows supersonically from the molecular cloud surface to the tenuous intercloud medium, carrying away dust grains. In this highly turbulent flow, grain-grain collision might be strong enough to fragment big grains. This process could explain spectro-imaging observations with CAM across several low-excitation PDRs (NGC 2023, 2068, 2071 and 7023; ρ Oph, see Abergel et al., 2002, 2003b; Rapacioli et al., 2005) showing that the ratio of the AIBs (especially around 7.7μm) to the underlying continuum is higher in the photodissociated interface than in more shielded regions. Variations related to a transition in the charge state of PAHs are also expected. Indeed, PAHs are predicted to be neutral inside dense clouds, and positively ionised in the low-density regions facing the illuminated surface of PDRs (e.g., Dartois and d’Hendecourt 1997). Laboratory data and theoretical calculations show that the single ionization of PAHs strongly enhances (by a factor of ∼10) the emissivity of the C C stretching (6.2 and 7.7 μm) and the C H in plane bending (8.6 μm) modes, with respect to the intensity of the C H out-of-plane bending features at 11.3 and 12.7 μm (Hudgins et al., 2000; Langhoff, 1996; Szczepanski and Vala, 1993). Yet, as already discussed in Section 3.1 a direct decomposition of the mid-IR dust emission spectrum shows only limited variations of the AIB ratios in the cool ISM (factor 2 or less, e.g. Uchida et al., 2000). Rapacioli et al. (2005) went one step further by applying the singular value decomposition method to CAM-CVF data in order to extract typical emission spectra for the different population of small grains in PDRs. Three elementary spectra could be extracted (Figure 3): – The first two spectra are dominated by the AIBs and attributed to neutral PAH and PAH cations. The differences between these two spectra follow the expected changes due to ionisation (increase of the (6.2–7.7–8.6 μm)/(11.3–12.7 μm) band ratio and blueshift of the 11.3 μm band in cations according to Hudgins and Allamandola (1999). The PAH cation spectrum is dominant in the low-density gas surrounding the star whereas the PAH neutral spectrum dominates at the interface with the molecular cloud. – The third spectrum contains a strong continuum and dominates behind the illuminated edge of the PDR (far from the illuminating star), and is attributed to very small carbonaceous grains. The fact that a similar spectrum has been seen on a few pixels of the CAM map of the illuminated edge of Ced 201 (Cesarsky et al., 2000a) lends confidence to the analysis performed by Rapacioli et al. (in press). Several features visible in the spectrum indicate a possible aromatic nature of the carriers. Interestingly, the 7.8 μm component of the broad 7.7 μm feature appears to be carried by this population. The authors also argued that these carbonaceous VSGs could be the progenitors of free PAHs. Indeed, the

THE COOL ISM

261

Figure 3. Elementary spectra extracted from spectro-imaging CAM observations of NGC 7023 (Rapacioli et al., in press). These spectra are attributed to PAH cations (upper panel), neutral PAHs (middle panel) and carbonaceous clusters (lower panel).

drop of the VSG emission at the cloud edge appears to be correlated with the increase of the neutral PAH emission. These VSGs may be clusters of PAHs which photoevaporate at the surface of clouds. This idea is consistent with the strong spatial variations observed in the IRAS 12/100 μm ratio at the surface of molecular clouds (Boulanger et al., 1990; Bernard et al., 1993). SST observations are underway to confirm these results on a larger sample of objects (the CAMCVF data suffer from stray light artefacts which limit their useful dynamics, see Boulanger et al., in press), and search for new features at longer wavelengths which can be attributed to these VSGs. It should be emphasized that the question of the formation of PAHs in the ISM is still open. Some PAHs could be formed in the outflows of evolved carbon-rich stars in a chemical kinetic scheme based on soot formation in hydrocarbon flames (Frenklach and Feigelson, 1989; Cherchneff et al., 1992). Cernicharo et al. (2001) detected C4 H2 , C6 H2 and C6 H6 in the proto-planetary nebula CRL 618, showing that aromatic molecules can indeed be formed in such objects and that energetic processes such as UV irradiation or shocks play an important role in the chemistry of these species. 3.4.2. Cirrus Clouds Important variations of the PAH abundance have been observed in diffuse regions like cirrus clouds. Based on a comparison between CAM-LW2 (6.7 μm broad band) and radio data, Miville-Deschˆenes et al. (2002) have shown variations of

262

A. ABERGEL ET AL.

Figure 4. The L1780 cloud (from Miville-Deschˆenes et al., 2003). Left: IRAS 100 μm (top) and 60 μm (middle), and CAM 6.7 μm (bottom) maps across L1780. Right: Average emission as a function of right ascension. The 6.7 and 60 μm were scaled to be compared to the 100 μm emission. The spectacular decrease of the 6.7 μm / 100 μm ratio cannot be explained by the attenuation of the radiation field. The abundance of PAHs likely decreases, a trend which is spatially correlated with the apparition of CO emission.

the abundance of PAHs by up to a factor 10 in a high latitude cirrus cloud. High PAH abundances were found in regions with strong turbulent motions while low abundances were associated with regions of shallower turbulent motions where CO emission is also detected (see Figure 4). This study highlights for the first time the important role of the gas turbulent motions on the evolution of the grain size distribution. The depletion of small grains observed in the denser parts of cirrus clouds is the first step of the dust coagulation seen in cold clouds. 3.4.3. Cold Clouds Depletion of PAHs and VSGs in the dense regions of molecular clouds has also been shown by ISO data suggesting the importance of dust coagulation processes in the ISM. IRAS images first showed that the emission from small grains at 12, 25 and 60 μm, relative to that from big grains (in thermal equilibrium with the radiation field) at 100 μm, varies by one order of magnitude among and within translucent clouds in the nearby ISM (Boulanger et al., 1990). Standard model calculations predict that the FIR emission of dust in thermal equilibrium with the radiation field follows a modified black body with a spectral index β between 1 and 2 (e.g., Hildebrandt, 1983; Draine and Lee, 1984). The brightness per hydrogen atom can be written: Iλ = λ0 × ( λλ0 )−β × Bλ (T ) = NτλH × Bλ (T ). The dust opacity τλ can be derived if the brightness, the column density and the temperature of the emitting particles are known at a given wavelength. However, the main unknown in our understanding of the big dust grains is the FIR emissivity law, due to the observational difficulty to accurately determine the temperature and β independently, together with the column density. FIR observations of a variety of mostly

THE COOL ISM

263

quiescent clouds (i.e. no high mass star formation) have been used to solve this problem. The longest wavelength that can be measured with PHOT and LWS is slightly more than 200 μm, which is not sufficient to allow an independent determination of β and T. On the other hand, the wavelength coverage of the balloon-borne experiment SPM/PRONAOS (λ = 200–580 microns, Lamarre et al., 1994) is extremely suitable for the determination of β, but PRONAOS is less sensitive to dust temperature (except in the coldest objects, T < 15 K, where the spectral energy distribution peaks around the shortest PRONAOS band), so IRAS or PHOT must also be used. For comparison with dust models, the FIR opacity needs to be gauged against the extinction curve in either the visible or near-infrared. This can be done by determining the optical extinction directly from colour excesses or from star counts. The former method is constrained to a limited amount of sightlines and a maximum extinction. The latter method is also limited to a maximum extinction (of AV < 8), and suffers from higher uncertainties in the denser regions due to a lack of statistical accuracy. A study by Arce and Goodman (1999) showed that for low and moderate density regions (AV < 4), the different extinction estimates are reliable and give similar results. With the advent of large stellar databases (e.g. USNO, 2MASS) it has become much easier to perform star counts over any area on the sky, providing valuable extinction information. Alternatively, one can estimate the visual extinction from the total gas column density and assuming a ratio AV /Ngas . This indirect method depends on the assumed ratio between gas and dust and the conversion between molecular line strength and column density. Recent observations by FUSE (Rachford et al., 2002) confirmed that ratio between the total hydrogen column density and reddening NH+H2 /E(B − V ) is constant up to E(B − V ) ∼ 1 and well described by the constant presented by Bohlin et al. (1978), obtained from the Copernicus satellite data. Very high extinctions can be measured from nearinfrared extinction of background emission, first noticed in the ISOCAM images (Perault et al. 1996; Abergel et al., 1996) and used to study the very dense cores (Nisini et al., this volume). For the diffuse ISM associated with atomic hydrogen, Boulanger et al. (1996) found a value of β close to 2.0 and a mean temperature of 17.5 K by correlating COBE data with Dwingeloo HI data. The derived FIR emissivity τ (λ)/NH = 1.0 × 10−25 (λ/250 μm)−2 cm2 is close to the value predicted by Draine and Lee (1984) based on a standard dust model. This result suggests that spherical grains are more likely to occur in the diffuse ISM than porous or fractal grains. Assuming RV = 3.1, the result by Boulanger et al. implies τ (λ)/AV = 2.9×10−4 (λ/200 μm)−2 . We will use this value as a reference in the discussion below. In case of grain processing towards fractal or porous grains, one would expect an increase of the FIR emissivity and a corresponding increase in the value of τ (λ)/AV (e.g., Ossenkopf and Henning, 1994). The first ISO studies of moderate density regions indicated dust temperatures of 13.5 ± 2 K assuming β = 2 (Laureijs et al., 1996; Lehtinen et al., 1997). The

264

A. ABERGEL ET AL.

observed ratios τ (200)/AV had large uncertainties but were not inconsistent with the predictions for a standard dust model. For increasing densities (in the range 100 to 104 cm−3 ) the temperature is generally found to decrease from ∼17.5 K to as low as 12 K (T´oth et al., 2000; Juvela et al., 2002; del Burgo et al., 2003; Lehtinen et al., 2004). The first strong indication that the FIR properties of the dust in moderate density high latitude clouds might deviate from the diffuse ISM was presented by Bernard et al., (1999) based on SPM/PRONAOS and ISO observations of a cirrus cloud in Polaris. Bernard et al. find cold dust at 13.0 K with β = 12.2 ± 0.3. Since the cloud has a low visual extinction ( AV < 1), only an anomalously low ambient radiation field could cause such a low temperature, which is ruled out. It was therefore concluded that the dust properties of the large grains have changed in the cloud. The inferred ratio τ (200)/AV = 17 ± 3×10−4 is more than 5 times the value for the diffuse ISM. Other PRONAOS observations combined with IRAS data of a dense filament in the Taurus molecular cloud by Stepnik et al., (2003) has evidenced temperatures as low as of 12.1 ± 0.2 K in the densest regions (with β = 2), together with an increase of the FIR emissivity by a factor larger than 3 compared with the diffuse ISM. Assuming β = 2, del Burgo et al. (2003) have analysed a sample of moderate density regions (AV < 6) observed with ISO. The regions have dust temperatures in the range from 13.5–15.6 K and one exceptional region with T = 18.9 K. Combination with extinction data from star counts shows that the ratio τ (200)/AV is systematically higher than for the diffuse ISM. In fact a trend was found where τ (200)/AV increases with decreasing temperature. A similar trend, although less apparent, was found by Cambr´esy et al. (2001) who correlated large scale extinction measurements from starcounts with far-infrared DIRBE data in the Polaris Flare region. They find τ (200)/AV in the coldest regions with T < 15.5 K to be several times higher than in the diffuse ISM. Lagache et al. (1998), using COBE data, and T´oth et al. (2000), using ISO serendipity mode data, found that the temperature of the dust can be significantly below 17.5 K over large areas. In these areas, the abundance of small grains is low as inferred from the absence of 60 μm emission. Earlier studies have revealed a strong anticorrelation between the 60 μm emission and dense gas traced by the millimetre transitions of 13 CO (Laureijs et al., 1991; Abergel et al., 1994). In the dense molecular cloud TMC2, a large scale cold component with a uniform temperature of about 12.5 K is evidenced from PHOT data, while only the densest cores show lower temperatures (del Burgo and Laureijs, 2005). The dust spatially associated with the 60 μm emission is significantly warmer (∼19 K with β = 2). In addition the morphology of this “warm” component is different from the cold component suggesting that the cold and warm components are spatially separate. The evolution of dust properties with the gas density is believed to be due to the process of gas accretion onto grains and grain coagulation. Gas accretion forms grain mantles typically for AV  2 mainly composed of water ice and carbon oxides

THE COOL ISM

265

(e.g., Teixeira and Emerson, 1999). However, Preibisch et al. (1993) has shown that the presence of such mantles on dust grains does not significantly increase the FIR emissivity. As the gas cools down and condenses, gas turbulent motions become shallow leading to grain-grain collisions with sufficiently low energy to enable grain sticking and coagulation, which can lead to the formation of clusters with an enhancement of the FIR emissivity (e.g., Ossenkopf and Henning, 1994; Stepnik, 2001; Stepnik et al., in press) compatible with the PHOT and SPM/PRONAOS observations. Such a scenario can also qualitatively explain the observed general trend in the extinction curve, i.e., an increase in RV with the density (e.g., Cardelli et al., 1989). At first order, the coagulation process appears to be compatible with the dynamical and cloud evolutionary time-scales. The formation of ice mantles is also believed to increase the sticking probability for coagulation. The physics of grain–grain collisions, coagulation and growth has been studied theoretically (Dominik and Tielens 1997) and in the laboratory (Blum, 2004) in the context of proto-planetary disks. However, the efficiency of these processes in the ISM still needs a dedicated modelling taking into account the velocity structure of the ISM (e.g., turbulent and intermittent processes). In the last decade laboratory measurements (Agladze et al., 1996; Mennella et al., 1998) have also indicated that, at low temperatures (T < 30 K), the dust optical properties can change significantly likely because of low-energy structural transformations. A systematic analysis of β from a large compilation of PRONAOS data by Dupac et al. (2003) has shown that β is not a constant but varies with dust temperature: high indices (1.6–2.4) being observed in cold regions (11–20 K) while low indices (0.8–1.6) are observed in warm regions (35–80 K). Although systematic variations in grains size and/or chemical composition as a function temperature cannot be ruled out, the most likely explanation is that the spectral index of dust can change with temperature due to a solid state quantum process. Such studies allow a quantitative understanding of the FIR emission of dust in our Galaxy and in outer galaxies (a cold dust component has also been detected with PHOT in non-active galaxies, see the chapter by Sauvage et al., this volume) and pave the way for the scientific analysis of the Herschel and Planck missions. 4. Structure of the Cool ISM The structure of the ISM strongly affects the chemistry and the star formation activity. Observations of the IR dust emission is unique for tracing interstellar matter over a wide range of physical conditions and, in particular, across the atomic to molecular transition where neither H I nor molecular millimetric lines are robust tracers because of the local variations of the abundance and the excitation conditions. CAM broad-band mapping of molecular clouds has shown the spatial distribution of the emission due to small grains with an angular resolution two orders of magnitude better than IRAS at 12 μm. The images are, to first order, anti-correlated with the visible emission (e.g., Figure 5), and contain small scale brightness fluctuations

266

A. ABERGEL ET AL.

Figure 5. L1630 in the visible (35 × 40 field from the UK Schmidt Telescope and extracted from the DSS produced at the STSI) and with CAM (blue: 5–8.5 μm, red: 12–18 μm) from Abergel et al. (2002). The central region (NGC 2023) is saturated because of the dynamic range.

THE COOL ISM

267

related to illuminated edges of dense regions. For nearby objects with edge-on geometry, IR filaments as sharp as ∼10 in. (or ∼0.02 pc at a distance of 400 pc) are detected due to the combined effects of steep density gradients and extinction in the dense regions (e.g., Abergel et al., 2003a in the illuminated edge of the Horsehead nebula). Associated measurements of the penetration length of the incident radiation field gives estimates of the density just behind the illuminated edges. More quantitative constraints on the density profiles can be derived from the comparison of PDR modelling with spatial profiles of the AIBs as well as of cooling lines or molecular emissions taken with ISO or from the ground (Habart et al., 2005; Pety et al., 2005; Habart et al., 2001). Yet, such analysis should be carried out bearing in mind the discrepancies between the line intensities predicted by different PDR codes (van Dishoeck, 2004). Moreover, caution should be also exercised in analysing the spatial profiles observed across PDRs because of the complex interface geometry seen in the high angular resolution data of the ISO cameras (e.g., Habart et al., 2005). This is also true of ground-based and of recent SST data. The spatial structure of the cool ISM appears self-similar in a wide range of scales. With PHOT images of cirrus emission at 90 μm, the power law relation established from IRAS data (Gautier et al., 1992) between the power of the fluctuations and the spatial scale has been extended to twice as high spatial fequencies, while at 170 μm fluctuations at arcmin scales are studied for the first time (Herbstmeier et al., 1998; Kiss et al., 2003). The spectral index is found to vary from field to field (−5.3 < α < −2.1), depending on the absolute surface brightness. Higher values of α are found at 170 μm rather than at 90 μm and are attributed to temperature variations of dust. Compared to ISO, the SST has the sensitivity, the angular resolution and the mapping capabilities to extend significantly these studies (see the first results by Ingalls et al., 2005).

5. Conclusions The ISO mission has provided an incredibly rich harvest of gas and dust features in the general interstellar medium away from hot stars, the cool ISM. These results have highlighted the physics and chemistry of interstellar matter and its evolution. Emission and absorption lines of gas phase species were used to constrain the excitation processes (shocks, stellar UV irradiation) of the ISM and represented a major benchmark for the modelling. Similarly, spectroscopy of dust features has allowed a detailed comparison to the properties of terrestrial analogues of interstellar grains. These studies have pointed out the importance of nanometric carbon clusters in space and the need to build a laboratory and/or theoretical database of spectra for a large size range of these species. The spatial distribution of the dust emission observed by the photometers has clearly shown and for the first time, the importance of evolution processes (e.g., photodestruction, grain coagulation) affecting the dust

268

A. ABERGEL ET AL.

size distribution. The joint observations of dust and gas features have highlighted the gas–grain couplings (photoelectric effect, H2 formation) and allowed quantitative studies of the underlying microphysics. Furthermore, the quantitative understanding of the gas-to-dust correlation for all dust components is a key issue for the analysis of the extended emission in the future surveys to be conducted with Planck and Herschel, allowing to disentangle brightness fluctuations due to the ISM from those of the cosmic IR and microwave backgrounds. References Abergel, A., Boulanger, F., Mizuno, A., et al.: 1994, Astrophys. J. 423, 59. Abergel, A., Bernard, J.-P., Boulanger, F., et al.: 1996, Astron. Astrophys. 315, 329. Abergel, A., Bernard, J. P., Boulanger, F., et al.: 2002, Astron. Astrophys. 389, 239. Abergel, A., Teyssier, D., Bernard, J. P., et al., 2003a, Astron. Astrophys. 410, 577. Abergel, A., 2003b, in C. Gry, S. Peschke, J. Matagne, P. Garcia-Lario, R. Lorente, and A. Salama (eds.), Exploiting the ISO Data Archive, ESA SP-511, p. 177. Agladze, N. I., Sievers, A. J., Jones, S. A., et al.: 1996, Astrophys. J. 462, 1026. Allamandola, L. J., Tielens, A. G. G. M., and Barker, J. R.: 1985, Astrophys. J. 290, L25. Arce, H. G. and Goodman, A. A.: 1999, Astrophys. J. 517, 264. Bakes, E. L. O. and Tielens, A. G. G. M.: 1994, Astrophys. J. 427, 822. Bauschlicher, C. W.: 2002, Astrophys. J. 564, 782. Bernard, J. P., Boulanger, F., and Puget, J. L.: 1993, Astron. Astrophys. 277, 609. Bernard, J.-P., Abergel, A., Ristorcelli, I., et al.: 1999, Astron. Astrophys. 347, 640. Bohlin, R. C., Savage, B. D., and Drake, J. F.: 1978, Astrophys. J. 224, 132. Boonman, A. M. S., van Dishoeck, E. F., and Lahuis, F.: 2003, Astron. Astrophys. 399, 1047. Boulanger, F., Falgarone, E., Puget, J.-L., et al.: 1990, Astrophys. J. 364, 136. Boulanger, F., Abergel, A., Bernard, J. -P., et al.: 1996, Astron. Astrophys. 312, 256. Boulanger, F., Abergel, A., Bernard, J.-P., et al., 1998a, in J. L. Yun and R. Liseau (eds.), Star Formation with ISO, Astronomical Society of the Pacific, San Francisco, p. 15. Boulanger, F., Boissel, P., Cesarsky, D., and Ryter, C., 1998b, Astron. Astrophys. 339, 194. Boulanger, F., Abergel, A., Cesarsky, D., et al.: 2000, in R. J. Laureijs, K. Leech, and M. F. Kessler (eds.), ISO Beyond Point Sources: Studies of Extended IR Emission, ESA SP-455. Boulanger, F., Lorente, R., Miville-Deschˆenes, M. A., et al., 2005, Astron. Astrophys. 436, 1151. Blum, J.: 2004, in A. N. Witt, G. C. Clayton, and B. T. Draine (eds.), Astrophysics of Dust, ASP Conference Series, vol. 309, p. 369. Bregman, J. D. and Temi, P.: 2001, Astrophys. J. 554, 126. Bregman, J. D.: 1989, in Interstellar dust, Proceedings of IAU Symposium, vol. 135, p. 109. Cambre´sy, L., Boulanger, F., Lagache, G., and Stepnik, B.: 2001, Astron. Astrophys. 375, 999. Cardelli, J., Clayton, G. C., and Mathis, J. S.: 1989, Astrophys. J. 345, 245. Caux, E., Ceccarelli, C., Pagani, L., et al.: 2002, Astron. Astrophys. 383, L9. Ceccarelli, C., Baluteau, J.-P., Walmsley, M., et al.: 2002, Astron. Astrophys. 383, 603. Cernicharo, J., Heras, A. M., Tielens, A. G. G. M., et al.: 2001, Astrophys. J. 546, L123. Cernicharo, J., Goicoechea, J. R., and Benilan, Y.: 2002, Astrophys. J. 580, 157. Cesarsky, D., Lequeux, J., Ryter, C., and Gerin, M., 2000a, Astron. Astrophys., 354, L87. Cesarsky, D., Jones, A. P., Lequeux, J., et al., 2000b, Astron. Astrophys. 358, 708. Chan, K.-W., Roellig, T. L., Onada, T., et al.: 2001, Astrophys. J. 546, 273. Cherchneff, I., Barker, J. R., and Tielens, A. G. G. M.: 1992, Astrophys. J. 401, 269. Chiar, J. E., Tielens, A. G. G. M., Whittet, D. C. B., et al.: 2000, Astrophys. J. 537, 749.

THE COOL ISM

269

Cohen, M., Tielens, A. G. G. M., Allamandola, L. J., et al.: 1985, Astrophys. J. 299, 93. Cohen, M., Allamandola, L. J., Tielens, A. G. G. M., et al.: 1986, Astrophys. J. 302, 737. Cohen, M., Tielens, A. G. G. M., Bregman, J., et al.: 1989, Astrophys. J. 341, 246. Cook, D. J., Schlemmer, S., Balucani, N., et al.: 1998, J. Phys. Chem. A, 102, 1465. Dartois, E. and d’Hendecourt, L. B.: 1997, Astron. Astrophys. 323, 534. D´esert, F.-X., Boulanger, F., and Puget, J.-L.: 1990, Astron. Astrophys. 237, 215. Dominik, C. and Tielens, A. G. G. M.: 1997, Astrophys. J. 480, 647. Draine, B. T.: 1989, in B. H. Kaldeich (ed.), Infrared Spectroscopy in Astronomy, ESA SP-290, p. 93. Draine, B. T.: 2003, Annu. Rev. Astron. Astrophys. 41, 241. Duley, W. W. and Williams, D. A.: 1981, Mon. Not. Roy. Astron. Soc 196, 269. Dupac, X., Bernard, J.-P., Boudet, N., et al.: 2003, Astron. Astrophys. 404, L11. Dwek, E., Arendt, R. G., and Fixsen, D. J.: 1997, Astrophys. J. 475, 565. del Burgo, C., Laureijs, R. J., Abraham, P., and Kiss, C., 2003, Mon. Not. Roy. Astron. Soc. 346, 403. del Burgo, C. and Laureijs, R. J.: 2004, MNRAS 360, 901. Falgarone, E., Panis, J. -F., Heithausen, A., et al.: 1998, Astron. Astrophys. 331, 669. Falgarone, E., Verstraete, L., Pineau des Forêts, G., Flower, D., and Puget, J.-L.: 1999, in F. Combes and G. Pineau des Forêts (eds.), H2 in Space, Cambridge University Press, Cambridge, p. 225. Falgarone, E., Verstraete, L., Pineau des Forêts, G., et al., in press, Astron. Astrophys. Feuchtgruber, H., Helmich, F. P., van Dishoeck, E. F., and Wright, C. M., 2000, Astrophys. J. 535, 111. F¨orster-Schreiber, N. M., Roussel, H., Sauvage, M., and Charmandaris, V.: 2004, Astron. Astrophys. 419, 501. Frenklach, M. and Feigelson, E. D.: 1989, Astrophys. J. 341, 372. Gautier, T. N. III, Boulanger, F., Perault, M., and Puget, J. L.: 1992, Astron. J. 103, 1313. Goicoechea, J. R., Rodriguez-Fernandez, N. J., and Cernicharo, J.: 2004, Astron. Astrophys. 600, 214. Gry, C., Boulanger, F., Nehm´e, C., et al.: 2002, Astron. Astrophys. 391, 675. Habart, E., Verstraete, L., Boulanger, F., et al.: 2001, Astron. Astrophys. 373, 702. Habart, E., Boulanger, F., Verstraete, L., et al.: 2003, Astron. Astrophys. 397, 623. Habart, E., Abergel, A., Walmsley, M., Teyssier, D., and Pety, J.: 2005, Astron. Astrophys. 437, 177. Harper, D. A., Low, F. J., Rieke, G. H., and Thronson, H. A. Jr., 1976, Astrophys. J. 205, 136. Herbstmeier, U., Abraham, P., Lemke, D., et al.: 1998, Astron. Astrophys. 332, 739. Herlin, N., Bohn, I., Reynaud, C., et al.: 1998, Astron. Astrophys. 330, 1127. Hollenbach, D. J. and Tielens, A. G. G. M.: 1999, Rev. Mod. Phys. 71, 173. Hony, S., van Kerckhoven, C., Peeters, E., et al.: 2001, Astron. Astrophys. 370, 1030. Hudgins, D. M. and Allamandola, L. J.: 1999, Astrophys. J. 516, L41. Hudgins, D. M., Bauschlicher, C. W., Allamandola, L. J., and Fetzer, J. C., 2000, J. Phys. Chem. A, 104, 3655. Ingalls, J. G., Reach, W. T., and Bania, T. M.: 2002, Astrophys. J. 579, 289. Ingalls, J. G., Miville-Deschˆenes, M. A., Reach, W. T., Noriega-Crespo, A., Carey, S. J., Boulanger, F., et al.: 2004, Astrophys. J. SS 154, 281. Joblin, C., Abergel, A., Bregman, J., et al.: 2000, in A. Salama, M. F. Kessler, K. Leech, and B. Schulz (eds.), ISO Beyond the Peaks: The 2nd ISO Workshop on Analytical Spectroscopy, ESA-SP 456. Joblin, C., Boissel, P., and L´eger, A.: 1995, Astron. Astrophys. 299, 835. Joblin, C., Toublanc, D., Boissel, P., and Tielens, A. G. G. M.: 2002, Mol. Phys. Jones, A. P., Tielens, A. G. G. M., Hollenbach, D. J., and McKee, C. F.: 1994, Astrophys. J. 433, 797. Jones, A. P., Tielens, A. G. G. M., and Hollenbach, D. J.: 1996, Astrophys. J. 469, 740. Jones, A. P. and d’Hendecourt, L. B.: 2000, Astron. Astrophys. 355, 1191. Juvela, M., Mattila, K., Lehtinen, K., et al.: 2002, Astron. Astrophys. 382, 583. Kahanp¨aa¨ , J., Mattila, K., Lehtinen, K., et al.: 2003, Astron. Astrophys. 405, 999.

270

A. ABERGEL ET AL.

Kemper, C., Spaans, M., Jansen, D. J., et al.: 1999, Astrophys. J. 515, 649. Kim, H. S., Wagner, D. R., and Saykally, R. J. 2001, Phys. Rev. Lett. 86, 5691. van Kerckhoven, C., Hony, S., Peeters, E., et al.: 2000, Astron. Astrophys. 357, 1013. Kiss, Cs., Abrahm, P., Klaas, U., et al.: 2003, Astron. Astrophys. 399, 177. Lagache, G., Abergel, A., Boulanger, F., et al.: 1998, Astron. Astrophys. 333, 709. Lamarre, J.-M., Pajot, F., Torre, J.-P., Guyot, G., Bernard, J. P., de Luca, A., et al.: 1994, Infrared Phys. Technol. 35, 277. Langhoff, S. R.: 1996, J. Phys. Chem. 100, 2819. Laureijs, R. J., Clark, F. O., and Prusti, T.: 1991, Astrophys. J. 371, 602. Laureijs, R. J., Haikala, L., Burgdorf, M., et al.: 1996, Astron. Astrophys. 315, 316. Le Bourlot, J., Pineau des Forˆets, G., Roueff, E., and Flower, D., 1993, Astron. Astrophys. 267, 233. L´eger, A. and Puget, J. L.: 1984, Astron. Astrophys. 137, L5. Lehtinen, K., Lemke, D., Mattila, K., and Haikala, L. K.: 1998, Astron. Astrophys. 333, 702. Li, A. and Draine, B. T.: 2001, Astrophys. J. 554, 778. Li, A. and Draine, B. T.: 2002, Astrophys. J. 572, 232. Liseau, R., White, G. J., Larsson, B., et al.: 1999, Astron. Astrophys. 344, 342. Lutz, D., Feuchtgruber, H., Genzel, R., et al.: 1996, Astron. Astrophys. 315, 269. Lutz, D., Spoon, H. W. W., Rigopoulou, D., et al.: 1998, Astrophys. J. 505, L103. Lutz, D.: 1999, in P. Cox and M. F. Kessler (eds.), The Universe as Seen by ISO, ESA-SP 427, vol. 2, p. 623. Mattila, K., Lemke, D., Haikala, L. K., et al.: 1996, Astron. Astrophys. 315, L353. Mathis, J. S., Mezger, P. G., and Panagia, N.: 1983, Astron. Astrophys. 128, 212. Melnick, G., Gull, G. E., and Harwit, M.: 1979, Astrophys. J. 227, 29. Mennella, V., Brucato, J. R., Colangeli, L., et al.: 1998, Astrophys. J. 496, 1058. Miville-Deschˆenes, M. A., Boulanger, F., Joncas, G., and Falgarone, E.: 2002, Astron. Astrophys. 381, 209. Miville-Deschˆenes M. A. 2003, in Chemistry as Diagnostic of Star Formation, NRC press, Canada, p. 363. Miville-Deschˆenes, M.-A., Joncas, G., Falgarone, E., and Boulanger, F.: 2003, Astron. Astrophys. 411, 109. Mizutani, M., Onaka, T., and Shibai, H.: 2004, Astron. Astrophys., 423, 579. Moutou, C., Verstraete, L., L´eger A., et al.: 2000, Astron. Astrophys. 354, 17. Neufeld, D. A., Zmuidzinas, J., Schilke, P., and Phillips, T. G.: 1997, Astrophys. J. 488, 141. Ossenkopf, V. and Henning, T.: 1994, Astron. Astrophys. 291, 943. Pauzat, F., Talbi, D., and Ellinger, Y.: 1997, Astron. Astrophys. 319, 318. Pech, C., Joblin, C., and Boissel, P.: 2002, Astron. Astrophys. 388, 639. Peeters, E., Hony, S., van Kerckhoven, C., and Tielens, A. G. G. M: 2002, Astron. Astrophys. 390, 1089. Peeters, E., Allamandola, L. J., Bauschlicher, C. W. Jr., et al.: 2004, Astrophys. J. 604, 252. Perault, M., Omont, A., Simon, G., et al.: 1996, Astron. Astrophys. 315, 165. Pety, J., Teyssier, D., Foss´e, D., Gerin, M., Roueff, E., Abergel, A., et al.: 2005, Astron. Astrophys. 435, 885. Polehampton, E. T., Baluteau, J.-P., and Ceccarelli, C.: 2002, Astron. Astrophys. 388, L44. Preibisch, T., Ossenkopf, V., Yorke, H.W., et al.: 1993, Astron. Astrophys. 279, 577. Puget, J.-L., L´eger, A., and Boulanger, F.: 1985, Astron. Astrophys. 142, L19. Rachford, B. L., Snow, T. P., Tumlinson, J., et al.: 2002, Astrophys. J. 577, 221. Rapacioli, M., Joblin, C., and Boissel, P.: in press, Astron. Astrophys. 429, 193. Roberts, H. and Herbst, E.: 2002, Astron. Astrophys. 395, 233. Roche, P. F., Aitken, D. K., and Smith, C. H.: 1989, Mon. Not. Roy. Astron. Soc. 236, 485. Roelfsema, P., Cox, P., Tielens, A. G. G. M., et al.: 1996, Astron. Astrophys. 315, L289.

THE COOL ISM

271

Rosenthal, D., Bertoldi, F., and Drapatz, S.: 2000, Astron. Astrophys. 356, 705. Schutte, W. A., Tielens, A. G. G. M., and Allamandola, L. J.: 1993, Astrophys. J. 415, 397. Schutte, W. A., van der Hucht, K. A., Whittet, D. C. B., et al.: 1998, Astron. Astrophys., 337, 261. Schneider, N., Simon, R., Kramer, C., et al.: 2003, Astron. Astrophys. 406, 915. Sellgren, K.: 1984, Astrophys. J. 277, 623. Sellgren, K., Allamandola, L. J., Bregman, J. D., et al.: 1985, Astrophys. J. 299, 416. Stepnik, B.: 2001, Thesis, Universit´e of Paris. Stepnik, B., Abergel, A., Bernard, J.-P., et al.: 2003, Astron. Astrophys. 398, 551. Stepnik, B., Abergel, A., Jones, A. P., et al.: in press, Astron. Astrophys. Szczepanski, J. and Vala, M.: 1993, Astrophys. J. 414, 646. Teixeira, T. C. and Emerson, J. P.: 1999, Astron. Astrophys. 352, 517. Tielens, A. G. G. M., Wooden, D. H., Allamandola, L. J., et al., 1996, Astrophys. J. 461, 210. Timmermann, R., K¨oster, B., and Stutzki, J.: 1998, Astron. Astrophys. 336, L53. T´oth, L. V., Hotzel, S., Krause, O., et al.: 2000, Astron. Astrophys. 364, 769. Uchida, K. I., Sellgren, K., and Werner, M. W.: 1998, Astrophys. J. 493, L109. Uchida, K. I., Sellgren, K., Werner, M. W., et al.: 2000, Astrophys. J. 530, 817. van Diedenhoven, B., Peeters, E., van Kerckhoven, C., et al., 2004, Astrophys. J. 611, 928. van Dishoeck, E. F.: 2004, Annu. Rev. Astron. Astrophys. 42, 119. Vastel, C., Polehampton, E. T., Baluteau, J.-P., et al.: 2002, Astrophys. J. 581, 315. Verstraete, L., Puget, J. L., Falgarone, E., et al.: 1996, Astron. Astrophys. 315, L337. Verstraete, L., Pech, C., Moutou, C., et al.: 2001, Astron. Astrophys. 372, 981. Weingartner, J. C. and Draine, B. T., 2001a, Astrophys. J. Suppl. 134, 263. Weingartner, J. C. and Draine, B. T., 2001b, Astrophys. J. 548, 296. Weingartner, J. C. and Draine, B. T., 2001c, Astrophys. J. 553, 581. Werner, M. W., Uchida, K. I., Sellgren, K., et al.: 2004, Astrophys. J. Suppl. 154, 309. Wright, C. M., van Dishoeck, E. F., Cox, P., et al.: 1999, Astrophys. J. 515, L29.

HIGH EXCITATION ISM AND GAS 2 ´ ELS PEETERS1,∗ , NIEVES LETICIA MART´IN-HERNANDEZ , 3 ´ NEMESIO J. RODR´IGUEZ-FERNANDEZ and XANDER TIELENS4 1 NASA

Ames Research Center, U.S.A. de Geneve, Switzerland 3 LUTH/LERMA – Observatoire de Paris, France 4 Kapteyn Astronomical Institute, The Netherlands ∗ ( Author for correspondence: E-mail: [email protected]) 2 Observatoire

(Received 6 October 2004; Accepted in final form 3 December 2004)

Abstract. An overview is given of ISO results on regions of high excitation ISM and gas, i.e. H II regions, the Galactic Centre and Supernova Remnants. IR emission due to fine-structure lines, molecular hydrogen, silicates, polycyclic aromatic hydrocarbons and dust are summarised, their diagnostic capabilities illustrated and their implications highlighted. Keywords: IR, H II regions, galactic centre, supernova remnants, LATEX

1. Introduction Massive stars provide much of the radiative stellar energy of the Milky Way. Their copious amount of UV radiation has a great impact on the surroundings of the massive star. Indeed, their UV radiation dissociates molecules and dust in the interstellar medium (ISM) and ionises hydrogen, creating large H II regions. In addition, their powerful winds and supernova explosions provide most of the mechanical energy of the Galaxy which dominate the structure of the ISM. Furthermore, the nuclear reactions during their lifetime and death scene synthesis most of the intermediate mass elements and likely the r-process elements. Much of these nucleosynthetic products may condense in the form of small dust grains. Indeed, supernovae may dominate the dust mass budget of the ISM. Because massive stars evolve so fast, they are generally associated with the remnants of their “cradle” and are heavily enshrouded in dust and gas. This dust and gas absorbs most of the stellar luminosity which is then re-emitted in the infrared. Hence, due to the high degree of obscuration, massive stars can be best studied at wavelengths longer than 2 μm. Offering unique combinations of wavelength coverage, sensitivity, and spatial and spectral resolutions in the infrared spectral region, ISO opened the infrared Universe. Its spectral coverage (from 2.3–196 μm) gives for the first time access to nearly all the atomic fine-structure and hydrogen recombination lines in the infrared Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands, and the United Kingdom), and with the participation of ISAS and NASA. Space Science Reviews (2005) 119: 273–292 DOI: 10.1007/s11214-005-8070-1

 C

Springer 2005

274

E. PEETERS ET AL.

range. In addition to the atomic lines, ISO revealed the shape and strength of the dust continuum and several emission features. This presented an unprecedented view of the luminous but dusty and obscured Universe, from the Galactic Centre to the most extincted regions of the Milky Way and other galaxies. This chapter reviews the major achievements of ISO on massive stars. Section 2 summarises the results on H II regions, the Galactic Centre is discussed in Section 3. Subsequently, Section 4 highlights the ISO results on supernova remnants. 2. H II Regions The overall mid-IR (MIR) spectrum of H II regions is dominated by a dust continuum which rises strongly towards longer wavelengths (Figure 1). On top of this continuum, there are a multitude of fine-structure lines and hydrogen recombination lines. The continuum emission is dominated by strong emission features at 3.3, 6.2, 7.7, 8.6 and 11.2 μm, generally attributed to Polycyclic Aromatic Hydrocarbons (PAHs). In some case, broad absorption features due to simple molecules (H2 O, CO, CO2 ) in an icy mantle and/or narrow emission or absorption lines due to gaseous molecules (H2 , H2 O, CO2 , OH, C2 H2 ) are present. 2.1. THE LINE S PECTRUM 2.1.1. The Ionised Gas H II regions are prime targets to derive the present-day elemental abundances of the ISM. They are bright and are characterised by a large number of emission

Figure 1. The ISO-SWS/LWS spectra of a typical H II region (Peeters et al., 2002b).

HIGH EXCITATION ISM AND GAS

275

lines. Optical studies of abundance gradients in the Galaxy, however, present the problem that many H II regions are highly obscured by dust lying close to the Galactic plane. The arrival of the Kuiper Airborne Observatory (e.g. Simpson et al., 1995; Afflerbach et al., 1997; Rudolph et al., 1997) and the Infrared Astronomical Satellite (e.g. Simpson and Rubin, 1990) permitted measurements of elemental abundances of embedded compact and ultra-compact H II regions, obscured at optical wavelengths but observable in the infrared, and gave access for the first time to the central regions of the Galaxy. Regarding the determination of elemental abundances, the use of infrared fine-structure lines present clear advantages with respect to the optical lines (e.g. Rubin et al., 1988; Simpson et al., 1995): (1) they are attenuated much less due to the presence of dust; (2) they are practically insensitive to the precise temperature of the emitting gas since they are emitted from levels with very low excitation energies; and (3) the infrared range is the only wavelength regime to measure the dominant form of nitrogen in highly excited H II regions (N2+ ). ISO provided a unique opportunity to measure the full spectral range from 2.3 to 196 μm of a large number of (ultra)compact H II regions with relatively good spectral and spatial resolution (e.g. Peeters et al., 2002b). Figure 1 shows the combined SWS and LWS spectrum of a typical H II region. The spectrum is dominated by recombination lines of H and fine-structure lines of C, N, O, Ne, S, Ar and Si. The lines of [C II], [O I], and [Si II] are produced by ions with ionisation potentials lower than 13.6 eV and are thus expected to be mostly emitted in the PDR surrounding the H II region (see Section 2.1.2). Two lines are present for some of the ions (O2+ , Ne2+ and S2+ ), providing a handle on the electron density. Typically, [O III] 52, 88 μm densities towards Galactic H II regions range between ∼100 and 3000 cm−3 (cf. Mart´ın-Hern´andez et al., 2002). The use of the [S III] 18.7, 33.5 μm and [Ne III] 15.5, 36.0 μm line ratios requires careful aperture corrections. When such corrections are applied, the [S III] and [Ne III] densities agree well with the [O III] densities within the errors (cf. Mart´ın-Hern´andez et al., 2003). N, Ne, Ar and S are observed in two different ionisation stages, which enormously alleviates the problem of applying ionisation correction factors (cf. Mart´ın-Hern´andez et al., 2002). The ISO observations of H II regions covered the Galactic plane from the centre to a Galactocentric distance of about 15 kpc, giving thus the possibility of investigating trends of relative and absolute elemental abundances across a large part of the Galactic disk. The gradients resulting from the full samples of Mart´ın-Hern´andez et al. (2002) and Giveon et al. (2002) are  log(Ne/H) = −0.039 ± 0.007,  log(Ar/H) = −0.047 ± 0.007,  log(S/H) = −0.023 ± 0.014 – re-computed by Giveon et al. (2002) applying the same extinction and Te corrections to both data sets – and  log(N/O) = −0.056 ± 0.009 dex kpc−1 . Figure 2 shows the neon abundance as a function of the distance from the Galactic Centre. The direct observation of two different ionisation stages not only facilitates the determination of elemental abundances, but allows us to probe the ionisation structure of the H II regions and constrain the stellar energy distribution (SED) of the ionising stars. Line ratios such as [N III]/[N II] 57/122 μm, [Ne III]/[Ne II]

276

E. PEETERS ET AL.

Figure 2. Neon abundance as a function of the distance to the Galactic Centre (cf. Mart´ın-Hern´andez et al., 2002a).

15.5/12.8 μm, [S IV]/[S III] 10.5/18.7 μm and [Ar III]/[Ar II] 9.0/7.0 μm probe the ionising stellar spectrum between 27.6 and 41 eV and depend on the shape of the SED and the nebular geometry (see e.g. Morisset et al., 2002, who present a detailed model of the well studied H II region G29.96–0.02 based on their infrared lines and Morisset et al., 2004, who compared predicted ionising spectra against ISO observations of Galactic H II regions). These line ratios are found to correlate well with each other for the large sample of H II regions observed by ISO (cf. Figure 3) and to increase with Galactocentric distance (cf. Giveon et al., 2002; Mart´ın-Hern´andez et al., 2002). The observed [Ne III]/[Ne II] and [S IV]/[S III] line ratios have been compared with diagnostic diagrams built from extensive photoionisation model grids computed for single-star H II regions using stellar atmosphere models from the WM-Basic code (Pauldrach et al., 2001) where the metallicities of both the star and the nebula have been taking into account (Morisset, 2004). This comparison finds no evidence of a gradient of the effective temperature of the ionising stars with the Galactocentric distance, attributing the observed increase of excitation with distance mainly to the effect of the metallicity gradient on the SED (see also Mart´ın-Hern´andez et al., 2002 and Mokiem et al., 2004). The relation between the [Ne III]/[Ne II] line ratio and the Ne/H elemental abundance for a combined sample of Galactic and extra-galactic H II regions is shown in the right panel of Figure 3. As it is evident from this figure, a clear correlation between excitation and metallicity exists, albeit with a large scatter. Extended emission of highly ionised species (fine-structure lines of N+ , N2+ , 2+ O and S2+ ) have been detected by LWS and SWS (Mizutani et al., 2002; Okada et al., 2003; Goicoechea et al., 2004). This reveals the presence of extended, highly ionised gas surrounding H II regions and probably ionised by the central O-star(s). The electron density has been derived from the [O III] 52 to 88 μm ratio, and values

HIGH EXCITATION ISM AND GAS

277

Figure 3. Left: Relation between the [S IV]/[S III] 10.5/18.7 μm and [Ne III]/[Ne II] 15.5/12.8 μm line ratios for a sample of H II regions (cf. Mart´ın-Hern´andez et al., 2002b). Indicated by various symbols are Galactic regions at Galactocentric distance Rgal < 7 kpc (solid triangles), Galactic regions at Rgal > 7 kpc (solid circles), LMC regions (open triangles, except 30 Doradus, which is indicated by an open square) and SMC regions (open diamonds). The positions of Sgr A (Lutz et al., 1996) and the Orion nebula (Simpson et al., 1998) are indicated by a solid star and a solid square, respectively. The dotted line is a least squares fit to the data. A typical error bar is given in the upper left corner. The arrow in the lower right corner indicates the correction due to an extinction AK = 2 mag. Right: Relation between the [Ne III]/[Ne II] 15.5/12.8 μm line ratio and the Ne/H elemental abundance for a combined sample of H II regions (Mart´ın-Hern´andez et al., 2003). Indicated by various symbols are Galactic H II regions (solid circles), Sgr A (solid star), the orior nebula (star), the Pistol and the Sickle (open and solid diamonds, respectively; Rodr´ıguez-Fern´andez et al., 2001a), a sample of H II regions in M33 (open squares; Willner and Nelson-Patel, 2002), LMC H II regions (reverse open triangles, except 30 Doradus, which is plotted as a plus sign) and 2 regions in the SMC (open triangles).

of a few 10 to a few 100 cm−3 have been found. This extended, ionised gas – denser than the galactic warm ionised medium – would represent a new phase of the ISM as stressed by Mizutani et al. (2002). Extragalactic studies of H II regions have been performed in the Magellanic Clouds (cf. Vermeij et al., 2002a; Vermeij and van der Hulst, 2002b) and M33 (Willner and Nelson-Patel, 2002). H II regions in the Magellanic Clouds are characterised by low metallicities and high excitation (see e.g. Figure 3). Towards M33, the distribution of neon abundances as a function of Galactocentric radius is best described as a step gradient, with a slope of −0.15 dex from 0.7 to 4.0 kpc and −0.35 dex from 4.0 to 6.7 kpc. 2.1.2. The Photodissociation Region Photodissociation Regions (PDRs) are regions where FUV (6 < hν < 13.6 eV) photons dissociate, ionise and heat neutral atomic/molecular gas surrounding H II regions (Tielens and Hollenbach, 1985; Hollenbach and Tielens, 1999). The gas reaches temperatures in the range 100–1000 K, much warmer than the dust (10– 100 K), and radiates its energy in low energy atomic fine-structure lines, particularly

278

E. PEETERS ET AL.

the [O I] 63 μm, [C II] 157 μm, and [Si II] 34 μm lines, and in pure rotational molecular hydrogen lines (at wavelengths between 28 and 2 μm). In addition, the strong FUV flux pumps molecular hydrogen molecules into excited vibrational states and their cascade produces strong fluorescent ro-vibrational lines in the near-IR. The dust gives rise to a continuum at long wavelengths (λ > 25 μm). In addition, the IR spectrum of PDRs shows broad emission features at 3.3, 6.2, 7.7, 8.6, 11.2, and 12.7 μm due to fluorescent emission of FUV pumped Polycyclic Aromatic Hydrocarbon (PAHs) molecules (see Section 2.2). Here, we will focus on PDRs associated to H II regions; other PDRs are extensively discussed elsewhere in this book. Because of their luke warm temperatures, PDRs were only ‘discovered’ when the IR window was opened by the Kuiper Airborne Observatory in the late 1970s and 1980s, but – because of limited sensitivities – these studies focused on the brightest objects in the sky (the Orion bar; Melnick et al., 1979; Storey et al., 1979; Russell et al., 1980, 1981). The increased sensitivities of the SWS and LWS and, in particular, their wide wavelength coverage allowed for the first time a systematic study of the properties of PDRs. Much of this data are however still resting in the archives. Except for the some of the ionic fine-structure lines and the H I recombination lines, much of the IR emission characteristics of H II regions (Figure 1), originate in the associated PDRs. The SWS is particularly particularly suited for observations of the pure rotational lines of molecular hydrogen. These lines provide a direct handle on the physical properties of the emitting gas. Because of the small Einstein A’s of the associated quadrupole transitions, the lowest levels of H2 have very low critical densities and hence their populations are in local thermodynamic equilibrium for typical PDR conditions (Burton et al., 1992). A rotational diagram provides then directly the temperature and column density of the emitting gas. This is illustrated in Figure 4, left panel, for the PDR associated with the blister H II region, S140 (Timmermann et al., 1996). The optically bright rim is produced by the ionisation of a dense clump in the molecular cloud, L1202/1204 by the nearby B0.5 star, HD 211880 (Blair et al., 1978; Evans et al., 1987). The SWS spectrum of this source shows pure rotational H2 lines up to 0 − 0 S(9). The column density and temperature derived from these observations are N 1020 cm−2 and Trot = 500 K (Timmermann et al., 1996). Similar studies of the pure rotational lines in the spectra of S106 and Cep A resulted in excitation temperatures of 500 and 700 K (van den Ancker et al., 2000; Wright et al., 1996). In addition, SWS has observed several ro-vibrational lines in S140 belonging to the v = 1–0, 2–1, and 3–2 series. These levels are populated by FUV pumping. Figure 4, left panel, compares the observations with a detailed model for the collisional and FUV-pumping excitation of H2 in this source (Timmermann et al., 1996). Good agreement is obtained for a density of 104 cm−3 in the PDR and an incident FUV flux of 400 Habing. These values are consistent with estimates based upon pressure equilibrium across the ionisation front coupled with the ionised gas density in the associated bright rim and the properties of the illuminating star.

HIGH EXCITATION ISM AND GAS

279

Figure 4. Left: The rotational diagram of molecular lines observed towards the bright rim cloud, S140 (Timmermann et al., 1996). Observed (filled symbols) and model calculated (open symbols) column densities (ratioed to the statistical weight of the level involved) are plotted as a function of the energy of the upper level. A Boltzmann distribution will show a straight line on this plot. The low lying pure rotational lines are consistent with gas at a kinetic temperature of 500 K. The higher excitation temperature of the vibrationally excited levels at high energies indicate the increased importance of FUV pumping for the populations of these levels. The models results are calculated for an incident FUV field of 400 times the average interstellar radiation field and a gas density of 104 cm−3 . A foreground extinction of 0.3 magnitudes in the K band has been adopted. To indicates the adopted peak temperature of the gas in the PDR. Right: A diagnostic diagram for PDRs based upon the intensity ratio of the [C II] 158 μm and the [O I] 63 μm lines and the overall cooling efficiency of the gas in the PDR. The lines present the results of detailed model calculations for a range in densities and incident FUV fields (Kaufman et al., 1999). The data are taken from LWS observations of a sample of H II regions (Peeters et al., 2002b; Peeters et al., 2005, in preparation; Vermeij et al., 2002a).

While this agreement between observations and models of H2 excitation is very comforting (Figure 4, left panel), PDR models have a problem in matching the observed kinetic temperatures and column densities derived from the low lying pure rotational lines. In particular, for such low FUV fields and densities as derived for S140, the calculated gas temperatures in the PDR are much lower than the observed 500 K (e.g. Draine and Bertoldi, 1999). This problem seems to be more general, although a detailed comparison of observations and models for regions spanning a wide range in physical properties is still wanting. Nevertheless, it seems that PDR models may be missing an essential part of the energetic coupling between the gas and the illuminating stellar FUV radiation field. The far-IR atomic fine-structure lines provide important probes of the physical conditions in PDRs. However, at present, little has been done to harvest the data provided by the LWS. The [O I] 63 μm and [C II] 157 μm lines are the dominant coolants of PDRs (Tielens and Hollenbach, 1985). Hence, the summed flux of these lines measures the total heating of PDR gas by the FUV radiation field. Comparison with the total incident FUV flux, as measured by the far-IR dust continuum

280

E. PEETERS ET AL.

provides a measure of the heating efficiency of PDRs. The levels involved in these two transitions have very different critical densities (∼105 and 3 × 103 cm−3 ) and excitation energies (92 and 228 K). Not coincidentally, these values are precisely the range expected for gas in PDRs and, hence, the ratio of these lines has evolved to a major diagnostic of the properties of PDRs. This is illustrated in Figure 4, right panel, where the results of PDR models for this line ratio are plotted against the calculated ratio of the cooling lines as a function of density and incident FUV field (Kaufman et al., 1999). Over much of the relevant range of these observables, the model result segregate out well and hence this diagram is very useful for the analysis of the conditions in PDRs. Observations towards a sample of H II regions are also shown in Figure 4, right panel. The conditions in these regions span a range in density, 30–3×103 cm−3 , and incident FUV fields, 3 × 102 –3 × 104 Habing. These values are within the range expected for PDRs associated with (evolved) compact H II regions. In addition, there seems to be a trend in the distribution of the observed points with location in the galaxy or, equivalently, metallicity. That is regions with lower metallicity seem to be characterised with a higher heating efficiency and lower [C II]/[O I] ratio than regions with higher metallicity. If the [O I] line is the dominant cooling line, a decrease in the metallicity is expected to result in a decreased [C II]/[O I] ratio (Tielens and Hollenbach, 1985). However, the decrease in the heating rate with increasing metallicity is not expected. Further analysis of these types of observations is clearly warranted. The fine-structure lines observed towards several individual high mass star forming regions have been analysed in detail. Analysis of the observations of W49N show that the PDR is illuminated by an intense FUV field (G 0 = 3 × 105 ). The density and temperature are, however, quite moderate (n = 104 cm−3 , T = 130 K). Reflecting this high FUV field and low density, the observed heating efficiency is comparatively low, ∼10−4 (Vastel et al., 2001). Towards S106IR, a hot (200–500 K) dense (n > 3 × 105 cm−3 ) gas component is present. Cooler gas is associated with the bulk of the emission of the molecular cloud and is characterised by two emission peaks which have densities of 105 cm−3 and are illuminated by a radiation field, G 0 , ranging from 102 to 103.5 (Schneider et al., 2003). The H II region S 125 is modelled self-consistently with a two-dimensional geometrical blister model. In order to fit the spatial profile, a systematic increase of the gas temperature along the PDR boundary with decreasing distance from the ionising star is necessary, which is not readily understood within present-day PDR models (Aannestad and Emery, 2003). Despite the rich concentration of massive stars in the stellar cluster Trumpler 14, in the Carina nebula, the physical conditions in the associated PDR are much less extreme than for Orion or M17 (Brooks et al., 2003). The data for the PDR in NGC 2024, on the other hand, are consistent with emission from dense (n 106 cm−3 ), coolish (T 100 K) clumps (Giannini et al., 2000). Finally, the lowest pure rotational J = 1 − 0 transition of the HD molecule at 112 μm was detected towards the Orion Bar using the LWS on ISO (Wright

HIGH EXCITATION ISM AND GAS

281

et al., 1999). The observations imply a total HD column density of (3 ± 0.8) × 1017 cm−2 for adopted temperatures in the range 85–300 K. This corresponds to an elemental D/H abundance ratio of (2 ± 0.6) × 10−5 , comparable to that derived from D I and H I ultraviolet absorption measurements in the solar neighbourhood.

2.2. PAHS

AND

DUST

Besides the well-known UIR bands at 3.3, 6.2, 7.6/7.8, 8.6, 11.2 and 12.7 μm, and very broad structures present underneath these bands, many discrete weaker bands and subcomponents can be found at 3.4, 3.5, 5.2, 5.7, 6.0, 6.6, 7.2–7.4, 8.2, 10.8, 11.0, 12.0, 13.2 and 14.5 μm. ISO extended the observable wavelength range and found there new feature at 16.4 μm and 17.4 μm as well as a weak, variable, emission plateau between 15 and 20 μm (Moutou et al., 2000; Van Kerckhoven et al., 2000; Van Kerckhoven, 2002). It should be emphasized that not all sources show all these emission features at the same time. ISO also allowed, for the first time, a systematic analysis of the UIR bands in a wide variety of environments. It is now firmly established that the detailed characteristics (intensity, peak position, profile) of the UIR features vary from source to source and also spatially within extended sources (for a recent review, see Peeters et al., 2004b). ISO-SWS spectra indicated that the UIR band profiles and positions of all H II regions are equal to each other and to those observed in Reflection Nebulae (RNe), non-isolated Herbig AeBe stars and the ISM, while they are clearly distinct from those found around evolved stars and isolated Herbig AeBe stars (Figure 5, Peeters et al., 2002a; van Diedenhoven et al., 2004). However, spatial variations within RNe (Bregman and Temi, 2005) and evolved stars (Kerr et al., 1999; Song et al., 2004; Miyata et al., 2004) are observed suggesting that in all sources profile variations occur on a small spatial scale. In addition – despite of what the integrated spectra of sources might suggest – the profiles are not unique to certain object types. In contrast, their relative strength in H II regions varies both from source to source and within sources (Figure 5). Integrated spectra of sources show variations in the relative strength of the CC modes (6–9 μm range) relative to the CH modes (at 3.3 and 11.2 μm), as is the 11.2/12.7 band ratio while the 3.3 μm feature correlates with the 11.2 μm feature and the 6.2 μm feature with the 7.7 μm feature (Figure 5, Roelfsema et al., 1996; Verstraete et al., 1996; Hony et al., 2001; Vermeij et al., 2002c). These variations are directly related to object type (Hony et al., 2001) and to the local environment (Figure 5, Verstraete et al., 1996; Vermeij et al., 2002c). Within extended sources, additional variations are found, nicely demonstrated by that of the 6.2/7.7 band ratio in S106 (Joblin et al., 2000). Within a single object, variations in G 0 are important in driving the UIR emission spectrum (Onaka, 2000; Bregman and Temi, 2005). However, the source to source variation does not seem to follow G 0 (Verstraete et al., 2001; Hony et al.,

282

E. PEETERS ET AL.

Figure 5. Left: Source to source variations for the 3–12 μm UIR bands. Profile A represent H II regions, Reflection Nebulae and the ISM, profile B isolated HAeBe stars and most PNe and profile C 2 post-AGB stars (Peeters et al., 2002a; van Diedenhoven et al., 2004). Right: PAH band strength ratios for H II regions. Galactic sources are shown as (), non-30 Dor sources as (), 30 Dor pointings as () and the SMC B11 molecular cloud (Reach et al., 2000a) by an asterisk (Vermeij et al., 2002).

2001; Peeters et al., 2002a, 2004c; Vermeij et al., 2002c; van Diedenhoven et al., 2004). The observed variations of the UIR bands provide direct clues to the physical and chemical properties of their carriers, the local physical conditions and/or the local history of the emitting population. These spectral variations have been attributed to a range of different physical or chemical characteristics of the carriers, including charge state, anharmonicity, different subcomponents with variable strength, family of related species with varying composition, substituted/complexed PAHs, isotope variations, clustering, molecular structure, molecular composition etc. (Verstraete et al., 1996, 2001; Joblin et al., 2000; Hony et al., 2001; Pech et al., 2002; Peeters et al., 2002a; Wada et al., 2003; Song et al., 2004; Bregman and Temi, in press; Hudgins and Allamandola, 2004; van Diedenhoven et al., 2004). In particular, the variations in the 11.2/6.2 and 3.3/6.2 μm ratios likely reflect variations in the average ionisation state of the emitting PAHs. In contrast, the variations in the peak position of the 6.2 μm band point towards variations in the molecular structure of the emitting species perhaps related to the incorporation of N into the ring structure of the PAHs. Since the UIR bands are omnipresent, they serve as important probes of the different emission zones. For example, the presence and strength of the UIR bands are generally thought to trace star formation and so they are used as qualitative and quantitative diagnostics of the physical processes powering Galactic nuclei (see this

HIGH EXCITATION ISM AND GAS

283

issue). In addition, deuterated PAHs are tentatively detected at 4.4 and 4.65 μm at high abundances in the Orion Bar and M17 (Peeters et al., 2004a). If born out, they can serve as probes of PAH D/H ratio in various regions. Several studies were devoted to the spatial distribution of the different emission components (Cesarsky et al., 1996; Verstraete et al., 1996; Pilbratt et al., 1998; Cr´ete et al., 1999; Klein et al., 1999; Zavagno and Ducci, 2001; Urquhart et al., 2003; Verma et al., 2003). It is overwhelmingly clear that the UIR bands peak in the PDR, though closer to the ionised star with respect to H2 , while the strong dust continuum peaks towards the H II region. Inside the H II region, where UIR bands are weak or absent, remains a broad underlying emission band which is attributed to very small grains (VSG, Cesarsky et al., 1996; Jones et al., 1999). The spatial distribution of the UIR bands is also displaced from that of the ERE emission in Sh 152 (Darbon et al., 2000), similar to the Red Rectangle (Kerr et al., 1999). It is clear that the carrier of the UIR bands is not also responsible for the ERE. Moreover, because the ERE emission can sometimes be traced to inside the ionised gas and recalling that the PAHs do not survive in the H II region, the carrier of the ERE is unlikely to be molecular in origin. The dust/UIR band emission has been modeled for several sources. Cr´ete et al. (1999) modeled M17 using a two-component model composed of PAHs as the carrier of the UIR bands and VSG responsible for the “warm” continuum and a two-component model composed of coal dust emission and big grain emission. In the frame of the coal dust model, the observations required nano-sized, transiently heated particles. The IR emission in the M17-SW H II region and Orion is modeled with a mixture of amorphous carbons, silicates and possibly crystalline silicates (see below, Jones et al., 1999; Cesarsky et al., 1996). PAHs seem to be depleted inside the H II region consistent with their destruction in the intense radiation field of M17 and Orion (Jones et al., 1999; Cesarsky et al., 2000; Cr´ete et al., 1999). Modelling of the FIR emission of Sh 125 also indicates the dust to be severely depleted in the H II region while normal dust/gas ratio is observed in the PDR, but with a major fraction in the form of VSG (Aannestad and Emery, 2001). ISO also increased the number of H II regions showing silicate in emission (Cesarsky et al., 2000; Peeters et al., 2002b; Peeters, 2002) and detected several new dust features: (1) A very strong, broad 8.6 μm band is detected towards the M17-SW H II region and three compact H II regions, clearly distinct from the “classical” 8.6 μm PAH band and originating inside the H II regions (Roelfsema et al., 1996; Cesarsky et al., 1996; Verstraete et al., 1996; Peeters et al., 2002a; Peeters et al., 2005). Detailed analysis reveals the feature having a profile peaking at 8.9 μm with a FWHM ∼1 μm and likely not being related to PAH emission (Peeters et al., 2005). (2) A broad 22 μm feature is detected in the M17-SW H II region (Jones et al., 1999), the Carina Nebula and two starburst galaxies (Chan and Onaka, 2000).

284

E. PEETERS ET AL.

The feature shape is similar to that of the 22 μm band observed in Cassiopeia A indicating that supernovae are probably the dominant production sources (Chan and Onaka, 2000). (3) Several (possible) detections of crystalline silicates associated with H II regions are reported, i.e. (i) emission at 34 μm in the M17 H II region attributed to Mg-rich crystalline olivine (Jones et al., 1999); (ii) emission bands at 9.6 μm in θ 2 Ori A and at ∼15–20 μm, ∼20–28 μm and longward of 32 μm in the Orion H II region attributed to crystalline Mg-rich silicate fosterite (Cesarsky et al., 2000). The latter emission bands are however a combination of instrumental effects and the PAH 15–20 μm plateau (Kemper et al., in preparation); (iii) a band at 65 μm in the Carina Nebula and Sh 171 attributed to diopside (Cabearing crystalline silicate – Onaka and Okada, 2003). An upper-limit for the crystallinity of the silicates in the ISM is derived being a few percents (Jones et al., 1999), less than 5% (Li and Draine, 2001) and less than 0.4% (Kemper et al., 2004). (4) To end, a broad emission feature is found at 100 μm in the Carina Nebula and Sh 171, possibly due to carbon onion grains (Onaka and Okada, 2003). A similar broad feature around 90 μm is reported in evolved stars and a low mass protostar, attributed to calcite, a Ca-bearing carbonate mineral (Ceccarelli et al., 2002; Kemper et al., 2002).

3. The Galactic Centre The 500 central pc of the Galaxy (hereafter Galactic center, GC) are an extended source of emission at all wavelengths from radio to γ -rays (see Mezger et al., 1996 for a review). The non-thermal radio emission as well as γ - and X-rays observations show the presence of a very hot plasma (107 –108 K) and a recent episode of nucleosynthesis (∼106 year ago) that suggest an event of violent star formation in the recent past. On the other hand, the cold medium in the GC is mainly known by radio observations of the molecular gas. In between these hot and cold phases, there is a warm (few hundred K) neutral medium and a thermally ionised medium. ISO has made important contributions to our understanding of the heating of the warm neutral gas and the properties of the ionised gas in the GC. Before ISO, the ionised gas has been mainly studied by radio continuum and hydrogen radio recombination lines observations. They showed an extended diffuse ionised gas component and a number of discrete H II regions, most of them associated with the Sgr A, B, . . . complexes. Only the most prominent H II regions (Sgr A) and ionised nebulae (the Sickle) had been observed in infrared fine-structure lines. ISO observations of fine-structure lines allowed studying the properties of the ionised gas and the ionising radiation over the whole GC region. In the vicinity of Sgr A∗ , where the ionising source is the central stellar cluster, Lutz et al.

HIGH EXCITATION ISM AND GAS

285

(1996) derived an effective temperature of the ionising radiation of ∼35000 K and Rodr´ıguez-Fern´andez et al. (2001a) derived a similar effective temperature in the Radio Arc region. They also showed unambiguously that the Quintuplet and the Arches clusters ionise the Sickle nebula and the thermal filaments that seem to connect Sgr A to the Radio Arc. Indeed, the combined effects of both clusters ionise a region of more that 40 × 40 pc2 . Their photoionisation model simulations showed that, in order to explain the long-range effects of the radiation, the ISM that surrounds the clusters must be highly inhomogeneous. A similar scenario has been invoked by Goicoechea et al. (2004) to explain the extended ionised gas component in the Sgr B molecular complex. Indeed, the whole GC region is permeated by relatively hot (∼35000 K) but diluted ionising radiation field (Rodr´ıguez-Fern´andez and Mart´ın-Pintado, 2005). The line ratios ([Ne III]/[Ne II] 15.5/12.8 μm, [N III]/[N II]57/122 μm . . .) measured in the GC are similar to those found in low excitation starburst galaxies as M 83 or IC 342 but somewhat lower than those measured in other starburst galaxies as NGC 3256 or NGC 4945. However, the low ratios do not imply intrinsic differences in the star formation activity as a lower upper mass cutoff of the initial mass function. On the contrary, they are probably due to starburst of short duration that produce hot massive stars but whose age quickly softens the radiation (Thornley et al., 2000). The ISO observations of the GC are consistent with a burst of star formation less than 7 Myr ago. On the other hand, in the GC there is a widespread neutral gas component with temperatures of ∼200 K and without associated warm dust. Before ISO, this warm neutral gas had only been studied by radio observations of symmetric top molecules like NH3 . How this gas is heated has been a long-standing problem. Radiative heating mechanisms are usually ruled out due to the apparent lack of continuum sources and the discrepancy of gas and dust temperatures. ISO was perfectly suited to study the heating of this gas. First, Rodr´ıguez-Fern´andez et al. (2001b) have measured for the first time the total amount of warm gas in the GC clouds by observing H2 pure rotational lines. These lines trace gas with temperature from 150 K to 500 K. The column density of gas with a temperature of 150 K is a few 1022 cm−2 and, on average, it is around 30% of the total gas column density. Second, ISO has observed the main coolants of the neutral gas (the [C II] 158 and [O I] 63 μm lines) with temperatures of a few hundred K and the full continuum spectrum of the dust emission around the maximum. The dust emission can be characterized by two temperature components, one with a temperature of ∼15 K and a warmer one with a temperature that varies from source to source from ∼25 to ∼45 K (Lis et al., 2001; Rodr´ıguez-Fern´andez et al., 2004; Goicoechea et al., 2004). A detailed comparison of the lines and continuum emission with shocks and PDR models show that the fine structure lines and the H2 emission from excited levels arise in a PDR with a far-ultraviolet incident radiation field 103 times higher than the average local interstellar radiation field and a density of 103 cm−3 (Figure 6; Rodr´ıguez-Fern´andez et al., 2004). The warm dust emission and

286

E. PEETERS ET AL.

Figure 6. Intensities of two fine-structure lines and two H2 pure-rotational lines versus the farultraviolet incident flux derived for the sources observed by Rodr´ıguez-Fern´andez et al. (2004). For comparison, the PDR model predictions (Tielens and Hollenbach, 1985) for different densities are also shown.

the fine-structure lines probably arise in PDRs located in the interface between the ionised gas discussed above and the cold molecular gas. Rodr´ıguez-Fern´andez et al. (2004) also showed that an important fraction (∼50%) of the [C II] 158 and [Si II] 35 μm can arise from the diffuse ionised gas instead of the PDR itself. Regarding the warm H2 , only 10–20% of the total column density of gas with temperature of 150 K can be accounted for in the widespread PDRs scenario. The most likely heating mechanism for the bulk of the warm neutral gas is the dissipation of magnetohydrodynamic turbulence or low velocity shocks. Spectroscopic imaging/mapping will be useful to investigate the interplay between the different ISM phases in the GC. This could be done with the IRS instrument onboard the Spitzer Space Telescope. In addition, SOFIA and Herschel will be able to map the ionised and the warm neutral gas with high spectral resolution, which is needed due to the crowded velocity fields in the GC region.

4. Supernova Remnants (SNRs) As the prototype of O-rich young SNRs, Cassiopeia A is well-studied by ISO (Lagage et al., 1996; Tuffs et al., 1997; Arendt et al., 1999; Douvion et al., 1999; Douvion et al., 2001b). Several fine-structure lines are observed at high velocity with [OIV] being the most prominent (e.g. Figure 7). However, no iron emission is observed (Arendt et al., 1999). These fine-structure lines are spatially correlated

HIGH EXCITATION ISM AND GAS

287

Figure 7. Contour maps of the [Ne II] 12.8 μm line emission (dotted contours) and of the 9.5 μm silicate dust emission (full contours) plotted onto an optical image of the Cassiopeia-A supernova remnant. Examples of the variety seen in the MIR spectra are shown on the right (Taken from Douvion et al., 1999).

with the Fast Moving Knots (FMKs) known to be made of nuclear burning products from the progenitor star. While fainter diffuse dust emission is present, the brightest continuum emission is spatially correlated with the fine-structure lines indicating that dust has freshly condensed in the FMKs. The high dust temperature derived is consistent with emission from collisionally heated dust in the post-shock region of FMKs. A new 22 μm feature is found and identified with Mg protosilicates (Arendt et al., 1999). However, Mg protosilicates give rise to a 10 μm feature quite different from the observed one (Douvion et al., 2001b). These authors fitted a 6–30 μm spectrum of Cas A with SiO2 , MgSiO3 , Al2 O3 . In addition to dust continuum emission, silicate emission is present at several positions within Cas A (Figure 7). The presence of silicate and the presence of neon is anti correlated in many knots while argon and sulfur are present in most of the neon knots (Figure 7), suggesting a different degree of mixing occurred (Douvion et al., 1999). These results also indicate an incomplete condensation of silicate elements. If born out for other core collapse SNe, these SNe can no longer be considered the dominant source of silicates (Douvion et al., 1999). Similar conclusion was reached by Tuffs et al. (1997) based upon the derived dust mass being less than a few hundredths of a solar mass. In contrast to Cas A, the dust emission in the Kepler and Tycho SNRs is spatially correlated with Hα emission and therefore attributed to shocked circumstellar (CS)

288

E. PEETERS ET AL.

material and CS or interstellar (IS) material respectively and not to SN condensates. Instead of dust signatures, the Crab SNR is dominated by synchrotron radiation with a spectral index of −0.3 to −0.65. Fine-structure lines of [Ne II] and [Ar II] are observed for the Kepler SNR and of [Ne II], [Ne III], [Ne V], [Ar II], [Ar III], [S IV], [Ni II] for the Crab SNR. The [Ne III] emission in the Crab SNR shows the same filamentary structure as seen in optical lines, known to be composed predominately of SN ejecta (Douvion et al., 2001a). The IR spectrum of RCW103, a young and fast SNR, is dominated by prominent lines from low-excitation species allowing estimates of electron density, temperature and abundances. The ionic lines are consistent with post-shock emission. The absence of significant emission from the pre-shock region suggests that shock models may overestimate the importance of the precursor. All – but one – abundances are solar and hence the emitting gas is ISM material in which the dust grains have been destroyed by the shock front (Oliva et al., 1999b). The interaction of SNRs with molecular clouds has been extensively studied with ISO. The IR spectra of the SNRs 3C 391, W44 and W28 exhibit fine-structure line emission, dust emission along the line of sight and shocked dust emission shortwards of 100 μm (Reach and Rho, 1996, 2000; Reach et al., 2002). In addition, shock-excited FIR emission of H2 O, CO and OH emission is detected in 3C 391 (Reach and Rho, 1998). These IR observations provide evidence for multiple pre- and post-shock conditions which are found to be spatially separated. Indeed, ISOCAM-CVF observations clearly resolves the ionic emission from the molecular emission tracing moderate-density to high density regions in 3C 391. Abundance analysis reveal partial dust destruction consistent with theoretical models of grain destruction. Similar, spectral variations are observed spatially within IC443 (Cesarsky et al., 1999; Oliva et al., 1999a; Rho et al., 2001). The northeasthern filamentary structure is dominated by prominent [Ne II], [Fe II], [S III], [O IV] and [O I] line emission while the southern clumpy region is dominated by H2 emission indicating different types of shock occurring in different conditions. Both regions can alone account for virtually all the observed IRAS flux in the 12 and 25 μm bands – likely due to the limited ISO field-of-view and limited IRAS spatial resolution. Despite this, the unusually blue IRAS colors of IC443 reflect line (molecular and/or atomic) contamination instead of a large population of small grains. The late emission of SN1987A probes directly the elemental abundances deep in the stellar ejecta and as such constrains models of SN explosion and the explosive nucleosynthesis. Particularly important for this are the 56 Ni, 57 Ni and 44 Ti masses. The former two are known, the latter is more uncertain and can be derived from the [Fe I]24.05 μm and [Fe II] 25.99 μm fine-structure lines. These lines are however not detected in the ISO spectra of SN1987A indicating – based upon time-dependent models – an upper limit of 1.1 × 10−4 M (Lundqvist et al., 1999, 2001). The central region around the SN position is resolved in the MIR and consistent with collisionally heated grains in the shocked CS gas around SN1987A (Fischera et al.,

HIGH EXCITATION ISM AND GAS

289

2002a). The MIR spectrum can be modeled by a dust mixture of silicate–iron or silicate–graphite or pure graphite grains and indicates a low dust-to-gas ratio (Fischera et al., 2002a). Hence, observations of ionic lines allow to diagnose the physical conditions of the medium in which the SNR blast wave is propagating, the shock kinematics, the elemental abundances and hence also dust destruction. ISO provided evidence that dust can form in supernovae and dust can be (partially) destroyed in supernovae.

Acknowledgements EP acknowledges the support of the National Research Council. NJR-F acknowledges support by a Marie Curie Fellowship of the European Community under contract number HPMF-CT-2002-01677.

References Aannestad, P. A. and Emery, R. J.: 2001, Astron. Astrophys. 376, 1040. Aannestad, P. A. and Emery, R. J.: 2003, Astron. Astrophys. 406, 155. Afflerbach, A., Churchwell, E., and Werner, M. W.: 1997, Astrophys. J. 478, 190. Arendt, R. G., Dwek, E., and Moseley, S. H.: 1999, Astrophys. J. 521, 234. Blair, G. N., Evans, N. J. II, vanden Bout, P. A., and Peters, W. L., III: 1978, Astrophys. J. 219, 896. Bregman, J. and Temi, P.: 2005, Astrophys. J. 621, 831. Brooks, K. J., Cox, P., Schneider, N., Storey, J. W. V., Poglitsch, A., et al.: 2003, Astron. Astrophys. 412, 751. Burton, M. G., Hollenbach, D. J., and Tielens, A. G. G. M.: 1990, Astrophys. J. 365, 620. Ceccarelli, C., Caux, E., Tielens, A. G. G. M., Kemper, F., Waters, L. B. F. M., and Phillips, T.: 2002, Astron. Astrophys. 395, L29. Cesarsky, D., Lequex, J., Abergel, A., Perault, M., Palazzi, E., et al.: 1996, Astron. Astrophys. 315, L309. Cesarsky, D., Cox, P., Pineau des Forets, G., van Dishoeck, E. F., Boulanger, F., and Wright, C. M.: 1999, Astron. Astrophys. 348, 945. Cesarsky, D., Jones, A. P., Lequex, J., and Verstraete, L.: 2000, Astron. Astrophys. 358, 708. Chan, K.-W., and Onaka, T.: 2000, Astrophys. J. 533, L33. Cr´ete, E., Giard, M., Joblin, C., Vauglin, I., L´eger, A., and Rouan, D.: 1999, Astron. Astrophys. 352, 277. Darbon, S., Zavagno, A., Perrin, J.-M., Savine, C., Ducci, V., and Sivan, J.-P.: 2000, Astron. Astrophys. 364, 723. Draine, B. T. and Bertoldi, F.: 1999, The Universe as Seen by ISO, ESA-SP 427, p. 553. Douvion, T., Lagage, P. O., and Cesarsky, C. J.: 1999, Astron. Astrophys. 352, L111. Douvion, T., Lagage, P. O., Cesarsky, C. J., and Dwek, E.: 2001a, Astron. Astrophys. 373, 281. Douvion, T., Lagage, P. O., and Pantin, E.: 2001b, Astron. Astrophys. 369, 589. Evans, Neal, J., II, Kutner, M. L., and Mundy, Lee, G.: 1987, Astrophys. J. 323, 145. Fischera, Jg., Tuffs, R. J., and Volk, H. J.: 2002a, Astron. Astrophys. 386, 517. Fischera, J.g., Tuffs, R. J., and Volk, H. J.: 2002b, Astron. Astrophys. 395, 189.

290

E. PEETERS ET AL.

Fuente, A., Mart´ın-Pintado, J., Rodr´ıguez-Fern´andez, N. J., Rodr´ıguez-Franco, A., de Vicente, P., and Kunze, D.: 1999, Astron. Astrophys. 518, L45–L48. Giannini, T., Nisini, B., Lorenzetti, D., Di Giorgio, A. M., Spinoglio, L., et al.: 2000, Astron. Astrophys. 358, 310. Giveon, U., Sternberg, A., Lutz, D., Feuchtgruber, H., and Pauldrach, A. W. A.: 2002, Astrophysical Journal 566, 880. Giveon, U., Morisset, C., and Sternberg, A.: 2002, Astron. Astrophys. 392, 501. Goicoechea, J. R., Rodr´ıguez-Fern´andez, N. J., and Cernicharo, J.: 2004, Astrophys. J. 600, 1214. Gruenwald, R. B. and Viegas, S. M.: 1992, Astrophys. J., Suppl. Ser. 78, 153–178. Hollenbach, D. J. and Tielens, A. G. G. M.: 1999, Rev. Mod. Phys. 71, 173. Hony, S., Van Kerckhoven, C., Peeters, E., Tielens, A. G. G. M., Hudgins, D. M., and Allamandola, L. J.: 2001, Astron. Astrophys. 370, 1030. Hudgins, D. M. and Allamandola, L. J.: 2004, Astrophys. of Dust/Astron. Soc. Pacif. Conference Series, vol. 309, 665. Joblin, C., Abergel, A., Bregman, J., d’Hendecourt, L., Heras, A. M., et al.: 2000, ISO Beyond the Peaks, ESA-SP 466. Jones, A. P., Frey, V., Verstraete, L., Cox, P., and Demyk, K.: 1999, The universe as seen by ISO, ESA-SP 427, p 679. Kaufman, M. J., Wolfire, M. G., Hollenbach, D. J., and Luhman, M. L.: 1999, Astrophys. J. 527, 795. Kemper, F., Molster, F. J., J¨ager, C., and Waters, L. B. F. M.: 2002, Astron. Astrophys. 394, 679. Kemper, F., Vriend, W. J., and Tielens, A. G. G. M.: 2004, Astron. Astrophys. 609, 826. Kerr, T. H., Hurst, M. E., Miles, J. R., and Sarre, P. R.: 1999, Mon. Not. R. Astron. Soc. 303, 446. Klein, R., Henning, T., and Cesarsky, D.: 1999, Astron. Astrophys. 343, L53. Li, A. and Draine, B. T.: 2001, Astrophys. J. 550, L213. Lagage, P. O., Claret, A., Ballet, J., Boulanger, F., Cesarsky, C. J., et al.: 1996, Astron. Astrophys. 315, L274. Lis, D. C., Serabyn, E., Zylka, R., and Li, Y.: 2001, Astrophys. J. 550, 761. Lundqvist, P., Sollerman, J., Kozma, C., Larsson, B., Spyromilio, J., et al.: 1999, Astron. Astrophys. 347, 500. Lundqvist, P., Kozma, C., Sollerman, J., and Fransson, C.: 2001, Astron. Astrophys. 374, 629. Lutz, D., Feuchtgruber, H., Genzel, R., Kunze, D., Rigopoulou, D., et al.: 1996, Astron. Astrophys. 315, L269. Mart´ın-Hern´andez, N. L., van der Hulst, J. M., and Tielens, A. G. G. M.: 2003, Astron. Astrophys. 407, 957. Mart´ın-Hern´andez, N. L., Vermeij, R., Tielens, A. G. G. M., van der Hulst, J. M., and Peeters, E.: 2002, Astron. Astrophys. 389, 286. Mart´ın-Hern´andez, N. L., Peeters, E., Morisset, C., Tielens, A. G. G. M., Cox, P., et al.: 2002, Astron. Astrophys. 381, 606. Mathis, J. S. and Rosa, M. R.: 1991, Astron. Astrophys. 245, 625. Melnick, G., Gull, G. E., and Harwit, M.: 1979, Astrophys. J. 227, L29. Mezger, P. G., Duschl, W. J., and Zylka, R.: 1996, Astron. Astrophys. Rev. 7, 289. Mizutani, M., Onaka, T., and Shibai, H.: 2002, Astron. Astrophys. 382, 610. Miyata, T., Kataza, H., Okamoto, Y. K., Onaka, T., Sako, S., et al.: 2004, Astron. Astrophys. 415, 179. Mokiem, M. R., Mart´ın-Hern´andez, N. L., Lenorzer, A., de Koter, A., and Tielens, A. G. G. M.: 2004, Astron. Astrophys. 419, 319. Morisset, C., Schaerer, D., Mart´ın-Hern´andez, N. L., Peeters, E., Damour, F., et al.: 2002, Astron. Astrophys. 386, 558. Morisset, C.: 2004, Astrophys. J. 601, 858. Morisset, C., Schaerer, D., Bouret, J.-C., and Martins, F.: 2004, Astrophys. J. 415, 577.

HIGH EXCITATION ISM AND GAS

291

Moutou, C., Verstraete, L., L’eger, A., Sellgren, K., and Schmidt, W.: 2000, Astron. Astrophys. 354, L17. Oliva, E., Moorwood, A. F. M., Drapatz, S., Lutz, D., and Sturm, E.: 1999b, Astron. Astrophys. 343, 943. Oliva, E., Lutz, D., Drapatz, S., and Moorwood, A. F. M.: 1999a, Astron. Astrophys. 341, L75. Okada, Y., Onaka, T., Shibai, H., and Doi, Y.: 2003, Astron. Astrophys. 412, 199. Onaka, T.: 2000, Adv. Space Res. 25, 2167. Onaka, T. and Okada, Y.: 2003, Astrophys. J. 585, 872. Pauldrach, A. W. A., Hoffmann, T. L., and Lennon, M.: 2001, Astron. Astrophys. 375, 161. Pech, C., Joblin, C., and Boissel, P.: 2002, Astron. Astrophys. 388, 639. Peeters, E.: 2002, PhD thesis, University of Groningen, The Netherlands. Peeters, E., Allamandola, L. J., Bauschlicher, C. W., Hudgins, D. M., Sandford, S. A., and Tielens, A. G. G. M.: 2004a, Astrophys. J. 604, 252. Peeters, E., Allamandola, L. J., Hudgins, D. M., Hony, S., and Tielens, A. G. G. M.: 2004b, Astrophys. of Dust/Astron. Soc. Pacif. Conference Series, vol. 309, 141. Peeters, E., Hony, S., Van Kerckhoven, C., Tielens, A. G. G. M., Hudgins, D. M., Allamandola, L. J., and Bauschlicher, C. W.: 2002a, Astron. Astrophys. 390, 1089. Peeters, E., Spoon, H. W. W., and Tielens, A. G. G. M.: 2004c, Astrophys. J., 613. Peeters, E., Tielens, A. G. G. M., Boogert, A. C. A., Hayward, T. L., and Allamandola, L. J.: 2005, Astrophys. J. 620, 774. Peeters, E., Mart´ın-Hern´andez, N. L., Damour, F., Cox, P., Roelfsema, P. R., et al.: 2002b, Astron. Astrophys. 381, 571. Peimbert, M. and Torres-Peimbert, S.: 1977, Mon. Not. R. Ast. Soc. 179, 217–234. Pilbratt, G. L., Altieri, B., Blommaert, J. A. D. L., Fridlund, C. V., Tauber, J. A., and Kessler, M. F.: 1998, Astron. Astrophys. 333, L9. Reach, W. T. and Tho, J.: 1996, Astron. Astrophys. 315, L277. Reach, W. T. and Tho, J.: 1998, Astrophys. J. 507, L93. Reach, W. T., Boulanger, F., Contursi, A., and Lequeux, J.: 2000a, Astron. Astrophys. 361, 895. Reach, W. T. and Tho, J.: 2000b, Astrophys. J. 544, 843. Reach, W. T., Tho, J., Jarrett, T. H., and Lagage, P.-O.: 2002, Astrophys. J. 564, 302. Rho, J., Jarrett, T. H., Cutri, R. M., and Reach, W. T.: 2001, Astrophys. J. 547, 885. Rodr´ıguez-Fern´andez, N. J., Mart´ın-Pintado, J., and de Vicente, P.: 2001a, Astron.y & Astrophysics 377, 631. Rodr´ıguez-Fern´andez, N. J., Mart´ın-Pintado, J., Fuente, A., de Vicente, P., Wilson, T. L., and H¨uttemeister: 2001b, Astron. Astrophys. 377, 631. Rodr´ıguez-Fern´andez, N. J. and Mart´ın-Pintado: 2005, Astron. Astrophys. 429, 923. Rodr´ıguez-Fern´andez, N. J., Mart´ın-Pintado, J., Fuente, A., and Wilson, T. L.: 2004, Astron. Astrophys. 427, 217. Roelfsema, P. R., Cox, P., Tielens, A. G. G. M., Allamandola, L. J., Baluteau, J.-P., et al.: 1996, Astron. Astrophys. 315, L289. Rubin, R. H., Simpson, J. P., Erickson, E. F., and Haas, M. R.: 1988, Astrophys. J. 327, 377. Rudolph, A. L., Simpson, J. P., Haas, M. R., Erickson, E. F., and Fich, M.: 1997, Astrophys. J. 489, 94. Russell, R. W., Melnick, G., Smyers, S. D., Kurtz, N. T., Gosnell, T. R., et al.: 1981, Astrophys. J. 250, L35. Russell, R. W., Melnick, G., Gull, G. E., and Harwit, M.: 1980. Astrophys. J. 240, L99. Schneider, N., Simon, R., Kramer, C., Kraemer, K., Stutzki, J., and Mookerjea, B.: 2003, Astron. Astrophys. 406, 915. Simpson, J. P. and Rubin, R. H.: 1990, Astrophys. J. 354, 165.

292

E. PEETERS ET AL.

Simpson, J. P., Colgan, S. W. J., Rubin, R. H., Erickson, E. F., and Haas, M. R.: 1995, Astrophys. J. 444, 721. Simpson, J. P., Witteborn, F. C., Price, S. D., and Cohen, M.: 1998, Astrophys. J. 508, 268. Sollerman, J.: 2002, New Astron. Rev. 46, 493. Song, I.-O., Kerr, T. H., McCombie, J. and Sarre, P. J.: 2004, Mon. Not. R. Astron. Soc. 346, L1. Stasi´nska, G.: 1990, Astron. Astrophys. Suppl. Ser. 83, 501. Storey, J. W. V., Watson, D. M., and Townes, C. H.: 1979, Astrophys. J. 233, 109. Tielens, A. G. M. M. and Hollenbach, D.: 1985, Astrophys. J. 291, 722. Timmermann, R., Bertoldi, F., Wright, C. M., Drapatz, S., Draine, B. T., et al.: 1996, Astron. Astrophys. 315, L281. Thornley, M. D., Schreiber, N. M. F., Lutz, D., Genzel, R., Spoon, H. W. W., et al.: 2000, Astrophysical Journal 539, 641–657. Torres-Peimbert, S. and Peimbert, M.: 1977, Rev. Mexicana de Astron. Astrof´ıs. 2, 181–207. Tuffs, R. J., Drury, L. O. ’C., Fischera, J., Heinrichsen, I., Rasmussens, I. et al.: 1997, First ISO Workshop on Analytical Spectroscopy, ESA-SP419, 177. Urquhart, J. S., White, G. J., Pilbratt, G. L., and Fridlund, C. V. M.: 2003, Astron. Astrophys. 409, 193. van den Ancker, M. E., Tielens, A. G. G. M., and Wesselius, P. R.: 2000, Astron. Astrophys. 358, 1035. van Diedenhoven, B., Peeters, E., Van Kerckhoven, C., Hony, S., Hudgins, D. M., et al.: 2004, Astrophysical Journal 611, 928. Van Kerckhoven, C., Hony, S., Peeters, E., Tielens, A. G. G. M., and Allamandola, L. J., et al.: 2000, Astron. Astrophys. 357, 1013. Van Kerckhoven, C.: 2002, PhD thesis, Catholic University of Leuven, Belgium. Vastel, C., Spaans, M., Ceccarelli, C., Tielens, A. G. G. M., and Caux, E.: 2001, Astron. Astrophys. 376, 1064. Verma, R. P., Ghosh, S. K., Mookerjea, B., and Rengarajan, T. N.: 2003, Astron. Astrophys. 398, 589. Vermeij, R., Damour, F., van der Hulst, J. M., and Baluteau, J.-P.: 2002a, Astron. Astrophys. 390, 649. Vermeij, R. and van der Hulst, J. M.: 2002b, Astron. Astrophys. 391, 1081. Vermeij, R., Peeters, E., Tielens, A. G. G. M., and van der Hulst, J. M.: 2002c, Astron. Astrophys. 382, 1042. Verstraete, L., Puget, F. L., Falgarone, E., Drapatz, S., Wright, C. M., and Timmermann, R.: 1996, Astron. Astrophys. 315, L337. Verstraete, L., Pech, C., Moutou, C., Sellgren, K., Wright, C. M., et al.: 2001, Astron. Astrophys. 372, 981. Wada, S., Onaka, T., Yamamura, I., Murata, Y., and Tokunaga, A. T.: 2003, Astron. Astrophys. 407, 551. Willner, S. P. and Nelson-Patel, K.: 2002, Astrophys. J. 568, 679. Wright, C. W., Drapatz, S., Timmermann, R., van der Werf, P. P., Katterloher, R., and de Graauw, Th.: 1996, Astron. Astrophys. 315, L301. Wright, C. W., van Dishoeck, E. F., Cox, P., Sidher, S. D., and Kessler, M. F.: 1999, Astrophys. J. 515, L29. Zavagno, A. and Ducci, V.: 2001, Astron. Astrophys. 371, 312.

THE ICE SURVEY OPPORTUNITY OF ISO EMMANUEL DARTOIS Institut d’Astrophysique Spatiale, Astrochimie Exp´erimentale, UMR-8617 Universit´e Paris-Sud, bˆatiment 121, F-91405 Orsay, France (E-mail: [email protected]) (Received 30 July 2004; Accepted in final form 27 October 2004)

Abstract. The instruments on board the Infrared Space Observatory have for the first time allowed a complete low (PHOT, CVF) to medium resolution (SWS) spectroscopic harvest, from 2.5 to 45 μm, of interstellar dust. Amongst the detected solids present in starless molecular clouds surrounding recently born stellar and still embedded objects or products of the chemistry in some mass loss envelopes, the so-called “ice mantles” are of specific interest. They represent an interface between the very refractory carbonaceous and silicates materials that built the first grains with the rich chemistry taking place in the gas phase. Molecules condense, react on ices, are subjected to UV and cosmic ray irradiation at low temperatures, participating efficiently to the evolution toward more complex molecules, being in constant interaction in an ice layer. They also play an important role in the radiative transfer of molecular clouds and strongly affect the gas phase chemistry. ISO results shed light on many other species than H2 O ice. The detection of these van der Waal’s solids is mainly performed in absorption. Each ice feature observed by ISO spectrometer is an important species, with abundance in the 10−4 –10−7 range with respect to H2 . Such high abundances represent a substantial reservoir of matter that, once released later on, replenishes the gas phase and feeds the ladder of molecular complexity. Medium resolution spectroscopy also offers the opportunity to look at individual line profiles of the ice features, and therefore to progressively reveal the interactions taking place in the mantles. This article will give a view on selected results to avoid to overlap with the numerous reviews the reader is invited to consult (e.g. van Dishoeck, in press; Gibb et al., 2004.). Keywords: ices, dust, infrared absorption

1. Ices Everywhere Interstellar ices have been observed with ISO-SWS in many different lines of sight among which the most common are:

r OH-IR (and by extension, evolved stars circumstellar shells) which are oxygen rich post main sequence stars loosing mass at a high rate. The dust and molecules formed in this mass loss ejecta feed a circumstellar shell, opaque to visible light and therefore re-emitting in the infrared. In these lines of sight, H2 O is likely to be the only ice present on dust grains, generally in crystalline Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands, and the United Kingdom), and with the participation of ISAS and NASA. Space Science Reviews (2005) 119: 293–310 DOI: 10.1007/s11214-005-8059-9

 C

Springer 2005

294

EMMANUEL DARTOIS

form (see Figure 1) as it was formed in the gas phase and then condensed at high temperature. r External galaxies (e.g. M82, Figure 2). In such lines of sight, the ocurrence of ices is strongly dependent on geometrical effects. Not surprisingly, galaxies do contain large amounts of ices inside molecular clouds.

Figure 1. Short Wavelength Spectrometer (AOT01) observations of two OH-IR circumstellar shells, displaying silicates features and water ice signatures. The insert contains the continuum divided spectra in the 3 μm region of the spectrum, in order to show the OH stretching mode of crystalline water ice. Extracted from the ISO database (http://isowww.estec.esa.nl/).

Figure 2. Short Wavelength Spectrometer (AOT01) observations of the Andromeda galaxy. The galactic continuum is absorbed by water ice present in the dense clouds of the galaxy. Extracted from the ISO database.

ICE SURVEY OPPORTUNITY OF ISO

295

Figure 3. Short Wavelength Spectrometer (AOT06) observation of the Field star Elias 3–16, located behind the Taurus dark cloud. The infrared continuum of the star is absorbed selectively by the ice mantles located in the cloud. Extracted from the ISO database.

r Field Stars are located behind molecular clouds (e.g. Elias 3–16, Figure 3) and thus allow one to use their infrared pencil to probe the foreground molecular cloud ice composition. They are crucial to understand ice evolution and chemistry during the earliest stage of starless clouds formation. The drawbacks are that they are statistically scarse and faint in the infrared. They are typically main sequence stars, often possessing intrinsic photospheric absorptions one has to take into account to interpret the data. r Embedded protostellar objects (e.g. Figure 4) which constitute by far the more abundant and richest database of ice species, with an observational bias toward the brightest high mass sources. The drawback of these sources lines of sight is that the central object has generally started the interaction with the ice containing parental cloud, stimulating new processes but also complicating the analysis. Such an extended panel of common line of sight, to which must be added the ones that to date escape detections such as optically thick disks, suggests ices are very common and abundant constituents in the life-cycle of dust in galaxies. 2. Inventory of the Detected Ice Features ISO has brought the number of ice features to more than 20 for clear detections, to which must be added a few ones to confirm. These features, the corresponding molecule, vibrational mode and integrated band strength are summarized in Table I. The large wavelength coverage of the ISO spectrometers allowed to observe in the same spectra several modes of the same ice constituent.

296

EMMANUEL DARTOIS

Figure 4. Short Wavelength Spectrometer (AOT06) observations of the rich ice spectrum of embedded intermediate to high mass protostars. From top to bottom: Elias 2–29, RCRA, GL2136, NGC7538 IRS9. Spectra have been shifted in flux by arbitrary constants for better clarity. The more common identified features are labeled below the spectra. Extracted from the ISO database

3. Molecule by Molecule 3.1. H2 O: THE LORD OF THE ICES One of the first but unsuccessful attempt to detect water ice was performed by Knacke et al. (1969) in the diffuse medium. Water ice was observed later on through

297

ICE SURVEY OPPORTUNITY OF ISO

TABLE I Line list of ices absorptions observed by ISO. Wavelength (μm)

Molecule

2.70 2.78 2.97 3.05 3.47 3.53 3.84–3.95 4.27 4.38 4.55 4.62 4.67 4.78 4.90 5.81 5.83 5.85 6.02 6.33 6.85 7.25 7.25 7.41 7.41 7.7 8.85 9.35 9.75 11–14 15.2

CO2 CO2 NH3 H2 O NH3 · · ·H2 O CH3 OH CH3 OH CO2 13 CO 2 H2 O OCN− CO 13 CO OCS H2 CO HCOOH CH3 CHO H2 O HCOO− NH+ 4 HCOOH HCOO− HCOO− CH3 CHO CH4 CH3 OH NH3 CH3 OH H2 O CO2

Vibrational mode

A (10−18 cm/mol)

Reference

Combination (ν1 + ν3) Combination (2ν2 + ν3) N H stretch (ν1) O H stretch (ν1 − ν3) Hydrate O H stretch CH3 s-stretch (ν3) Combination Antisym. stretch (ν3) Antisym. stretch (ν3) Libration overt. (3νL ) C N stretch (ν3) CO stretch CO stretch CO stretch (ν1) CO stretch (ν2) CO stretch (ν3) CO stretch O H bend (ν2) CO stretch NH def (ν4) C H bend. (ν4) CO stretch

– – 10–11 200 ∼200 5.3–7.6 2.6–3.2 76 78 ∼10 40–80/(130 ?) 11 13 150–170 9.6 67 13 8.4–12 100 40–44 2.6 8 17 1.5 6.4–7.3 1.3–1.8 12–20 18 28–31 11

(1,2) (1,3) (4) (1,5) (1,5) (3) (3) (3) (1,6,7,8)/(9) (3) (3) (5) (10) (11) (11) (1,3) (11) (14) (11) (11) (11) (11) (1,12) (1,5) (2,13) (1,5) (1,5) (3)

CO stretch CH def. (ν4) CH3 rock (ν7) Umbrella (ν2) CO stretch (ν8) Libration(νL ) CO2 bending (ν2)

Note: The italicized detected molecules are either tentative, need confirmations or their abundance are still subject to strong debate. References: (1) Dhendecourt and Allamandola (1986); (2) Sandford and Allamandola (1993); (3) Gerakines et al. (1995); (4) Dartois and d’Hendecourt (2001); (5) Hudgins et al. (1993); (6) Grim and Greenberg (1987); (7) Schutte and Greenberg (1997); (8) Demyk et al. (1998); (9) van Broekhuizen et al. (2004) a value 3 σ above the other determinations; (10) Schutte et al. ((1996a)); (11) Schutte et al. (1999); (12) Boogert et al. (1997); (13) Kerkhof et al. (1999); (14) Schutte and Khanna (2003).

298

EMMANUEL DARTOIS

its OH stretching mode at 3 μm, from the ground by Gillett and Forrest (1973). Since then, many line of sight were surveyed from the ground in this mode (e.g. Merrill et al., 1976; L´eger et al., 1979; Willner et al., 1982; Whittet et al., 1983; Smith, Sellgren et al., 1989; Eiroa and Hodapp, 1989; Tanaka et al., 1990; Sato et al., 1990; Chen and Graham, 1993; Smith et al., 1993). After the discovery of the vibrational fundamental modes, phonon modes around 44 and 66 μm were found in emission by Omont et al. (1990) and followed by laboratory analysis (Moore and Hudson, 1992; Hudgins et al., 1993; Smith et al., 1994; Maldoni et al., 1999). So many others authors contributed to this search that a full list is out of scope. In this context, the insight in a better understanding of water ice performed by ISO has first been to substantially extend the number of astrophysical lines of sight where water ice is observed and, more interestingly, to offer a complete and simultaneous coverage of the six principal modes/combinations occurring in the infrared: the OH stretch (∼3.05 μm), the libration overtone (∼4.5 μm), the OH bend (∼6.0 μm), the libration (∼11–13 μm), the longitudinal and transverse optical mode at ∼44 μm and ∼66 μm, respectively. The two latter modes are detected either in absorption (Dartois et al., 1998) or emission (Malfait et al., 1999; Molinari et al., 1999; Hoogzaad et al., 2002; Molster et al., 2002; Maldoni et al., 2003) betraying the radiative transfer effects at long wavelength. For the water ice seen in emission toward some lines of sight, care must however be taken to analyse the spectra, as silicates far infrared features (especially enstatite) contribute to the lines. In addition to their detections, ISO has brought truly new spectroscopic constraints on the water ice modes. As an example, the OH bending mode profiles reveal the contribution of other features, unaccessible before (Keane et al., 2001), and whose identification will be discussed later on. Water ice is the dominant solid state frozen species and consequently the abundances of other ices are referred to this one on a relative scale (see Figure 6). On an absolute scale, in clouds with substantial visual extinction, water ice is the second most abundant species after H2 ([H2 O]/[H2 ] ≈ 10−4 –10−5 ), comparable to or even more abundant than gas phase CO. 3.2. UNRAVELING THE UBIQUITOUS CO2 Before the launch of ISO, the carbon dioxide existence in ice mantles was founded on very few astrophysical data. Its inference was indirect and based on the possible influence of CO2 interacting with other molecules on their line profiles (Sandford et al., 1988; Kerr et al., 1991; Tielens et al., 1991). Indeed, besides an infrared oscillator strength amongst the strongest of ISM solid molecules, the atmospheric carbon dioxide prevents any ground based or airborne observations detection of its space parent. The very first direct detection was performed in 1989, on unresolved CO2 bending mode profiles, using IRAS Low Resolution Spectrometer spectra (d’Hendecourt and Jourdain de Muizon, 1989). The ubiquitous nature of carbon dioxide was revealed and firmly assessed with ISO spectrometers (Guertler et al.,

ICE SURVEY OPPORTUNITY OF ISO

299

1996; Gerakines et al., 1999; Alexander et al., 2003; de Graauw et al., 1996). CO2 then appears as the most abundant molecule after H2 O in ice mantles, with a mean CO2 /H2 O ratio ranging generally from 10 to 50% in protostellar environments, with a mean around 15–20%. The 13 CO2 stretching mode was detected, and its line profile analysed by Boogert et al. (2000). This is a valuable information, because being about 60 times less abundant than the main molecule, this isotopomer is experiencing and therefore tracing the ice matrix environment field. It allows to a certain point to distinguish between different thermal processing having affected in the past the ice mantles. Some combination modes (symmetric+antisymmetric stretching and twice the bending+symmetric stretching modes) have also been detected in S140 IRS1 (Keane et al., 2001), being the first detection of solid state identified combinations in space ice mantles. The position and profiles of these combinations implies that the carbon dioxide is mixed in roughly equal proportions with methanol and water ice. The ability to observe combination modes, with low integrated absorption cross sections further demonstrate these ices are abundant components of molecular clouds. Another results of ISO medium resolution spectra is the first detection of interstellar molecular complexes implying the interaction of a carbon atom of the carbon dioxide with the lone electron pairs of the methanol oxygen (Ehrenfreund et al., 1999; Dartois et al., 1999; Klotz et al., 2004). This interaction deforms the CO2 molecule in such a way that the degenerescence of the bending mode is broken, giving rise to two individual absorption components around 15.2 μm (see Figure 5, right panel), whereas the stretching mode is only slightly affected. The combination of this complexation with the thermal evolution of ice mantles gives rise to characteristic substructures, possessing from one to three subpeaks (Figure 5, left panel, Dartois, et al., 1999) in the bending mode regions of carbon dioxide infrared

Figure 5. Left: SWS01 absorption spectrum of RAFGL7009 in the CO2 bending mode region displaying a triple substructure, compared with laboratory ice spectrum + gas phase model. Right: This structure partly arises due to the break of degeneracy of the in-plane and out-of-plane CO2 bending mode (see text for details).

300

EMMANUEL DARTOIS

spectra. Such a spectroscopic constraint is of major importance for the understanding of the structure in which are organised here the carbon dioxide and methanol ices. These structures are not fortuitous, requires a specific stœchiometry, and the premice of better understanding of their formation history. 3.3. CO Solid CO is a molecule observed almost routinely from ground based telescopes and recently its principal isotopomer 13 CO was reported (Boogert et al., 2002). Groundbased observations allow to study the carbon monoxide line profiles (Pontoppidan et al., 2003) with unprecedented spectral resolution. The reader is referred to the huge literature on the subject, notably on the various profiles induced by the polar or apolar environment of this molecule. One of the results of ISO was to allow a comparison with the same infrared pencil beam of gas-to-solid state CO ratio (Dartois et al., 1998; van Dishoeck et al., 1996), showing that even if it is an abundant component of ice mantles (typically 3–20%), it resides principally in the gas phase in the general case. 3.4. OCS

AND

S ULFUR-CONTAINING MOLECULES

Sulfur is an abundant element of interstellar medium and one expects therefore to observe it in ices. Geballe et al. (1985) detect toward W33 A two absorptions at 3.9 and 4.9 μm, respectively attributed by these authors to H2 S and another sulfurcontaining molecule produced in laboratory experiments. The 3.9 μm feature will be attributed later on to solid methanol combination modes, the only really containing sulfur molecule giving rise to the feature at 4.9 μm. Smith (1991) discuss the search for H2 S in six objects and put an upper limit of about 1% on its presence. Palumbo et al. (1997) and Palumbo et al. (1995) observe the 4.9 μm feature, identify it with carbonyl sulfide and discuss its abundance toward various lines of sight. The line profile is best reproduced when OCS is mixed with methanol. In fact there is an overlap of the second overtone 2ν8 of methanol with this line. With ISO, OCS has been evaluated in many lines of sights (see references in Gibb et al., 2004), but most of the time without evaluating the methanol overtone contribution, especially towards RAFGL 7009 and W33 A. The CO stretch of OCS is one of the more intense infrared bands and therefore OCS is the molecule detected with the lowest abundance with respect to H2 O, of about 0.05–0.15%. This can be considered as the limit of detection of ices. 3.5. CH4 The ground based very difficult detections of methane were started by Lacy et al. (1991). With ISO, the methane molecule was detected in several sources via its

ICE SURVEY OPPORTUNITY OF ISO

301

deformation mode at 7.67 μm mode, with abundances up to 4% with respect to H2 O (d’Hendecourt et al., 1996), but more generally at the 1–2% level (Boogert et al., 1996). The gas-to-solid methane ratio in such lines of sight is less than unity, favouring a formation of this hydride by hydrogenation of atomic carbon directly on the surface of grains. The actual abundance of methane does not reflect its initial abundance in the solid phase as it is one of the easiest molecule to dissociate upon UV secondary photons photolysis or ion irradiation, typical of dark clouds. 3.6. H 2 CO Most of the molecules containing a carbonyl group strongly absorb in the 5.7– 6 μm region of the spectrum, the simplest of them being the formaldehyde (e.g. Schutte et al., 1993). The first attempt to detect it was made by Schutte et al., (1996a) in GL 2136 but the abundance controversed due to a mixing of formaldehyde modes with methanol ones, also present in large amount in the same line of sight. An estimate was also given for W33 A by Brooke et al. (1999). These “pre-ISO” estimates were based on C H stretching modes rather difficult to disentangle from the methanol CH stretches present in the spectra. Formaldehyde strongest modes fall at 5.8 and 6.68 μm, both inaccessible from the ground. ISO therefore allowed the search of these reliable modes for the first time, even if protostellar sources do possess strong absorptions in the same wavelength region due to the H2 O bending mode and the so-called 6.85 μm feature. Formaldehyde abundance in ISO spectra was estimated in five high mass protostellar envelopes (W33 A, AFGL7009, GL2136, GL989, NGC7538 IRS9) to range between 1 and 3% (see Keane et al., 2001 (Table III); Dartois et al., 1999; Gibb et al., 2000; and references therein). 3.7. CH 3 OH CH3 OH is mainly detected from ground-based observations of its ν3 mode around 3.53 μm . It was early detected through this mode (Baas et al., 1988; Grim et al., 1991; Allamandola et al., 1992) in massive protostars lines of sight. The presence of the methanol–carbon dioxide complex, discussed above in the CO2 section, raised a new interest for its search (e.g. Dartois et al., 1999). Only recently and from the ground, it was shown by Pontoppidan et al. (2003) that methanol is also present around some low mass protostars. Methanol is a key molecule in ices, as it sometimes represent up to several tenth of the water ice abundance. Its formation route is under a controversial debate. Some authors think it is formed directly from grain surface hydrogenation of CO, but to date laboratory experiments performed give contradictory answers to the hypothesis (Hiraoka et al., 2002; Watanabe et al., 2003).

302

EMMANUEL DARTOIS

Others think that it can be created by energetic processes (UV or cosmic rays) (Hudson and Moore, 1999; Moore and Hudson, 1998), eventually followed by selective desorption enhancement effects, methanol being a rather refractive ice. 3.8. NITROGEN B EARING S PECIES 3.8.1. N2 Nitrogen being an abundant cosmic element and N2 a stable molecule, its presence would be logical. In practice, this homonuclear molecule is, due to its symmetry, inactive in the infrared. However, when embedded in an ice matrix, the crystal field breaks this symmetry and an infrared transition is activated around 4.295 μm . This weakly intense transition, with an integrated absorption cross section of 10−6 – 10−4 the OH stretching one of water ice in the best activated mixtures (Sandford et al., 2001; Ehrenfreund and van Dishoeck, 1998 and references therein), is a challenge to detection limits. In addition, this feature falls in the red wing of the strong CO2 antistretch absorption, both in the solid and gas phase. To date, the only few upper limits derived for the presence of N2 when embedded into water ices, and using ISO SWS data (see above cited references), are not sufficiently low to provide strong constraints. Only very (too ?) specific ice mantles compositions (such as pure N2 :CO2 ice mixtures) could lead to push down this limit by enhancing the activation of the otherwise inactive N2 stretching mode, which remains to be proved in an ice dominated by water ice. 3.8.2. NH3 After N2 , the most abundant nitrogen bearing molecule is certainly the ammonia one. Knacke et al. (1982) were among the firsts to claim for an identification of NH3 in interstellar grains from an infrared absorption at 2.97 μm (NH stretching mode), a detection refuted later on Knacke and McCorkle (1987). Hagen et al. (1983) show that the H2 O stretching mode profile is best reproduced with an ice containing a significant fraction of ammonia. Since then, many authors entered into the still opened debate of ammonia ice abundance in grain mantles. Ammonia possess three main modes at 2.97 μm (NH stretch), 6.2 μm (NH bend) and 9. μm(umbrella mode, which shifts to longer wavelength if pure ammonia). Importantly, and often underestimated, when mixed with H2 O, the major ice mantle constituent, ammonia forms an hydrate that gives rise to a large and intense band around 3.47 μm (with an integrated absorption cross section similar to the water OH stretch). All modes, except for the hydrate one, fall in spectroscopic regions overlapping with the dominant water ice or silicate features, which renders the detection task difficult. In addition, toward lines of sight where methanol is present, the CH3 rock can contribute significantly to the ammonia umbrella mode. Detecting the relatively intense 9 μm umbrella mode in the wing of the strongly absorbed silicates features has always led to high overestimates of the true ammonia contents.

ICE SURVEY OPPORTUNITY OF ISO

303

A school case to support this view comes from the well-studied W33 A line of sight. The methanol contribution at 8.9–9 μm, inferred from other modes, should provide an optical depth around ∼0.1. To show consistency in interpreting spectra of this source, and before any ammonia search, this CH3 OH mode must be evidenced and subtracted to the data. Based on ISO PHOT40 and SWS spectra extracts around 9 μm, various authors derive quite different column densities of NH3 , namely 1.7 × 1018 cm−2 (Gibb et al., 2001), 1.2–2.2 × 1018 cm−2 (G¨urtler et al., 2002), representing a high percentage (11–20%) as compared to water ice. These numbers were in fact all contradicted by the recent ground based upper limit toward this source by Taban et al. (2003), which lies about two to three times lower. In the analysis presented in Dartois and d’Hendecourt (2001) and Gibb et al. (2001), on 20 different sources lines of sight (mainly massive YSOs), only in three of the Gibb’s article are claimed NH3 detections. For these three sources : (i) W33 A detection was later on rejected by Taban et al. (2003); (ii) NGC7538 IRS9 possess a considerable uncertainty on ammonia abundance given the scatter in the derived NH3 /H2 O ratios of 10% (Lacy et al., 1998), 15% (Gibb et al., 2001), 1011 L  )) were discovered that emit most of their bolometric luminosity in the infrared (Houck et al., 1984). AGN and massive bursts of star formation are the only known mechanisms that are capable of generating such luminosities. However, without spectroscopic information of the obscured components, classifying the underlying power source has proved to be difficult. Furthermore, quantitatively assessing the contribution to the total power from the two possible mechanisms is non-trivial. It is known that dust is formed during the late stages of stellar evolution and it is destroyed by intense radiation fields and shocks. Consequently, a detailed understanding of the physical mechanisms responsible for the observed infrared emission is an essential first step into revealing the nature of the underlying exciting source. As a result, since the most ‘active’ regions in our Universe appear to be enshrouded in large quantities of gas and dust, addressing the aforementioned points would have serious consequences in estimating star formation and black hole activity in the Universe. In a cosmological context, both the reported increase in the star formation density of the Universe for 0.1  z  2 (e.g. Madau et al., 1996; Barger et al., 2000) and the high frequency of starbursts hosted in galaxies with disturbed morphologies or in interacting/merging systems, imply that starbursts play an important role in galaxy formation and evolution. In addition, black hole activity could be buried in any luminous infrared system. Ultraluminous infrared galaxies (ULIGs, L IR > 1012 L  ) have been proposed to be an evolutionary stage in the life of a quasar and the local analogues of the z > 2 (sub-)mm population. Quantifying the amount of obscured activity at earlier times and its contribution to the infrared and X-ray backgrounds remain open issues requiring a sound understanding of the properties of local active galaxies. Following on from the coarse, photometric legacy of IRAS, ISO provided the first means to investigate the physical conditions of local active galaxies in significant detail at wavelengths where the obscured, and often most active, regions could be probed. Equipped with imaging, photometric and spectroscopic capabilities, ISO surveyed a wide range of emission properties from broad-band spectral energy distributions (SEDs), through imaging to detailed spectral analysis. We discuss the infrared emission properties of obscured active galaxies that ISO surveyed during its lifetime. In the context of this chapter, ‘active’ refers to both star formation and black hole activity and encompasses the range of non-normal galaxies: interacting/merging galaxies; starbursts; radio galaxies; AGN; quasars; low ionisation nebular emission regions (LINERs); and the full suite of LIGs–ULIGs and hyperluminous infrared galaxies (HyLIGs > 1013 L  ).

OBSCURED “ACTIVITY”: AGN, QUASARS, STARBURSTS AND ULIGs

357

2. Activity Manifest in the Infrared High energy UV (and visible) photons emanating from active regions can heat or excite environmental dust, located around active sites or distributed throughout the interstellar medium (ISM), resulting in re-radiation in the infrared. Grains ranging from the very small (radius a  10 nm) to the large (a ∼ 30 μm) contribute to emission over the 2.5–200 μm range. The type, size and distribution of dust grains together with the incident interstellar radiation field (ISRF) shape the observed spectral features and the overall form of the SED which arise from both thermal and non-thermal processes. At long infrared wavelengths (λ  25 μm), emission predominantly arises from grains that re-radiate in thermodynamic equilibrium. In young radio galaxies, thermal bremsstrahlung may also provide a contribution to the infrared continuum. Emission related to transient rather than steady-state heating of dust grain complexes can dominate the SEDs of some active galaxies giving rise to continua and features over 3–18 μm. Non-thermal components include the featureless far-infrared tail (increases with increasing wavelength) of the synchrotron radiation spectrum in strong radio galaxies. 2.1. INFRARED CONTINUA: T HERMAL

AND

NON -THERMAL

The mid-infrared spectra of galaxies display two continua: A flat continuum for λ  5 μm and, in some galaxies, an unrelated continuum for λ  11 μm (Helou et al., 2000). These continua are not produced by grains heated to thermal equilibrium by the radiation field but arise from the stochastic heating of small dust grains (amorphous silicates and graphites, a ∼ 0.01–0.1 μm) which are transiently heated by the absorption of a single photon (Sellgren, 1984; Beichman, 1987; Boulanger et al., 1998b; Draine and Li, 2001; Draine, 2003). A single photon with energy of only a few electron volts (λ  0.4 μm), is sufficient for the onset of infrared emission (Uchida et al., 1998). The steeply rising (near-thermal) continuum longward of 11 μm, which is strong in some active galaxies (but weak in quiescent galaxies), is thought to originate from the excess transient heating of very small fluctuating dust grains (VSG, a ≤ 10 nm). This component appears to be a characteristic of intensely star-forming regions (Desert et al., 1990; Verstraete et al., 1996; Cesarsky et al., 1996b; Laureijs et al., 1996). Under this transient heating regime, the mid-infrared continuum is directly proportional to the underlying radiation field over several orders of magnitude and varies with the conditions of the H II regions from which it originates (Boulanger et al., 1998a; Laurent et al., 2000; F¨orster Schreiber et al., 2004). For example, in the close vicinity of the dense starburst knots in the Antennae, this continuum is shifted to shorter mid-infrared wavelengths implying higher temperatures (Vigroux et al., 1996). The 3 μm < λ < 5 μm continuum is likely to arise from a featureless fluctuating component (Helou et al., 2000). Although often weak in active galaxies, the direct

358

A. VERMA ET AL.

stellar continuum can also contribute (λ < 5 μm) and should be considered in careful spectral decomposition analysis. The strength of this component increases with decreasing wavelength and thus the transition between direct stellar to interstellar dust emission occurs within the near- to low-mid-infrared (Boselli et al., 1997, 1998; Alonso-Herrero et al., 2001). The mid-infrared also includes the transition from the transient to the steady-state heating regime. The wavelength at which this transition occurs is determined by the incident radiation field above which grains of a certain size no longer suffer large temperature changes. Dust grains are heated by the general ISRF and attain a ‘steady state’ where emission and absorption reach an equilibrium. The grain temperature at which this equilibrium occurs characterises the emission giving rise to a thermal black-body continuum. Qualitatively, the mid- to far-infrared continua of active galaxies are often approximated by a series of superposed black-bodies, each originating from a body of dust heated to a characteristic temperature (Haas et al., 1998a; Ivison et al., 1998; Klaas et al., 2001; Bendo et al., 2003; Stevens et al., 2005). This approximation provides only an indication of typical temperatures and is not an accurate description of the complex heating of a variety of dust grains. More accurate descriptions are achieved by considering a multi-grain dust model and complex radiation fields using, for example, semi-empirical models (e.g. Dale et al., 2001a) or radiative transfer theory (e.g. Acosta-Pulido et al., 1996; Silva et al., 1998; Alexander et al., 1999a; Siebenmorgen et al., 1999b; Efstathiou et al., 2000; Verma et al., 2002; Farrah et al., 2003; Freudling et al., 2003; Gonz´alez-Alfonso et al., 2004; Siebenmorgen et al., 2004a). For sources that are spatially unresolved by the ISO instruments, SED decomposition (using one of the methods described earlier) is the only means to investigate the underlying physical processes. Some AGN-hosting galaxies display a relatively broad and flat (in ν Fν ) hot continuum that dominates their mid-infrared SEDs. This emerges from grains in the putative torus that are heated up to the grain sublimation temperature (Tsub ∼ 1500 K) by the strong radiation field of the accretion disk of the central AGN. The ‘warm’ IRAS galaxy criterion (S60 /S25 > 0.2) selects galaxies displaying this component which are mostly identified with AGN (de Grijp et al., 1985). The extended wavelength range of ISO over IRAS enabled (a) the precise determination of the turnover in the far-infrared SEDs of active galaxies and (b) the identification of two dust components that comprise the far-infrared emission that are associated with large grains in thermal equilibrium. The first (or ‘cold’) component represents dust surrounding the star-forming regions and thus the strength of the component is directly related to the star formation rate (however, see Section 4.5). While the SEDs of normal quiescent galaxies peak at ∼150 μm, the enhanced heating in actively star-forming regions produces a strong far-infrared peak at ∼60– 100 μm that dominates the infrared SED and corresponds to dust heated to 30–60 K (e.g. Calzetti et al., 2000; Dale et al., 2001a; Spinoglio et al., 2002). This component also enhances the mid-infrared continuum (e.g. see Figure 1 in Sanders and Mirabel, 1996). In AGN, the fractional contribution of this ‘cold’ far-infrared

OBSCURED “ACTIVITY”: AGN, QUASARS, STARBURSTS AND ULIGs

359

component can be weaker in comparison to the warmer mid-infrared component (λ ∼ 30 μm) than in starbursts, reflective of the intense heating power of the AGN. The second (‘very cold’ or ‘cirrus’) component is dust that is spatially more extended and is associated with the H I disk or the molecular halo (Trewhella et al., 2000; Radovich et al., 2001; Stickel et al., 2004). Located far and, in some galaxies, shielded (Calzetti et al., 2000) from the intense emission from the central engine, the extended component is heated by the ambient interstellar radiation field to temperatures of 10–20 K (Odenwald et al., 1998; Alton et al., 1998; Davies et al., 1999; Haas et al., 1998b; Haas, 1998; Trewhella et al., 2000). These temperatures are consistent with the theoretically predicted heating/cooling of dust grains in thermal equilibrium by a diffuse radiation field. Both profile measurements for resolved sources and far-infrared-sub-mm SED analysis for more distant galaxies reveal that a very cold component is required to explain all the emission beyond 100 μm (Rodr´ıguez Espinosa et al., 1996; Radovich et al., 1999; Siebenmorgen et al., 1999a; Domingue et al., 1999; Calzetti et al., 2000; Haas et al., 2000a, 2003; Klaas et al., 2001; P´erez Garc´ıa and Rodr´ıguez Espinosa, 2001). This component, that IRAS was insensitive to, may be a significant contributor to the far-infrared luminosity in some active galaxies (e.g. 60% in local starbursts, Calzetti et al., 2000). 2.2. FEATURES The advent of ISO spectroscopy enabled the first sensitive and high-resolution mid- to far-infrared spectroscopic measurements of active galaxies providing a better understanding of the physical conditions within visually obscured regions of systems such as Circinus (Moorwood et al., 1996; Sturm et al., 2000), the starburst M82 (Sturm et al., 2000) and the Seyfert 2 NGC 1068 (Lutz et al., 2000b). The full 2.5–200 μm spectra of these three galaxies are shown in Figure 1, and display the richness of the infrared range and the differences between these active galaxy templates (Sturm et al., 2000). The composite source Circinus displays broad midinfrared features, fine structure lines of low excitation, H-recombination lines and a thermal continuum that all predominantly trace the starburst component these are intermingled with the strong VSG and mid-infrared AGN continua and high excitation fine structure lines that are predominantly caused by the hard radiation field of an AGN. In addition, warm and cold molecular gas and warm atomic gas are probed by rotational lines of H2 , molecular absorption features (e.g. OH, CH, XCN and H2 O) and very low excitation (0.5) in active

362

A. VERMA ET AL.

regions and galaxies than in more quiescent regions such as in normal galaxies and the disks of some active galaxies (ratio ∼1 with a large dispersion, see Figure 1 of Vigroux et al. (1999) and Figure 12 in Dale et al. (2000)). Weak detections of UIBs in galaxies with AGN probably originate from the cooler material such as circumnuclear starburst regions or diffusely heated extended regions rather than from the immediate vicinity of the AGN torus. In nearby Seyferts, spatially resolved mid-infrared spectroscopy suggested that the absence or suppression of UIB emission is due to the fact that the dust is predominantly heated by processes related to the central AGN (e.g. in NGC 1068 (Le Floc’h et al., 2001), Circinus (Moorwood, 1999), NGC 4151 (Sturm et al., 1999), Mrk279 (Santos-Lle´o et al., 2001)). This was confirmed through a comparison of ISO and high-spatial resolution ground-based data of AGN and starbursts, that demonstrated the absence of UIBs in the nuclei of AGN-hosting galaxies (Siebenmorgen et al., 2004b). For unresolved sources, the VSG continuum may be so strong that the UIB features are diluted by the overpowering continuum resulting in their non-detection (Lutz et al., 1998b). Thus, at high redshifts, where low-metallicity- and/or AGN-dominated systems may be more prevalent, the intrinsic weakness of UIB features seen in local systems of these types would imply that the use of UIB features as a tracer of similar systems at high redshift may be problematic.

2.2.2. Fine Structure Lines Fine structure lines are tracers of nebular conditions such as excitation and the intrinsic ionising spectrum. By virtue of their differing excitation potentials and critical densities they provide an insight into the energetics and chemical composition of the regions from which they originate. Starburst galaxies commonly show low excitation fine structure lines (e.g. [Ar II], [Ar III], [Fe II], [Ne II] and [S III]) that have ionisation potentials between 13.6 and ∼50 eV and mostly arise from H II regions that have been photoionized by their central massive stars. A PDR or shock origin may also contribute to the flux of some low-excitation lines. High-excitation lines with potentials exceeding 50 eV (e.g. [Ne V, VI], [S IV], [Mg V, VII, VIII] [O IV], [Si V, VI, VII]) are present in the spectra some low-metallicity galaxies but are more commonly found in AGN where high-energy photons, produced by non-thermal processes related to the central black hole, photoionise the gas. Excitation to these high-energy levels (>50 eV) is difficult to attain by photons emitted by stars in typical H II regions. However, the mid-infrared spectra of some starbursts do show the high-excitation [O IV] and [S IV] lines. The presence of these lines is associated to collisionally excited, shocked or coronal gas (e.g. Lutz et al., 1998a). Active galaxy spectra also show fine structure lines with excitation potentials below the ionisation potential of hydrogen (e.g. [O III]5288 μm, [O I]63,145 μm and [C II]158 μm), that trace mostly cool atomic and molecular clouds. [C II]158 μm and [O I]63 μm are important cooling lines in PDRs but also originate in widespread atomic gas (see Section 4.6).

OBSCURED “ACTIVITY”: AGN, QUASARS, STARBURSTS AND ULIGs

363

As infrared fine structure lines are fairly insensitive to uncertainties of the effective temperature of the gas, and because they are less affected by extinction than optical lines, they provide an unequalled insight into the properties of the ISM (e.g. electron temperature and density, ISRF, metallicity) of active galaxies and their embedded power source. Ratios of two lines of the same species provide information on the nebular conditions such as electron temperatures and densities. Ratios of high- to low-excitation states of the same element probe the hardness of the ISRF and ratios of high- to low-excitation lines of different elements have been used as tracers of AGN presence and thus contribute to nebular diagnostics (see Section 6.1). ISO-spectroscopy enabled a major advance over IRAS by allowing the dominant fuelling mechanism in a given source to be identified (Genzel et al., 1998; Laurent et al., 2000; Sturm et al., 2000). 2.2.3. Molecular Features Molecular features trace the cold dense molecular material in active galaxies. Several active gas-rich starbursts, Seyferts and ULIGs display a range of molecular lines: the rotational lines of molecular hydrogen (Section 2.2.3.2), OH, H2 O, NH3 and CH are seen in absorption or emission (see Figure 1 of Fischer et al., 1999b; also refer to Hur et al., 1996; Fischer et al., 1997, 1999a,b; Bradford et al., 1999; Colbert et al., 1999; Sturm et al., 2000; Rigopoulou et al., 2002; Spoon et al., 2002; Spinoglio et al., 2005). Molecular absorption features appear to be more common (and prominent) in sources of high luminosity (e.g. ULIGs) than those of lower luminosity (Fischer et al., 1999a,b; Spoon et al., 2002). Spectra displaying molecular absorption features show a wide variety of excitation: from starburst galaxies like M82 which display features indicating that most of the O and C molecules occupy the ground state, to objects such as ULIGs where a significant population exists at higher energy levels (Fischer et al., 1999a,b). Additional absorption features at 6.85 and 7.25 μm are seen in the spectra of some active galaxies and ULIGs (Spoon et al., 2002). They are attributed to CH-deformation modes of carbonaceous material because of their similarity to features seen along lines of sight towards Sgr A* (Chiar et al., 2000). ULIG and Seyfert spectra also exhibit an absorption feature at 3.4 μm that is likely to be aliphatic CH stretch absorptions of refractory carbonaceous material (Imanishi, 2002; Dartois et al., 2004). 2.2.3.1. Silicate Absorption. The 9.7 and 18 μm absorption features arise from the stretching and bending modes of SiO. The extinction suffered by a galaxy can be estimated by the depth of these lines. For example, the extreme 9.7 μm silicate absorption feature seen in the starburst-ULIG Arp220 provides a lower limit to the optical depth of AV  45 (Charmandaris et al., 1997), but see Spoon et al. (2004) for a more detailed analysis of this system). Though it is possible that the depth of the 9.7 μm may be overestimated in the feature-dense mid-infrared spectra of starbursts exhibiting UIB emission (Sturm et al., 2000; F¨orster Schreiber et al., 2003). The

364

A. VERMA ET AL.

extreme extinction of the starburst M82 (AV = 15–60) derived by Gillett et al. (1975) was ascribed to this problem as ISO data could not support such a high extinction (Sturm et al., 2000). 2.2.3.2. Molecular Hydrogen and Warm Molecular Gas. Rotational transitions of H2 that originate in dense molecular media were detected in the spectra of several active galaxies: NGC 3256 (Rigopoulou et al., 1996); NGC 891 (Valentijn and van der Werf, 1999); NGC 1068 (Lutz et al., 2000b); NGC 4945 (Spoon et al., 2000); NGC6 240 (Lutz et al., 2003). Low rotational transitions originate from warm gas (T ∼ 100–200 K) and the higher and rovibrational lines from much hotter gas (T  1000 K). Rigopoulou et al. (2002) investigated a sample of starbursts and Seyferts observed by ISO presenting pure rotational lines from S(7) to S(0). Irrespective of starburst or Seyfert type, temperatures of ∼150 K are derived from the S(1)-to-S(0) ratio. A similar temperature was found for NGC 4945 (Spoon et al., 2000). While ∼10% of the galactic gas mass in a starburst system can be accounted for by the warm component, on average higher fraction (18%, range 2–35%) are observed for Seyferts (Rigopoulou et al., 2002). The temperature of the molecular hydrogen implies that it arises from a combination of emission from fairly normal PDRs (at least for starbursts), low velocity shocks and X-ray heated gas around the central AGN. The latter is more probable in Seyfert galaxies which accounts for the larger fraction of warm gas found in these systems. Shock heating appears to be the most likely origin of the extremely strong rotational H2 lines detected in the merging ‘double active nucleus’ system NGC 6240 (Lutz et al., 2003). While, the origin of shock heating can be largely associated to winds originating from massive stars and supernovae, recently Haas et al. (2005) have found the first observational evidence of pre-starburst shocks in the overlap region of the early merger, the Antennae, (see Section 4.1). From a re-analysis of archival ISOCAM-CVF data, they find that the strongest molecular hydrogen emission (traced by the H2 S(3) line) is displaced from the regions of active star formation. This, together with the high line luminosity (normalised to the far-infrared luminosity) indicates that the bulk of excited H2 gas is shocked by the collision itself. 2.2.3.3. Ice Absorption and Cold Molecular Gas. An interesting discovery made by ISO was the first extragalactic detection of absorption features due to ices (H2 O, CH4 and XCN) present in cold molecular components of starbursts, Seyferts and predominantly ULIGs. Sturm et al. (2000) first reported the detections of weak water ice absorption features at 3 μm in the spectra of local starbursts M82 and NGC 253. Very strong water ice, as well as the first detections of absorptions due to CO and CO2 ices, were detected in the actively star-forming galaxy NGC 4945 (Spoon et al., 2000). The presence of ices is related to the presence of cold material and the radiation environment. One would expect cold molecular clouds in starbursts to be likely hosts, while more intense environments, such as

OBSCURED “ACTIVITY”: AGN, QUASARS, STARBURSTS AND ULIGs

365

those predominantly fuelled by AGN, to be less favoured. This is corroborated by the lack of absorption features in the spectrum of NGC 1068 (Sturm et al., 2000) and the limits on the strength of the CO absorption feature set from ISOSWS spectra of 31 AGN (Lutz et al., 2004b). However, water ice absorption has been detected in Seyfert galaxies, e.g. in the UIB-free spectrum of the Seyfert galaxy NGC 4418 (Spoon et al., 2001). The absorption-dominated spectrum bears strong similarities to the spectra of embedded protostars, and the depth of absorption features implies that its Seyfert nucleus is so deeply embedded that ices are shielded from the AGNs intense radiation field. In a heterogeneous sample of 103 active galaxies with high signal-to-noise mid-infrared spectra, approximately 20% display absorption features attributed to ices (Spoon et al., 2002). While ices are weak or absent in the spectra of starbursts and Seyferts, in ULIGs they are strong, considerably stronger than the weak features found previously in M82 and NGC 253. This implies that rather than the radiation environment, it is the amount of cold material that determines their presence in these most luminous and massive infrared galaxies that are known to contain large concentrations of molecular material (e.g. Solomon et al., 1997; Downes and Solomon, 1998). Although plausible, more sensitive (and spatially resolved) mid-infrared spectroscopy is required to understand these differences, and the distribution and energetic conditions of cold molecular material. Moreover, higher spectral resolution over a wider mid-infrared wavelength range is necessary for accurate decomposition of the spectra into their various components, which will aid the identification of subtle and/or blended features (e.g. Spoon et al., 2004). Spoon et al. (2002) proposed an evolutionary sequence for their ice spectra ranging from a highly obscured beginning of star formation where ices dominate the spectra, to a less obscured stage of star formation where the UIBs become stronger. At any stage, the star formation may be accompanied by AGN activity. 2.2.3.4. OH Features and (Mega-)Masers. Transitions of the OH molecule were also detected in the ISOSWS and ISOLWS spectra of some active galaxies (Fischer et al., 1999b), such as in the starbursts NGC 253 (Bradford et al., 1999) and NGC 4945 (Brauher et al., 1997) permitting estimation of the OH column density and abundance. The depths of the OH features detected in the ISOLWS far-infrared spectra of galaxies hosting known OH (mega-)masers were found to be sufficient to provide all of the infrared photons required to radiatively pump the maser activity (e.g. in the starburst-ULIG Arp220 (Skinner et al., 1997; Suter et al., 1998); AGN-ULIG Mrk231 (Suter et al., 1998); the merging double systems ULIG IRAS 20100-4156 and LIG 3 Zw 35 (Kegel et al., 1999); the starburst NGC 4945 (Brauher et al., 1997)). In Arp220, the fact that the far-infrared continuum source is extended and that the maser must be located in front of it favours the interpretation that a starburst provides the continuum photons for its maser (Skinner et al., 1997). The high pump rate determined for this source suggests that pumping mechanisms other

366

A. VERMA ET AL.

than radiative pumping (such as collisional pumping) may contribute (He and Chen, 2004).

2.3. MAGNETICALLY ALIGNED DUST GRAINS Polarised emission detected from active galaxies is ascribed to dichroic absorption by the ISM of a galaxy i.e. by elongated dust grains that are magnetically aligned. This has been observed in galactic sources where the position angle of the measured polarisation indicates the orientation of the projected magnetic field. Polarisation is an important probe of the physical conditions in active galaxies, as it can provide constraints on the formation of the putative torus (Alexander et al., 1999b). Such polarisation studies were performed for some ULIGs and revealed 3–8% of the 5–18 μm flux to be polarised (Siebenmorgen and Efstathiou, 2001). The archetypal starburst-like ULIG Arp220 has the lowest polarisation, while the warm AGN-like ULIG Mrk231 has the highest polarisation. The first polarisation profile of a starburst galaxy was made at 6 μm for NGC 1808 showing a 20% increase towards the outer regions (Siebenmorgen et al., 2001). This, together with a complementary measurement at 170 μm (unresolved), are best explained by the large-scale ˚ (∼500 pc) magnetic alignment of (non-spherical) large grains (≥100 A).

3. Very Cold Dust and Diffuse Radiation Fields: ISM ISOPHOT-data clearly confirmed the presence of very cold dust (T ∼ 10–20 K) in a variety of infrared galaxies. Often in conjunction with supplementary (sub-)mm photometry, this component was detected in spatially resolved maps or intensity profiles of local galaxies (e.g. Haas et al., 1998b), and was isolated as a significant contributor to the far-infrared luminosity in SED analyses of unresolved active galaxies (Calzetti et al., 2000; Klaas et al., 2001; Spinoglio et al., 2002; Haas et al., 2003). Although cold ‘cirrus’ or ‘disk’ components in galaxies were indicated by IRAS colours, IRAS was insensitive to this very cold component due to its shorter wavelength coverage ( 50 μm) using a single modified black-body. Invariably, for active galaxies, this was achieved with temperatures ∼30–50 K and emissivity values of 1.5–2 (e.g. Colbert et al., 1999; Siebenmorgen et al., 1999a). However, doing so raised conflicts with other observational constraints (see Klaas et al. (2001) for a full description). Moreover, by leaving the dust emissivity as a free parameter presupposes that intrinsic dust properties differ from source to source, with no physical basis supporting this assumption. By choosing a Galactic value of γ = 2 for the grain emissivity (i.e. assuming that dust properties do not vary between sources), single black-bodies fail to explain all the far-infrared emission – additional cold components (T ∼ 10–20 K) are required that are associated with the spatially resolved extended very cold components described earlier. The SEDs of active galaxies are warmer than those of normal galaxies and the far-infrared part is generally fit by two components representing active star formation and very cold (diffusely heated) material. The very cold dust can be a significant contributor in terms of galaxy mass and energy output. For example, two of the three components used to fit the SED of the LINER NGC 3079 have very cold temperatures of 12 and 20 K, respectively (Klaas and Walker, 2002). Calzetti et al. (2000) find that two dust components are required at T = 40–55 K and T ∼ 20–23 K to explain the far-infrared SEDs of eight local starbursts. The latter comprises up to 60% of the total flux and the associated mass is as much as 150 times that of the former. For the most actively star-forming galaxies or those with strongest radiation fields, the contribution to the bolometric luminosity from cold emission (as measured in the 122–1100 μm range) was shown to be smaller in actively star-forming galaxies (5%) than in cirrus-dominated galaxies (∼40%) (Dale et al., 2001a). SED analysis of a wide range of active galaxies indicates that very cold components are common whether they are centrally fuelled by AGN or starbursts, suggesting that the central activity does not directly affect the far-infrared emission.

368

A. VERMA ET AL.

4. Cold Dust and Low Excitation Radiation Fields: Starburst Regions The T = 40–45 K temperature component found in local starbursts by Calzetti et al. (2000) mentioned earlier represents the emission from dust surrounding sites of active star formation. For starburst galaxies, this emission (peaking between 60 and 100 μm) dominates the infrared SED and is the major contributor to the infrared luminosity. This ‘cold’ infrared emission is generally less extended than the very cold component described in the preceding section. In some nearby galaxies, the far-infrared morphology traces star-forming regions well. For example, the strongly far-infrared-emitting dark molecular cloud coincides with a cluster of starforming regions in the galaxy NGC 4051 (P´erez Garc´ıa et al., 2000). Similarly, farinfrared emission could be resolved for the pair members of interacting spirals where enhanced star formation has been established (Xu et al., 2000; Gao et al., 2001). The correlation of the warmer far-infrared emission with the R-band morphology of local Seyferts hosting compact nuclei and circumnuclear star formation further supports that this far-infrared component is related to star formation activity (Rodr´ıguez Espinosa and P´erez Garc´ıa, 1997). Moreover, far-infrared emission has also been spatially associated with regions exhibiting signs of strong star formation (e.g. strong UIBs). Roussel et al. (2001) demonstrated that the integrated far-infrared flux and the star formation related mid-infrared flux are strongly correlated in the disks of normal star-forming galaxies, indicating that the grains responsible for the mid- and far-infrared share a common heating source.

4.1. THE I MPORTANCE

OF I NTERACTIONS /M ERGERS

Interactions and mergers enhance starburst activity, both nuclear and extra-nuclear (see review by Struck, 1999). The final merger stage plays a critical role in how active galaxies form stars and may influence the appearance of AGN and the onset of the ULIG phase. The fraction of disturbed or interacting systems in infrared selected samples of galaxies appears to increase with luminosity with nearly 100% of all ULIGs displaying evidence of interaction (see Sanders and Mirabel, 1996, and references therein). Numerical simulations have also helped in establishing that interactions and mergers play an important role in the formation and evolution of such galaxies (Mihos and Hernquist, 1996). The induced gravitational instabilities form bars which strip in-falling gas of angular momentum and enable radial inflows to the nuclear regions that can feed starbursts or AGN (Combes, 2001). Moreover, mergers appear to be the triggering mechanism for the ultraluminous phase. Dynamical shocks and tidal forces are the only mechanisms that can cause transference and compression of sufficient quantities of matter to fuel luminosities in excess of 1011 L  . ISOCAM’s spatial scale was sufficient to map the morphologies of local interacting active systems (see Mirabel and Laurent, 1999 for a review) and agree with

OBSCURED “ACTIVITY”: AGN, QUASARS, STARBURSTS AND ULIGs

369

results from ground-based high-resolution images that (optically faint or invisible) compact sources located in deeply obscured nuclear regions or interaction interfaces dominate the mid-infrared emission (e.g. Soifer et al., 2000, 2001; Charmandaris et al., 2002; Gallais et al., 2004; Siebenmorgen et al., 2004b). For some sources, diffuse emission between these ‘hot-spots’ can also contribute significantly to the total mid-infrared emission (e.g. Le Floc’h et al., 2002; Gallais et al., 2004). This diffuse emission can be enhanced in merging systems that are common among active galaxies. By probing through the famous characteristic dust lane of the AGN-hosting elliptical galaxy Centaurus A (NGC 5128), imaging with ISOCAM revealed a barred ‘mini-spiral’ located in its central 5 kpc. The mini-spiral was proposed to be tidal debris from a gas-rich object that was accreted in the last gigayear (Mirabel et al., 1999; Block and Sauvage, 2000). Evidence of accreted companions in Centaurus A exists from previous studies of its stellar and gas kinematics (Charmandaris et al., 2000, and references therein). ISOPHOT contributed to the accretion scenario through the detection of the northern cold dust shell that is possibly the ISM remnant of a captured disk galaxy (Stickel et al., 2004). As a result, Centaurus A is a prime example of a giant elliptical that must have experienced several minor mergers. It is likely that, as the interactions took place, tidal forces funnelled gas into the dynamical centre and boosted the star formation rates in the gas-rich components forming the dust observed in the infrared and visible.

4.1.1. Nuclear and Extra-Nuclear Starbursts Disturbed morphological signatures such as rings, tidal tails, bridges and arcs are clearly detected in ISOCAM observations of numerous merging galaxies (e.g. the Antennae (Mirabel et al., 1998) Figure 2; UGC12915, 12914 (Jarrett et al., 1999); NGC 985 (Appleton et al., 2002); Arp299 (Gallais et al., 2004)). The mid-infrared morphologies depend on the mass ratio of the progenitor galaxies and the merger stage. As well as enhancing star formation in the nuclear regions, interactions cause a general increase in the widespread activity of the parent galaxies. At the early stages of interaction, wide scale star formation can be induced and dust grains are heated by a generally warmer radiation field. For example, ISOCAM observations of the VV 114 galaxy show that 60% of the mid-infrared flux arises from a diffuse component that is several kpc in size (Le Floc’h et al., 2002). Nearly 40% of the 7 and 15 μm emission detected in the early-stage merger (nuclei separation 5 kpc) galaxy pair Arp299 (NGC 3690 and IC 694) is diffuse and originates from the interacting disks (Gallais et al., 2004). In more advanced mergers, shocks and gravitational instabilities encourage the onset of star formation. The transfer of matter to nuclear regions and compression in interaction zones gives rise to enhanced mid-infrared emission in nuclear and often extra-nuclear sites. A clear example of extra-nuclear star formation was uncovered by ISOCAM in the intra-group medium of Stephan’s Quintet (Xu et al., 1999).

370

A. VERMA ET AL.

Figure 2. This composite figure of the Antennae is from Mirabel et al. (1998) showing the ISOCAMCVF contours overlaid on a HST V and I band image. Half of the mid-infrared emission arises from starburst activity that is completely obscured in the optical, including the brightest mid-infrared knot (knot A). The CVF spectrum of this knot is shown in the lower part of the image displaying strong signatures of massive star formation – the [Ne III] line and a strong rising continuum above 10 μm.

The best example of ISOCAM’s capability in probing through highly extincted regions is imaging of the spectacular major merger in the Antennae (Arp244, NGC 4038/4039). The completely obscured overlap region in the HST-WFPC2 image (AV ∼ 35 mag; Verma et al., 2003) was shown by ISOCAM to contain a

OBSCURED “ACTIVITY”: AGN, QUASARS, STARBURSTS AND ULIGs

371

massive star-forming region (50 pc in radius) comprising more than 15% of the total 12–17 μm luminosity which outshines the two nuclei by a factor of 5 (Mirabel et al., 1998). The warm mid-infrared emission traces the molecular gas that fuels star formation (Wilson et al., 2000, 2003), whereas the cold (T ∼ 30 K) and very cold dust (T ∼ 20 K) are traced by far-infrared ISOPHOT and sub-mm observations (Haas et al., 2000a). Haas et al. (2000a) find that the sub-mm knots, in particular K2, are likely to be in a pre-starburst phase and the simultaneous presence of powerful off-nuclear starbursts and pre-starbursts may be a general feature of colliding galaxies. The authors postulate that once star formation has commenced in these clouds, the Antennae may evolve from a luminous IR galaxy into an ultraluminous one. An interesting feature for the Antennae (also seen for Centaurus A) is that the most luminous 15 μm knots are not coincident with the darkest, most prominent dust absorption lanes seen in the HST image. Presumably the darkest optical dust lanes trace the coldest emission (i.e. that probed by the far-infrared) rather than the warm dust from which 15 μm emission arises. This is plausibly the cause of the displacement. Alternatively, projection effects may be responsible for this offset (Mirabel et al., 1998). Extranuclear overlap starbursts emitting in the mid-infrared are not unique to the Antennae. In a sample of eight spiral–spiral pairs with overlapping disks, Xu et al. (2000) found extranuclear, overlap region starbursts in five galaxies classified as advanced mergers or having severely disturbed morphologies (but not in lessdisturbed systems). However, the starbursts in the bridges or tidal tails are generally fainter than those in the nuclear regions. 4.1.2. Collisional Rings Collisional ring galaxies are produced when a compact galaxy passes through the centre of a larger disk galaxy giving rise to a radial density wave that creates a massive star-forming ring (see review by Appleton and Struck-Marcell, 1996). If the collision is off centre then the ring morphology is asymmetric e.g. as in VII Zw 466 (Appleton et al., 1999). These systems are excellent local examples of induced star formation in colliding galaxies. The effects of collisions on the ISM can be investigated in detail, and may be relevant for higher redshift galaxies including ULIGs (Appleton et al., 2000). Azimuthal variations in over-density in the rings are often noted as well as extranuclear star-forming knots that show strong UIBs (Charmandaris et al., 1999, 2001b; Appleton et al., 1999, 2002). When detected, star formation in the ring contributes at a level of 10% of the bolometric output (Appleton et al., 1999, 2000). The mid-infrared emission is clearly enhanced in the nucleus and the expanding ring, even though diffuse emission from the intervening medium was also observed (Charmandaris et al., 1999, 2001b). The diffuse components coincide with regions where only weak Hα had been previously measured. The mid-infrared colours measured in collisional ring galaxies are similar to those of late-type galaxies, indicative of more modest star formation than is found

372

A. VERMA ET AL.

for extranuclear starbursts (e.g. in the nucleus inner ring and spokes of Cartwheel, the entirety of Arp10 and the ring of Arp118 (Charmandaris et al., 2001b)). The morphological agreement between the mid-infrared and Hα or radio emission is good, although differences have been noted for VII Zw 466 (Appleton et al., 1999). A deviation from this picture is seen for the extranuclear starburst knot in the archetypal collisional ring galaxy, the Cartwheel, that coincides with the strongest Hα and radio knot. The regions have an unusually red mid-infrared colour (15/7 μm ∼ 5.2) which is the highest among the extragalactic regions observed with ISOCAM and it is even more extreme than the powerful starburst region (knot A) revealed in the Antennae (Charmandaris et al., 1999). Interestingly, this region also hosts the remarkable hyperluminous X-ray source (L 0.5–10 keV  1041 ergs/s; Gao et al., 2003) which dominates the X-ray emission from the Cartwheel ring.

4.2. HOT STAR P OPULATION , S TARBURST EVOLUTION

AND

M ETALLICITY

Starburst galaxies, defined as having star formation rates that cannot be sustained over a Hubble time, are likely the birthplace of a large fraction of massive stars. Locally, four starburst galaxies can account for 25% of the local massive star population within 10 Mpc (Heckman et al., 1998) and are important as the providers of the Lyman continuum photons that ionise hydrogen. ISO spectroscopy enabled the stellar populations of starbursts to be probed using fine structure lines that are sensitive to the nebular conditions from which they originate (see F¨orster Schreiber et al. (2001) for a detailed analysis of the stellar populations and radiation environment of the local starburst M82). The ratio of [Ne III]15.6 μm (E p = 41 eV) and [Ne II]12.8 μm (E p = 22 eV) is sensitive to the hardness of the stellar energy distributions of the OB stars which excite them. In a survey of 27 starburst galaxies, Thornley et al. (2000) confirmed earlier ground-based results that the excitation measured by the [Ne III]15.6 μm/[Ne II]12.8 μm ratio was lower in starburst galaxies than measured for Galactic H II regions, indicating a deficiency of massive stars in the starbursts. Thornley et al. suggest a lower upper mass cut-off to the initial mass function (IMF) or ageing of the massive star population to be responsible for the low excitation. Through extensive quantitative photo-ionisation modelling, Thornley et al. found the ratios to be consistent with the formation of massive stars (50–100M ). This, together with the known presence of very massive stars in local starburst regions, led Thornley et al. to prefer an ageing scenario to explain the low excitation. This explanation was further supported by modelling presented in Giveon et al. (2002). The modelling of Thornley et al. indicates a short starburst timescale of only a few O star lifetimes (106 –107 yrs) and suggests that strong negative feedback regulates the star-forming activity. Disruption due to stellar winds and SNe may cease the starbursts before the depletion of gas and suggests that periodic starburst events are likely (e.g. in M82 F¨orster Schreiber et al. (2001)).

OBSCURED “ACTIVITY”: AGN, QUASARS, STARBURSTS AND ULIGs

373

Verma et al. (2003) confirmed that for a given nebular abundance, local starburst nuclei are of lower excitation than Galactic and Magellanic cloud H II regions. This was shown for the primary products of nucleosynthesis argon, sulphur and neon confirming the Thornley et al. (2000) result was not restricted to Ne only. Additionally, the observed excitation was confirmed to be metallicity dependent, with the lowest metallicity compact dwarfs showing the highest excitations, at a similar level to local H II regions. Ageing, a modified IMF and metallicity are all possible contributors to the metallicity-excitation relation. In the same study, the abundance of sulphur was observed to be lower than expected for a primary product of nucleosynthesis. Wolf–Rayet and BCD galaxies did not show such an underabundance (Verma et al., 2003; Nollenberg et al., 2002). This under-abundance and weakness in the sulphur line strength implies that (a) the fine structure lines of neon are favoured over the those of sulphur as a star formation tracer in future spectroscopic surveys of galaxies and (b) that sulphur is an unsuitable tracer of metallicity.

4.3. WOLF–R AYET S TARBURSTS The Wolf–Rayet phase is a signature of massive star formation as the progenitors are believed to be massive young O stars (3  tsb  8 Myr, M  20M ). Wolf–Rayet features are found in a wide range of galaxies such as BCDs, H II galaxies, AGN, LINERs and starbursts (see Schaerer et al., 1999 for a compilation). Recently, L´ıpari et al. (2003) identified possible Wolf–Rayet features in the spectra of some PG quasars. The identifying Wolf–Rayet signatures are in the visible region and thus are only representative of the unobscured components. Nevertheless, a correlation between the star formation indicator UIB 7.7 μm/15 μm continuum with Hα in Wolf–Rayet galaxies suggests that the infrared and visible trace, at least partly, the same components (e.g. in Haro 3 Metcalfe et al., 1996). The short-lived Wolf– Rayet phase is commonly detected in low metallicity systems with a strong ISRF such as blue compact dwarfs (BCDs, see Section 4.4). ISO observations of known Wolf–Rayet galaxies reveal differences in their infrared properties compared to non-Wolf–Rayet galaxies. In comparison to starbursts, Wolf–Rayet galaxies are of higher excitation and lower abundance contrary to expectation of a more important role of Wolf–Rayet stars at higher metallicities from stellar evolution models (Verma et al., 2003). The Wolf–Rayet galaxy NGC 5253 has a weaker excitation determined from mid-infrared ISO data than expected by stellar models (Crowther et al., 1999) implying that the UV spectrum may not be as hard as commonly thought. However, Crowther et al. (1999) show that the effect is due to a young compact starburst surrounded by an older starburst both of which lie within the SWS aperture. As a result, measured line strengths are diluted and the apparent effective temperature of the emitting region is lowered, thus reducing the measured excitation. This effect, which is fully consistent with an ageing/decaying starburst scenario suggested by

374

A. VERMA ET AL.

Thornley et al. (2000), can explain the low excitation of starbursts described in Section 4.2 (Thornley et al., 2000; Verma et al., 2003). 4.4. BLUE C OMPACT DWARFS A number of low metallicity (Z  Z  ) blue compact ( 50 mag) and hydrogen column density (NH > 1023 cm−2 ). Although little can be stated regarding favoured torus models or constraining gas properties based on this information, Brα was identified as being the preferred indication for future studies and Lutz et al. (2002) found broad line components in approximately one-fourth of their sample of Seyfert galaxies from ground-based high-resolution spectroscopy. 6. Ambiguous Sources: Ultra- and Hyperluminous Infrared Galaxies As we have previously discussed, ULIG samples contain galaxies of both starburst and AGN type. Establishing whether these types are linked in evolution or nature is a key issue in the study of active galaxies raising several questions: what is the role of star formation and black holes; which is/are present; do super-massive black holes and starburst activity concurrently exist in all ULIGs and how much do they contribute to the bolometric luminosity? Prior to ISO, few observational tools were available to probe through the dust obscuration to reach the source of the infrared power. ISO’s spectroscopic and photometric capabilities enabled significant advances in this field, largely elucidating the previously termed ‘starburst-AGN controversy’ which is discussed in this section. 6.1. QUANTITATIVE S PECTROSCOPY 6.1.1. Fine Structure Line Diagnostics Fine structure lines directly trace the excitation states of the ISM from which they originate. The radiation fields generated by AGN/QSO greatly exceed that of starbursts and thus can excite interstellar gas to higher ionisation species (as described in Section 2.2.2). The difference in excitation between AGN and starbursts is the basis of fine structure line diagnostic diagrams. Such excitation diagrams are well established in the optical (Veilleux and Osterbrock, 1987; Kewley et al., 2001), for which Sturm et al. (2002) constructed the first infrared analogues (Figure 4) that successfully distinguish starbursts from AGN. Diagnostic diagrams from Sturm et al. (2002) are shown in Figure 4 both involving the fine structure line [O IV]25.91 μm. While this line was expected in the high ionisation potential of AGN, it was also often weakly detected in star-forming galaxies and ULIGs (Lutz et al., 1998a). In our galaxy, the [O IV]25.91 μm line has not been observed in H II regions surrounding young, hot, massive stars. However, its detection in ISO spectra of starbursts and the fact that it was spatially resolved in the low excitation starburst M82 implies it is plausibly produced by ionising shocks or extremely hot ionising stars (Lutz et al., 1998a; Schaerer and Stasi´nska, 1999). Its diagnostic capability thus arises from the fact that the measured strength of the line in starburst galaxy spectra is at least 10 times fainter than those measured in

386

A. VERMA ET AL.

Figure 4. Diagnostic diagrams based on fine structure lines from Sturm et al. (2002). (a) [Ne VI]/[O IV] vs. [Ne VI]/[Ne II]: (diamonds) Seyfert 1s, (asterisks) Seyfert 2s: the composite Seyfert galaxies that exhibit UIBs are encircled and are mostly separated from the remaining galaxies. (b) [O IV]26 μm/Brβ vs. [Si II]34 μm/Brβ: (diamonds) AGN, (asterisks) starbursts: the starbursts occupy a different regions of the diagnostic diagram than the AGN.

spectra of AGN. This absolute difference in the strength of [O IV] relative to a low excitation starburst line such as [Ne II] provides a straightforward indicator of AGN versus starburst activity (Sturm et al., 2002). Combining the high:high excitation fine structure line ratio [Ne VI]/[O IV] with the high:low excitation fine structure line ratio [Ne VI]/[Ne II], provides a basic diagnostic of composite sources (Figure 4a). By using the limiting values of the high:low excitation ratio [O IV]/[Ne II] (always 0.2). Type B sources display a relatively flat SED from 1 to 10 μm, that is followed by a step rise towards the peak of emission in the far-infrared. These SED characteristics are displayed by ULIGs with optical spectroscopic classifications of Seyfert 2, LINER, H II/starburst. These are ‘cool’ ULIGs with infrared colour f 25 μm / f 60 μm < 0.2. The grouping of LINERs in the ‘cool’ starburst-like ULIG class is consistent with the analysis of mid-infrared versus optical spectroscopic classification performed by Lutz et al. (1999) discussed in the previous section. This, together with the near-infrared colours (see Figure 6 in Klaas et al. (2001)) of ULIGs, suggest that the far-infrared emission from LINER ULIGs is of starburst origin. Klaas et al. (2001) interpreted the emission of ‘warm’ ULIGs (type A) to arise from dust heated to high temperatures

392

A. VERMA ET AL.

directly from the central AGN, which in turn dominates the mid-infrared emission. The second SED feature is ascribed due to thermal re-radiation by dust heated by starburst activity. The indistinguishable far-infrared SEDs of AGN- and starburst-dominated ULIGs suggests that the far-infrared emission largely comes from less active or shielded regions that are generally not heated by the AGN (Klaas et al., 2001). Thus, Klaas et al. (2001) proposed a three-stage dust model consisting of hot AGN-heated dust, warm starburst-heated dust (50 K > T > 30 K) and cold, pre-starburst or cirrus-like dust (30 K > T > 10 K). The general consensus from ISO spectroscopic results of warm ULIGs being AGN dominated and cold ULIGs starburst dominated is corroborated by the findings of Klaas et al. (2001). 6.5. HyLIGs AND

THE I NCREASING

ROLE

OF

AGN

AT

HIGH L UMINOSITY

The increasing importance of AGN with increasing luminosity (Shier et al., 1996; Veilleux et al., 1997) has been confirmed by ISO observations of ULIGs (Lutz et al., 1998b; Tran et al., 2001; Charmandaris et al., 2002). Tran et al. (2001) quantified that the change from starburst- to AGN-dominated systems occurs at log10 (L IR ) ∼ 12.4–12.5L  . While ISO results show that 1012 L  can be powered by star formation alone (Genzel et al., 1998; Lutz et al., 1998b; Rigopoulou et al., 1999; Tran et al., 2001), beyond 1013 L  an AGN contribution seems inevitable. Moreover, the production of such a luminosity would require star formation rates greater than or approximately few ×103 M yr−1 . Such rates require huge concentrations of molecular gas to be present (most likely the result of mergers) and highly ineffective negative feedback (Takagi et al., 2003). The prevalence of AGN in HyLIG samples partially arises from a selection effect. These rare luminous sources have been discovered with heterogeneous selection methods, most commonly through a correlation of known quasar or radio catalogues with IRAS or ISO surveys resulting in the known population being biased towards AGN. Constraining the properties of this sample and full consideration of the biases involved is unfortunately not possible due to the paucity of known HyLIGs. Nevertheless, ISO observed several hyperluminous galaxies in a range of programs. Only a few HyLIGs were sufficiently bright to permit low resolution spectroscopic observations. Analyses of IRAS 09104+4109 (Tran et al., 2001) and IRAS F15307+3252 (Aussel et al., 1998) showed that the mid-infrared emission predominantly originates from very compact regions 0.2 intervals (taken from Pozzi et al. (2004)). Both plots show significant evolution.

the full 15 μm number counts (Figure 1, right) and the redshift distributions from the deeper ISO-CAM surveys (Aussel et al., 1999; Flores et al., 1999, 2004). We have also calculated the 90 μm luminosity function from the Preliminary Analysis (Serjeant et al., 2001) and again from the Final Analysis with more spectroscopic redshifts in Serjeant et al. (2004), Figure 5 (right). This function is also significantly different from the redshift zero IRAS counterpart. The ELAIS ISO data has also helped in the understanding of the star-forming galaxies seen by SCUBA. Using constraints on the SEDs of 19S850 μm > 8 mJy SCUBA sources. Fox et al. (2002) concluded that all had z > 1 and half had z > 2. 6. Active Galactic Nuclei As we have already said, SED modeling suggests around 11% of the infrared sources have SEDs dominated by emission from dusty tori around AGN. Also around 24% of the sources contain an AGN, even if it does not necessarily dominate the infrared luminosity. So it is clear that AGN are related, even if not causally, with the infrared galaxy phenomenon. It is interesting to note that the fraction of AGN seen in infrared galaxies increases with the infrared luminosity Figure 6 (left) an effect that had previously been noted in IRAS infrared galaxies at lower redshifts (Sanders and Mirabel, 1996). We have also calculated the 15 μm luminosity functions for AGN (Type-1 and 2, Matute et al., 2002, 2005). These show very significant evolution (see Figure 7).

418

S. OLIVER AND F. POZZI

Figure 6. Left: Fraction of infrared galaxies containing (bottom, or not containing – top) an AGN as a function of infrared luminosity. Right: Infrared luminosity as a function of black hole mass. Taken from Afonso-Luis et al. (2004).

Figure 7. AGN luminosity functions at 15 μm. Points represent the observed space density at different redshift, while lines represent the best-fit model predictions. Left: objects spectroscopically classified as Type-1 AGN; Right: objects spectroscopically classified as Type-2 AGN (taken from Matute et al. (2005)).

As yet we have too few AGN to determine the luminosity function in different redshift bins, but the IRAS sample of Rush et al., (1993) gives a low-z counterpart. The Type-1 AGN evolution is consistent with pure luminosity evolution at a rate of (1 + z) Q with Q ∼ 2.6 similar to that already seen for this type of sources at other wavelengths and the infrared galaxies. A similar evolution scenario is found for Type-2 AGN with Q ∼ 2–2.6 (depending on the assumed SED) and this is the first time the shape and evolution for Type-2 AGN has been derived. Alexander et al. (2001) have investigated the 15 ELAIS sources that were also detected by ROSAT. Six of these were broad-line QSOs, four were narrow-line star-forming galaxies or Type-2 AGN, three were stars, and two were unidentified optically. They did not find any particularly hard sources, arguing against heavily obscured AGN. Basilakos et al. (2002) conducted a 2–10 keV survey over 2.5 square

THE EUROPEAN LARGE AREA ISO SURVEY

419

degree in S1 using Beppo-SAX. They found 17 hard X-ray sources, 10 of which had 15 μm counterparts and eight of these have spectral classifications: six QSOs, one Type-2 AGN and one apparently normal galaxy. These observations were too faint to investigate the bulk of the ISO populations. Manners et al. (2004) have investigated the associations between 15 μm sources and deep Chandra observations in ELAIS N1 and estimate an AGN fraction of 19% with the emission from the other ISO sources consistent with star formation. A similar AGN fraction has been observed in deeper 15 μm fields using the same X/MIR correlation method (HDF-N and Lockman Hole: Fadda et al., 2002). To-date, the X-ray surveys are either too small or too shallow to address this AGN fraction definitively. Afonso-Luis et al. (2004) have investigated optically selected QSOs in the ELAIS N1 field and found that their optical properties are largely indistinguishable from the QSOs, 15 μm, selected sample. They do find an apparent correlation between the black hole mass and the infrared luminosity (Figure 6, right). 7. Hyper-Luminous Galaxies Objects as luminous as those seen in the SCUBA surveys are very rare. The wide areal coverage and large volume (5×106 Mpc−3 ) of ELAIS makes it ideal for discovering rare objects such as these. So far we have discovered nine galaxies whose infrared luminosities probably exceed L > 1013.22 L  (Rowan-Robinson et al., 2004) (60% of these are based on photometric redshifts and await spectroscopic confirmation). It is very likely that this number increases as the follow-up proceeds but this is already almost as many as have ever been discovered from the IRAS database. The bolometric emission from first of these, ELAISP90 J1640101410502 (Morel et al., 2001), has been modeled with an almost equal contribution from star formation and accretion disk processes. Such objects may provide us with accessible “local” counterparts to the elusive SCUBA galaxies or, if they are unrelated, will provide interesting additional challenges to galaxy evolution models. 8. Large-Scale Structure Very recently we have been able to analyze the clustering within ELAIS. In the southern field, S1, we have estimated the angular correlation function (Figure 8, Gonzales-Solares et al., 2005). Using the median redshift z ∼ 0.2 (La Franca et al., 2004) and inverting Limber’s Equation we have estimated the 3D infrared galaxygalaxy autocorrelation function with amplitude, encapsulated in the correlation length r0 = 4.3+1.0 −0.6 h − 1 Mpc. This is consistent with measurements from local infrared samples (r0 = 3.79 ± 0.14 at z ∼ 0.02, IRAS (Saunders et al., 1992); r0 = 3.7 at z ∼ 0.02, PSC-z (Jing et al., 2002) and weaker than optical (r0 = 5.7 at z ∼ 0.05 – APM, Maddox et al., 1990; r0 = 4.92 ± 0.27 at z ∼ 0.08 – 2dFGRS, Norberg et al., 2001; r0 = 5.7 ± 0.2 at z ∼ 0.18 – SDSS, Zehavi et al., 2002).

420

S. OLIVER AND F. POZZI

( )

1.00

0.10

0.01

0.10

(deg)

1.00

Figure 8. The angular correlation function measured in our southern field S1 (taken from GonzalesSolares et al. (2005)).

Our results thus conform to the existing picture i.e., the clustering strength of infrared samples is less than optically selected samples. This result appears to contradict that presented by Elbaz et al. (2004) who found that in deep infrared surveys the infrared galaxies were more strongly clustered that the optical galaxies. However, this may suggest that within the ISO surveys we have already detected the evolution of clustering. Within the ELAIS survey itself Vaisanen and Johansson (2004) find a significant over-density of EROs around faint ELAIS mid-IR sources. A radial distribution of the EROs suggests a physical connection of these EROs and ISOCAM sources, which also suggests that the faintest ELAIS sources are more strongly clustered. This would be qualitatively consistent with the predictions of hierarchical models where star formation occurs first in the strongly clustered peaks of the density field and takes longer to initiate in the lower density, more weakly clustered “field.” 9. The ELAIS Legacy The ELAIS fields were carefully chosen to avoid contamination from dust emission from our own Galaxy (Oliver et al., 2000). Coupled with the ever expanding multiwavelength data available the ELAIS fields are a prime target for new surveys. A number of surveys and projects have been carried out whose goals are relatively independent of the ISO observations themselves. Notable amongst these are the SCUBA 8 mJy (Scott et al., 2002) and MAMBO surveys (Greve et al., 2004) and X-ray surveys (Willott et al., 2003, 2004; Manners et al., 2003). These fields are particularly ideal for Spitzer and the ISO 15 μm band data (inaccessible to Spitzer) will be extremely valuable. For example the GOODS validation field (Chary et al., 2004) was executed in ELAIS fields. More significantly, three of the ELAIS fields are core components of the SWIRE fields (Lonsdale et al., 2003, 2004).

THE EUROPEAN LARGE AREA ISO SURVEY

421

10. Discussion We have demonstrated that infrared emitting galaxies evolve strongly. The indications are that the star forming galaxies make up the bulk of the excess over no evolution predictions. Low star formation rate (infrared luminosity) systems have similar infrared/optical ratios which is a measure of activity of the star formation (star-formation rate per unit stellar mass) so their total star-formation rate is determined by the total stellar mass. However, at high star-formation rates there is a correlation between the infrared/optical excess and the infrared luminosity (total star-formation rate) and a corresponding lack of correlation with the optical luminosity (total stellar mass), indeed very high infrared/optical excess galaxies are found with very modest optical luminosities. This suggests that active star formation is associated with factors external to the galaxy. AGN are found in 20–25% of the sources, but star formation may still be the dominant source of the infrared luminosity in all but 11% of the sources. The AGN and infrared phenomena are however, closely linked: the AGN fraction increases at higher luminosity; the infrared luminosity functions evolve at a similar rate; and the infrared luminosity is correlated to the black hole mass. This may imply a common factor in the AGN and star-formation phenomena. At the extreme end we have discovered nine hyper luminous galaxies, comparable with the total number discovered by IRAS. We detect clustering with an amplitude consistent with lower redshift infrared samples and lower than optical samples, this contrasts with the higher clustering of infrared galaxies in a deeper sample suggesting that an evolution of clustering may have been detected in the ISO surveys. It seems quite plausible that the environment provides the external factor determining active star formation and the common factor linking star formation and AGN activity, thus we might expect the evolution in the luminosity functions to be linked to an evolution in the clustering. The ELAIS samples will provide invaluable templates for understanding the deeper samples detected by the Spitzer GTO and legacy surveys (e.g., Lonsdale et al., 2003; Dickinson et al., 2004) and the ELAIS fields themselves provide a valuable legacy for surveys with Spitzer and other facilities. The complete ELAIS survey comprising a band-merged catalogue with multi-wavelength associations is now publicly available at http://astro.imperial. ac.uk/Elais/

Acknowledgements The authors would like to thank Michael Rowan-Robinson and the whole ELAIS team for their enormous efforts in making the ELAIS survey a reality and “we” in the text refers to the ELAIS team. We would like to thank Phillippe H´eraudeau, Israel

422

S. OLIVER AND F. POZZI

Matute, Alejandro Afonso Luis and Evanthia Hatziminaoglou for providing figures in advance of publication and Petri V¨ais¨anen for useful comments. This chapter was based in part on an original conference proceeding (Oliver et al., 2004). References Afonso-Luis, A., et al.: 2004, MNRAS 354, 961. Alexander, D., et al.: 2001, AJ 554, 18. Aussel, H., et al.: 1999, A&A 342, 313. Babbedge, T., et al.: 2004, MNRAS 353, 654. Basilakos, S., et al.: 2002, MNRAS 331, 417. Cesarskey, C., et al.: 1996, A&A 315, L32. Chary, R., et al.: 2004, ApJS 154, 80. Ciliegi, P., et al.: 1999, MNRAS 302, 222. Condon, J.: 1992, AnRA&A 30, 575. Dickinson, M., et al.: 2004, in A. Renzini and R. Bender (eds.), Proceedings of Multiwavelength Mapping of Galaxy Formation and Evolution Conference, Venice (Italy), October 2003. Dole, H., et al.: 2001, A&A 372, 364. Efstathiou, A., et al.: 2000, MNRAS 316, 1169. Elbaz, D., et al.: 2002, A&A, 384, 848. Elbaz, D., et al.: 1999, A&A 351, L37. Elbaz, D., et al.: 2004, in A. Renzini and R. Bender (eds.) Proceedings of Multiwavelength Mapping of Galaxy Formation and Evolution Conference, Venice (Italy), October 2003, astro-ph/0403209. Fadda, D., et al.: 2002, A&A 383, 838. Flores, H., et al.: 1999, ApJ 517, 148. Flores, H., et al.: 2004, A&A 415, 885. Fox, M., et al.: 2002, MNRAS 331, 839. Gonzales-Solares, E., et al.: 2004, MNRAS 352, 44. Gonzales-Solares, E., et al.: 2005, MNRAS 358, 333. Greve, T., et al.: 2004, MNRAS 354, 779. Gruppioni, C., et al.: 2001, MNRAS 335, 297. Gruppioni, C., et al.: 2002, MNRAS 335, 831. Gruppioni, C., et al.: 2003, MNRAS 341 1L. Guiderdoni, B., et al.: 1998, MNRAS 295, 877. H´eraudeau, P., et al.: 2004, MNRAS 354, 924. Hauser, M. G., and Dwek, E.: 2001, ARA&A 39, 249. Hughes, D., et al.: 1998, Nature 394, 241. Jing, Y. P., B¨orner, G., and Suto, Y.: 2002, ApJ 564, 15. Johansson, P. H., V¨ais¨anen, P., and Vaccari, M.: 2004, A&A 427, 795. Kessler, M., et al.: 1996, A&A 315, L27. La Franca, F., et al.: 2004, AJ 127, 3075. Lari, C., et al.: 2001, MNRAS 325, 1173. Lemke, D., et al.: 1998, A&A 315, L64. Lonsdale, C., et al.: 2003, PASP 115, 897. Lonsdale, C., et al.: 2004, ApJS 154, 54. Maddox, S. J., et al.: 1990, MNRAS, 242, 43P. Manners, J., et al.: 2003, MNRAS 343, 293. Manners, J., et al.: 2004, MNRAS 355, 97. Matute, I., et al.: 2002, MNRAS 332, L11.

THE EUROPEAN LARGE AREA ISO SURVEY

423

Matute, I., et al.: 2005, in preparation. McMahon, R. G., et al.: 2001, New Astron Rev 45(Issue 1–2), 97. Metcalfe, L., et al.: 2003, A&A 407, 791. Morel, T., et al.: 2001, MNRAS 327, 1187. Norberg, P., Baugh, C. M., Hawkins, E., et al.: 2001, MNRAS 328, 64. Oliver, S., et al.: 1997, MNRAS 289, 471. Oliver, S., et al.: 2000, MNRAS 316, 749. Oliver, S., et al.: 2002, MNRAS 332, 356. Oliver, S., et al.: 2004, in A. Renzini and R. Bender (eds.), Proceedings of Multiwavelength Mapping of Galaxy Formation and Evolution Conference, Venice (Italy), October 2003. Poggianti, B. and Wu, H.: 2000, ApJ, 529, 157. Pozzi, F., et al.: 2003, MNRAS 343, 1348. Pozzi, F., et al.: 2004, ApJ 609 122. Puget, J.-L., et al.: 1996, A&A 308L, 5. Rodighiero, G., et al.: 2003, MNRAS 343 1155. Rodighiero, G., et al.: 2004, A&A 427, 773. Rowan-Robinson, M., et al.: 2004, MNRAS 351, 1290. Rowan-Robinson, M.: 2001, ApJ 549 745. Rush, B., Malkan, M. A., and Spinoglio, L.: 1993, ApJS, 89, 1. Sanders, D. B. and Mirabel, I. F.: 1996, AnRA&A 34, 749. Saunders, W., Rowan-Robinson, M., and Lawrence, A.: 1992, MNRAS 258, 134. Scott, S., et al.: 2002, MNRAS 331, 817. Serjeant, S., et al.: 2000, MNRAS 316, 768. Serjeant, S., et al.: 2001, MNRAS 322, 262. Serjeant, S., et al.: 2004, MNRAS 355, 813. Serjeant, S. and Harrison, D.: 2004, in M. Plionis (ed.) Proceeding of the “Multi-Wavelength Cosmology” Conference, Mykonos, Greece, June 2003 (Kluwer). V¨ais¨anen, P., et al.: 2002, MNRAS 337, 1043. V¨ais¨anen, P. and Johansson, P. H.: 2004, A&A 421, 821. V¨ais¨anen, P. and Johansson, P. H.: 2004, A&A 422, 453. Vaccari, M., et al.: 2005, MNRAS 358, 397. Willott, C., et al.: 2003, MNRAS 339, 297. Willott, C., et al.: 2004, ApJ 610, 140. Zehavi, I., Blanton, M. R., Frieman, J. A., et al.: 2002, ApJ 571, 172.

ISO’s CONTRIBUTION TO THE STUDY OF CLUSTERS OF GALAXIES∗ LEO METCALFE1,∗ , DARIO FADDA2 and ANDREA BIVIANO3 1 XMM-Newton

Science Operations Centre, European Space Agency, Villafranca del Castillo, P.O. Box 50727, 28080 Madrid, Spain 2 Spitzer Science Center, California Institute of Technology, Mail code 220-6, 1200 East California Boulevard, Pasadena, CA 91125, U.S.A. 3 INAF – Osservatorio Astronomico di Trieste, via G.B. Tiepolo 11, 34131, Trieste, Italy (∗Author for correspondence: E-mail: [email protected]) (Received 10 November 2004; Accepted in final form 20 December 2004)

Abstract. Starting with nearby galaxy clusters like Virgo and Coma, and continuing out to the furthest galaxy clusters for which ISO results have yet been published (z = 0.56), we discuss the development of knowledge of the infrared and associated physical properties of galaxy clusters from early IRAS observations, through the “ISO-era” to the present, in order to explore the status of ISO’s contribution to this field. Relevant IRAS and ISO programmes are reviewed, addressing both the cluster galaxies and the still-very-limited evidence for an infrared-emitting intra-cluster medium. ISO made important advances in knowledge of both nearby and distant galaxy clusters, such as the discovery of a major cold dust component in Virgo and Coma cluster galaxies, the elaboration of the correlation between dust emission and Hubble-type, and the detection of numerous Luminous Infrared Galaxies (LIRGs) in several distant clusters. These and consequent achievements are underlined and described. We recall that, due to observing time constraints, ISO’s coverage of higher-redshift galaxy clusters to the depths required to detect and study statistically significant samples of cluster galaxies over a range of morphological types could not be comprehensive and systematic, and such systematic coverage of distant clusters will be an important achievement of the Spitzer Observatory.

1. Introduction 1.1. ISO LOOKS DEEP The most strongly star forming galaxies are heavily dust obscured, and estimates of their star formation rates (SFR) made at visual wavelengths often fall one or two orders of magnitude below their true values. Frequently, very active star forming galaxies occur in associations or groups. At the same time, it is believed that in the dense environments of galaxy clusters the interactions of galaxies with each other, with the cluster tidal field, and with the intra-cluster medium (via ∗ Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, the Netherlands, and the United Kingdom) and with the participation of ISAS and NASA.

Space Science Reviews (2005) 119: 425–446 DOI: 10.1007/s11214-005-8065-y

 C

Springer 2005

426

L. METCALFE ET AL.

ram pressure) strip galaxies of their reserves of gas, and eventually suppress star formation. Much has already been learned about the evolution of the cosmic star formation rate in field galaxies from observations with the European Space Agency’s Infrared Space Observatory (ISO) satellite (Kessler et al., 1996). Thanks to deep surveys in the mid-infrared (MIR) and far-infrared (FIR) (e.g. Elbaz et al., 1999, 2002; Serjeant et al., 2000, 2004; Lari et al., 2001; Gruppioni et al., 2002; Metcalfe et al., 2003; Rodighiero et al., 2003; Sato et al., 2003; Kawara et al., 2004; RowanRobinson et al., 2004; and several others) conducted, respectively, with ISOCAM1 (Cesarsky et al., 1996) and ISOPHOT (Lemke et al., 1996), we now know that the comoving density of IR-bright galaxies has a very rapid evolution from z ∼ 0 to z ∼ 1. This evolution has been interpreted as the result of an increased rate of galaxy–galaxy interactions, coupled with an increase in the gas content of the galaxies (Franceschini et al., 2001). Dust obscuration plays an important role in concealing star formation in field galaxies. Its role in relation to the star formation activity of cluster galaxies is less known. Published ISO results to-date addressing the fields of galaxy clusters (Pierre et al., 1996; L´emonon et al., 1998; Altieri et al., 1999; Barvainis et al., 1999; Quillen et al., 1999; Soucail et al., 1999; Fadda et al., 2000; Duc et al., 2002, 2004; Metcalfe et al., 2003; Biviano et al., 2004; Coia et al., 2005a,b) concern a dozen clusters at z < 0.6, and point to an important role for dust, evolving with redshift, also in cluster galaxies. The fraction of IR-bright, star-forming cluster galaxies changes significantly from cluster to cluster, without a straightforward correlation with either the redshift, or the main cluster properties (such as the cluster mass and luminosity). The presence of dust in cluster galaxies may have hampered our efforts to fully understand the evolution of galaxies in clusters. Optical-band observations have provided a few, but very fundamental results, which cannot yet however be combined in a unique, well constrained scenario of cluster galaxy evolution.

1.2. SOME P ROPERTIES

OF

C LUSTER G ALAXIES

Perhaps the most fundamental phenomenology that all scenarios of cluster galaxy evolution (e.g. Dressler, 2004) must explain is the so called morphology density relation (hereafter MDR; Dressler, 1980), whereby early-type galaxies, i.e. ellipticals and S0s, dominate rich clusters, while late-type galaxies, i.e. spirals and irregulars, are more common in the field. The fact that early-type galaxies seem to reside in the cluster centres since z ≥ 1–2, while the colour-magnitude relation (CMR, Visvanathan and Sandage, 1977) remains very tight even at z ∼ 1 suggests that the MDR is established at the formation of galaxy clusters and that the early-type 1 Throughout this paper the ISOCAM filters having reference wavelengths 6.75 and 14.3 μm will respectively be referred to as the 7 and 15 μm filters. Both bandpasses are referred to as mid-infrared (MIR).

ISO’s CONTRIBUTION TO THE STUDY OF CLUSTERS OF GALAXIES

427

galaxies defining the CMR are uniformly old and passively evolving since their formation redshift, z f > 2 (e.g. Ellis et al., 1997). Similar conclusions are obtained by analysing the fundamental plane (FP; Dressler et al., 1987), relating basic properties of early-type galaxies (their effective radius, internal velocity dispersion, and effective surface brightness). The FP, like the CMR, still holds for early-type galaxies in z ∼ 1 clusters, and its scatter is similar to that seen in nearby clusters (van Dokkum and Stanford, 2003). However, these conclusions could only be true on average. Independent analyses suggest that at least part of the cluster galaxies have undergone significant evolution over the last 3–8 Gyr. First and foremost is the observational evidence for an increasing fraction of blue cluster galaxies with redshift, the so called ‘ButcherOemler’ (BO) effect (Butcher and Oemler, 1978, 1984; Margoniner et al., 2001). Approximately 80% of galaxies in the cores of nearby clusters are ellipticals or S0s, i.e. red galaxies (Dressler, 1980), but the fraction of blue galaxies increases with redshift. These blue galaxies are typically disk systems with ongoing star formation (Lavery and Henry, 1988), with spectra characterized by strong Balmer ˚ and no emission lines, and have been lines in absorption (typically, EW(Hδ) > 3 A) named ‘E+A’ (or also ‘k+a’) galaxies (Dressler and Gunn, 1983). Modelling of their spectra indicates that star formation stopped typically between 0.05 and 1.5 Gyr before the epoch of observation, in some cases after a starburst event (Poggianti et al., 1999, 2001). Similarly, the fraction of spirals increases with redshift (Dressler et al., 1997; Fasano et al., 2000; van Dokkum et al., 2001), at the expense of the fraction of S0s. Maybe, the colour and the morphological evolution of the cluster galaxy population are two aspects of the same phenomenon. Field spirals are being accreted by clusters (Tully and Shaya, 1984; Biviano and Katgert, 2004), and the accretion rate was higher in the past (Ellingson et al., 2001). It is therefore tempting to identify the blue galaxies responsible for the BO-effect in distant clusters with the recently accreted field spirals. Possibly, these evolutionary trends could be reconciled with the passive evolution inferred from the CMR and FP studies by taking into account the so-called ‘progenitor bias’ (van Dokkum et al., 2000), namely the fact that only the most evolved among cluster early-type galaxies are indeed selected in studies of the colour-magnitude and fundamental plane relations.

1.3. A HOSTILE E NVIRONMENT There is no shortage of plausible physical mechanisms that could drive the evolution of a galaxy in a hostile cluster environment. Among these, the most popular today are ram pressure, collisions, and starvation from tidal stripping (Dressler, 2004), all in principle capable of depleting a spiral of its gas reservoir, thereby making it redder and more similar to a local S0.

428

L. METCALFE ET AL.

Ram pressure from the dense intra-cluster medium can sweep cold gas out of the galaxy stellar disk (Gunn and Gott, 1972) and induce star formation via compression of the gas that remains bound to the galaxy. Collisions or close encounters between galaxies generate tidal forces that tend to funnel gas towards the galaxy centre (Barnes and Hernquist, 1991) eventually fueling a starburst that ejects gas from the galaxy. The cumulative effect of many minor collisions (named ‘harassment’, Moore et al., 1996), can lead to the total disruption of low surface brightness galaxies (Martin, 1999). The collision of a group with a cluster can also trigger starbursts in cluster galaxies, as a consequence of the rapidly varying tidal field (Bekki, 2001). Finally, the so called ‘starvation’ mechanism (Larson et al., 1980) affects a galaxy’s properties by simply cutting off its gaseous halo reservoir. This can occur because of tidal stripping, a mechanism effective in galaxy–galaxy encounters, but also when galaxies pass through the deep gravitational potential well of their cluster. The common outcome of all these processes is galaxy gas depletion, ultimately leading to a decrease of the star formation activity for lack of fuel, and, hence, to a reddening of the galaxy stellar population. However, some of these processes induce a starburst phase before the gas depletion, and some do not. To date, it remains unclear which physical process dominates in the cluster environment. Useful constraints can be obtained by finding where the properties of cluster galaxies change with respect to the field, since the different processes become effective at different galaxy or gas densities. Recently it has been found (Kodama et al., 2001; Lewis et al., 2002; G´omez et al., 2003) that a major change in the star-formation properties of cluster galaxies occurs in the outskirts of clusters (at ∼1.5 cluster virial radii). These results would seem to exclude ram-pressure stripping as a major factor in cluster galaxy evolution, since the density of the intra-cluster medium is too low in the cluster outskirts. Recent observations of high-z clusters seem to have complicated, rather than simplified, the issue of cluster galaxy evolution. A surprisingly high fraction of red merger systems has been found in these distant clusters (van Dokkum et al., 2001). The red colours of these merging galaxies and the lack of emission lines in their spectra suggest that their stellar populations were formed well before the merger events, but the occurrence of relatively recent starburst events in these galaxies is instead suggested by detailed analyses of their spectra (Rosati, 2004).

1.4. A N U NCLUTTERED VIEW A better understanding of the evolutionary processes affecting cluster galaxies can come from observations in the infrared. Dust, if present, is capable of obscuring most of a galaxy’s stellar radiation, making the observed galaxy red and dim at optical wavelengths, and affecting optical estimates of the galaxy star formation activity. The effects of dust can be particularly severe if the galaxy is undergoing a starburst (Silva et al., 1998), so that we might be missing a substantial part of the

ISO’s CONTRIBUTION TO THE STUDY OF CLUSTERS OF GALAXIES

429

evolutionary history of cluster galaxies by observing them at optical wavelengths. Since the dust-reprocessed stellar radiation is re-emitted at IR wavelengths, the IR luminosity is a much more reliable indicator of a galaxy’s star formation activity (Elbaz et al., 2002). The plan of this review is as follows: in Section 2 the development of knowledge of the infrared properties of galaxy clusters from early IRAS observations, through the “ISO-era” to the present is described. Section 2.1 considers the accumulation of data on the Virgo cluster, while Section 2.2 addresses other nearby clusters, e.g. the Fornax, Hydra, Coma and Hercules clusters. Section 2.3 discusses the significant progress that has been possible with ISO in the study of cluster galaxy properties out to moderate redshifts ( 1011 L  ). The sample consisted almost entirely of IR normal galaxies (L FIR < 1010 L  ) in contrast to the rather high percentage of LIRGs (20%) detected by IRAS in the field. Moreover, the lack of a strong correlation between galaxy HI content and IR emission led them to conclude that SFR is not enhanced by interaction with the ICM, and might even be quenched by it. On the contrary, the suppressed FIRto-radio ratio of spiral galaxies found in rich clusters with respect to poor clusters (Andersen and Owen, 1995) seemed to suggest that ram pressure enhances the radio emission in rich clusters while galaxy–galaxy interactions play a more important role in poor clusters where velocity dispersion, and so encounter velocities, are smaller. Quillen et al. (1999) observed 7 E+A galaxies plus one emission-line galaxy at 12 μm with ISOCAM. They found that E+A galaxies have mid- to near-IR flux ratios typical of early-type quiescent galaxies, while the emission-line galaxy had enhanced 12 μm emission relative to the near-IR. Galaxies with ongoing star formation have a different velocity distribution in the cluster from galaxies with stopped SF, suggesting that the ongoing infall of field spirals into the cluster potential may first trigger and then quench star formation. Further observations of Coma cluster galaxies in the MIR came from Boselli et al. (1998) and Contursi et al. (2001). They also observed the cluster A1367, located in the Coma supercluster, detecting, in total, 18 spiral/irregular galaxies in the MIR and FIR with ISO. Confirming results found in Virgo galaxies, these authors concluded that most IR-detected Coma galaxies display diffuse MIR emission unrelated to their Hα emission. The aromatic carriers are not only excited by UV photons, but also by visible photons from the general ISM. When the UV radiation field is too intense, it can even destroy the aromatic carriers, and overall the MIR emission is dominated by photo-dissociation regions rather than HII-like regions. A cold dust component was detected in all galaxies, at a temperature of ∼22 K, more extended than the warm dust. Only a very weak trend was found between the total dust mass and the gas 4 A262,

Cancer, A1367, A1656 (Coma), A2147, A2151 (Hercules), and Pegasus.

434

L. METCALFE ET AL.

content of the galaxies, even if some galaxies are very HI-deficient, and there was no detection of any relation between the MIR/FIR properties and the environment. All these results seemed to suggest very little (if any) dependence of the IR properties of galaxies on the environment. If anything, IR emission was thought to be quenched in the cluster environment. However, only nearby clusters had been studied, in which most of the galaxies are early type with little gas (and dust). With the launch of ISO it became possible for the first time to study galaxy clusters out to redshifts where a significant change in the composition of the cluster galaxy population had already been (Butcher and Oemler, 1984) or was soon to be (Dressler et al., 1999) observed in the optical. 2.3. GALAXY CLUSTERS

AT I NTERMEDIATE

REDSHIFT

ISO’s mid-infrared camera, ISOCAM, with its vastly improved sensitivity and spatial resolution with respect to IRAS, has successfully observed several galaxy clusters out to redshifts at which significant evolution might be expected to occur. At the same time, while studies of galaxies in nearby clusters are frequently done by targeting galaxies selected at other wavelengths, distant clusters can be completely surveyed due to their smaller angular size, producing an unbiased sample of infrared-emitting galaxies. Published ISO observations to date for clusters at redshifts above 0.1 (Pierre et al., 1996; L´emonon et al., 1998; Altieri et al., 1999; Barvainis et al., 1999; Fadda et al., 2000; Duc et al., 2002, 2004; Metcalfe et al., 2003; Biviano et al., 2004; Coia et al., 2005a,b) address seven clusters (see Table I)

Clustersa

TABLE I in the redshift range 0.17 < z < 0.6 studied with ISOCAM.

Cluster

z

A2218 A1689 A1732 A2390 A2219 A370 Cl0024+1654 J1888.16CL

0.175 0.181 0.193 0.23 0.228 0.37 0.39 0.56

a The

No. 15 μm-only sources 5 3 – 1 3 1 13 6

No. 7 μm-only sources

No. 7-and-15 μm sources

No. all sources

18 20 4 11 – 5 – –

4 9 0 3 – 0 – –

27 32 4 15 3 6 13 6

content of the columns in the table is as follows: name and redshift of the cluster; number, respectively, of 15 μm-only, 7 μm-only and 7-and-15 μm confirmed cluster sources detected in each case, and total number of MIR sources detected. Results are gathered from Coia et al. (2005a,b), Biviano et al. (2004), Metcalfe et al. (2003), Duc et al. (2002, 2004) and Fadda et al. (2000).

ISO’s CONTRIBUTION TO THE STUDY OF CLUSTERS OF GALAXIES

435

spanning the redshift range 0.17 < z < 0.6 and yield MIR data for around 110 cluster galaxies, slightly over 40 of these seen at 15 μm, and the rest only in the 7 μm bandpass. Almost half of the cluster galaxies detected at 15 μm prove to be LIRGs. (At the higher redshifts of the above sample only LIRGs fall above the sensitivity limit of the observations.) About 60% of these cluster galaxies were detected in observations originally intended to study distant field galaxies via the gravitational lensing amplification of the foreground clusters (Altieri et al., 1999; Barvainis et al., 1999; Metcalfe et al., 2003), and which, being generally very deep spatiallyoversampled measurements, were able to provide insights about the lensing cluster galaxy populations (Biviano et al., 2004; Coia et al., 2005a,b). Figure 2, taken from Coia et al. (2005b) is a V-band image of the z = 0.39 galaxy cluster Cl0024+1564 overlaid with contours of an ISO 15 μm map. The capacity of ISOCAM to detect and assign MIR counterparts unambiguously to numerous galaxies in the field is evident. The first published ISOCAM observations of a distant cluster were those of the z = 0.193 cluster A1732 by Pierre et al. (1996), which they observed at 7 and 15 μm over an 8 × 8 arcmin2 field. They found some evidence for a deficiency of spirals and star forming galaxies in the cluster, identifying only four cluster sources

Figure 2. From Coia et al. (2005b): Contours of a 15 μm map of the (gravitationally lensing) galaxy cluster Cl0024+1654 overlaid on a V-band FORS2 VLT image. 15 μm Sources numbered in order of increasing R.A. Dark circles identify spectroscopically confirmed cluster galaxies. Greek letters denote four prominent gravitational lensing arcs. North is up and East to the left, with the centre of the map falling at R.A. 00 26 37.5 and DEC. 17 09 43.4 (J2000).

436

L. METCALFE ET AL.

(at 7 μm, no cluster sources were detected at 15 μm) and 10 MIR galaxies in total in the field, most of them judged to be foreground. Nevertheless, these were the faintest MIR extragalactic sources reported up to that point and underlined the need for ultra-deep observations to detect cluster members at 15 μm. L´emonon et al. (1998) reported evidence for an active star-forming region in a cooling flow (later ‘cool-core’) from 7 and 15 μm observations of the inner square arcmin of the well known lensing cluster A2390 (z = 0.23), with an attendant SFR of as much as 80M yr−1 in the central cD. But this was later found to be compatible with non-thermal emission from a jet associated with the cD (Edge et al., 1999). The estimated cluster mass deposition rates in cooling flows have since been lowered by one or two orders of magnitude (B¨ohringer et al., 2002). A1689 (z = 0.181) was the first distant cluster for which detailed ISO observations were reported. Fadda et al. (2000) detected numerous cluster members (30 at 7 μm and 16 at 15 μm) within 0.5 Mpc of the cluster centre, and they found a correlation between the B-15 μm colour and cluster-centric distance of the galaxies. The 15 μm galaxies are blue outliers with respect to the colour/magnitude relation for the cluster and become brighter going from the center to the outer parts of the cluster. Coupled with the systematic excess of the distribution of the B-15 μm colours with respect to nearby clusters (Virgo and Coma), this suggested the existence of an IR analogue of the Butcher–Oemler effect in A1689 (see Figure 3). A follow-up optical study of these infrared galaxies (Duc et al., 2002) showed that the morphology of the 15 μm sources in A1689 is generally spiral-like, with disturbances reminiscent of tidal interactions. No LIRGs were found in A1689. The highest total IR luminosity found for a cluster galaxy was 6.2 × 1010 L  , corresponding

Figure 3. Fadda et al. (2000) found the [BT – 7 μm] colour distribution of A1689 cluster galaxies (filled circles) to be compatible with that of the nearby Virgo (crosses) and Coma (empty circles) clusters, but the [BT – 15 μm] colour distribution showed a systematic excess with respect to those nearer clusters. The cluster contains an excess of 15 μm sources relative to the field, suggesting that the environment of A1689 triggers starburst episodes in galaxies in the cluster outskirts that have similar IR luminosities and FIR/optical colours to those of field starburst galaxies.

ISO’s CONTRIBUTION TO THE STUDY OF CLUSTERS OF GALAXIES

437

Figure 4. This plot, taken from Duc et al. (2002), compares the A1689 SFR derived from optical [OII] measurements, and from ISO MIR measurements. It illustrates the limitations of purely optical indicators of star formation rates. A major part, at least 90%, of the star formation activity taking place in Abell 1689 is hidden by dust at optical wavelengths.

to a SFR of ≈11M per year. The median SFR for A1689, derived from 15 μm measurements, was 2M per year, while the median found from [OII] (optical) measurements was only 0.2M per year. This paper revealed the importance of IR observations in the study of starformation in clusters. About one-third of the 15 μm sources show no sign of star formation in their optical spectra. Moreover, comparing the star-formation estimates from IR and [OII] (see Figure 4), Duc et al. (2002) deduced that at least 90% of the star formation activity taking place in A1689 is obscured by dust. The ISO gravitational lensing survey programme (Metcalfe et al., 2003), and related work, led to deep observations of the core of several distant clusters. Large numbers of cluster galaxies were detected in the fields of A2218, A2390 and Cl0024+1654. So far, the cases of A2218 and Cl0024+1654 have been treated in dedicated papers. In the analysis of the galaxies in the field of A2218, a rich cluster at z = 0.175, Biviano et al. (2004) found nine cluster members at 15 μm inside a radius of 0.4 Mpc. In contrast to the case of A1689, which is at almost identical redshift (z = 0.181) and for which the median SFR is about 2M yr−1 and median L IR is around 1010 L  , only one of the A2218 MIR galaxies is a blue Butcher–Oemler galaxy. The MIR luminosity of A2218 galaxies is moderate and the inferred star

438

L. METCALFE ET AL.

TABLE II Summary of ISOCAM observations and results at 15 μm for five clustersa of galaxies. Sensitivity (μJy)

n scesc

LIRGsf

Cluster

z

Area (2 )

(5σ )

min.b

C

T

Rvir d

Mvir e

Obs

Exp

Cl0024 A370 A1689 A2390 A2218

0.39 0.37 0.18 0.23 0.18

37.8 40.5 36.0 7.0 20.5

140 350 450 100 125

141 208 320 54 90

13 1 11 4 6

35 20 18 28 46

0.94 0.91 1.1 1.62 1.63

6.42 5.53 5.7 20.35 18.27

10 1 0 0 0

– 8 1 1 1

a The table is adapted from Coia et al. (2005b), and the data originates from Metcalfe et al. (2003) for Abell 370, Abell 2218 and Abell 2390, Fadda et al. (2000) and Duc et al. (2002) for Abell 1689, and Coia et al. (2005b) for Cl 0024+1654. The content of the columns in the table are as follows: name and redshift of the cluster, total area scanned, sensitivity reported at the 5σ level, flux of the weakest reported source in μJy. Then number of cluster galaxies, total number of sources detected including sources without redshift and stars, virial radius of the cluster, virial mass, number of sources with L IR > 9 × 1010 L  detected and expected. The expected number of sources was obtained by comparison with Cl0024+1654 as described in the text. Virial radii and masses are from Girardi and Mezzetti (2001) and King et al. (2002). b Faintest source considered in publication. c Number of cluster sources, and total number of IR sources. d Cluster virial radius (h−1 Mpc). e Cluster virial mass (h−1 1014 M ).  f Number of LIRGs (or near LIRGs (L 10 IR > 9×10 L  )) detected vs. number expected if cluster were to be similar to Cl0024+1654.

formation rate is typically less than 1M yr−1 with a median L IR of only 6×108 L  . The absence of a MIR BO effect in A2218 might be a consequence of the small area observed, about 20.5 square arcmin (r < 0.4 Mpc), and yet the area studied for A1689 was not much larger, at about 36 square arcmin. Coia et al. (2005b) suggest that the difference between these two clusters may be traced to their having different dynamical status. The same sort of comparison can be drawn between the clusters Cl0024+1654 (z = 0.39) and A370 (z = 0.37). These two clusters at similar redshift and mapped in almost identical ways, exhibit very different numbers of LIRGs (Table II), probably, according to Coia et al. (2005b), because an ongoing cluster merger gives rise to enhanced star-forming activity in Cl0024+1654. As can be appreciated from Figure 5 taken from Coia et al. (2005b), the median infrared luminosity of ISO-detected cluster galaxies in Cl0024+1654 is around 1 × 1011 L  . Star formation rates derived from the 15 μm data range from 8 to 77M yr−1 , with median(mean) value of 18(30)M yr−1 . Because of the different sky areas mapped with ISO for several of the clusters discussed here, and the different sensitivities achieved, it is not straightforward

ISO’s CONTRIBUTION TO THE STUDY OF CLUSTERS OF GALAXIES

439

Figure 5. The infrared luminosity distribution for MIR galaxies in Cl0024+1654. The median value for L IR is ∼1.0 × 1011 L  . (From Coia et al., 2005b.)

to compare the results for different clusters. A useful approach is to compare the number of LIRGs detected in each cluster, since these IR-bright galaxies would be expected to be seen for any of the observations considered. Such a comparison, taken from Coia et al. (2005b), is presented in Table II. For each cluster an “expected” number of LIRGs is derived to test the hypothesis that the cluster is similar to the LIRG-rich cluster Cl0024+1654.5 The LIRG count in Cl0024+1654 is multiplied by the ratios of (a) virial mass per unit area of the cluster to that of Cl0024+1654, the square of the respective distances to the cluster and to Cl0024+1654, and the observed solid-angle for the cluster to that of Cl0024+1654. The resulting column of the table can then be compared with the column listing the actual observed number of LIRGs for each cluster. The two most distant clusters observed with ISO are Cl0024+1654 (z = 0.39, Coia et al., 2005b) and J1888.16CL (z = 0.56, Duc et al., 2004). These are among the deepest ISOCAM observations and could detect several cluster members. (For Cl0024+1654, 13 out of 35 sources found at 15 μm are spectroscopically confirmed to be cluster sources. For J1888.16CL, 6 out of 44 sources found at 15 μm are so confirmed.) A common feature of these two clusters is the high star formation rate inferred from their MIR luminosities. These two observations were also the most extended cluster maps performed in terms of absolute cluster area covered at 5 In fact, to avoid throwing away several sources close to the LIRG flux threshold of 1 × 1011 L and  thereby degrading the statistics of the comparison, the flux threshold for the comparison of detected luminous sources was set to 9 × 1010 L  .

440

L. METCALFE ET AL.

the cluster, 2.3 × 2.3 and 1.3 × 6 square Mpc for Cl0024+1654 and J1888.16CL respectively. This fact, and evolutionary effects detectable around z ∼ 0.5 (see Dressler et al., 1999), may explain the high IR luminosity of the 15 μm sources found in these clusters. In the case of J1888.16CL, Duc et al. (2004) estimate star-formation rates ranging between 20 and 120M per year. At least six galaxies belong to the cluster and have IR luminosities above 1.3 × 1011 L  . In Cl0024+1654, Coia et al. (2005b) report 10 sources brighter than 9 × 1010 L  . The star formation rates inferred from the MIR flux are one to two orders of magnitudes greater than those based on the [OII] flux (though in this case the comparison was only possible for the three sources for which [OII] data was available.) This is compatible with the result in A1689 (Figure 4) and implies similar dust extinction characteristics. Interestingly, the galaxies emitting at 15 μm appear to have a spatial distribution and a velocity dispersion slightly different from the other cluster galaxies. Galaxies in Cl0024+1654 are detected preferentially at larger radii, with the velocity dispersion of 15 μm sources being greater than that of the galaxies in the cluster. In J1888.16CL, Duc et al. (2004) estimate that to explain the number of sources detected on the basis of infall of galaxies from the field an infall rate of about 100 massive galaxies per 100 Myr is required, which seems unrealistic. Numerical simulations and X-ray observations show however that accretion onto clusters from the field is not a spherically symmetric process, but occurs along filaments or via mergers with other groups and clusters. One therefore cannot exclude the possibility that the LIRGs observed in these distant clusters belonged to such a recently accreted structure. An alternative possibility is that the collision with an accreted group of galaxies stimulated star formation in the galaxies of the group as a consequence of a rapidly varying tidal field (Bekki, 1999). This could be the case for Cl0024+1654 and A1689, clusters which show evidence of accreting groups of galaxies in their multi-modal velocity distributions. Cl0024+1654 is in the process of interacting with a smaller cluster.

2.4. A DIFFUSE INTRA-C LUSTER DUST COMPONENT? The hot intra-cluster material contains metals, and so is not entirely primordial. Might not the stars which produced the metals also have deposited dust in the ICM? The first to note that emission from intra-cluster material might be observable were Yahil and Ostriker (1973), based on a galactic dust-to-gas ratio and the observed intra-cluster gas. Ostriker and Silk (1973) and Silk and Burke (1974) developed expressions for the lifetime of dust in a hot intra-cluster medium. Pustilnik (1975), drawing upon contemporaneous reports of optical absorption in clusters, attributed it to dust and estimated that cluster emission at 100 μm would be in the 103 to 104 Jy range for six nearby clusters. Voshchinnikov and Khersonskii (1984) also attributed claimed reddenning of galaxies in distant clusters to dust absorption, and estimated

ISO’s CONTRIBUTION TO THE STUDY OF CLUSTERS OF GALAXIES

441

that the total FIR emission from the Coma or Perseus clusters should be 105 – 106 Jy (tens of Jy/arcmin2 ) in the 50–100 μm range. They estimated the sputtering lifetime of intra-cluster dust grains to be up to 108 years. Hu et al. (1985), noting that intra-cluster dust must be short-lived, predicted FIR dust emission of a few Jy per square degree, close to the IRAS limit, for a sample of X-ray luminous clusters. IRAS measurements failed to bear out even the most modest of the above predictions. Kelly and Rieke (1990) co-added IRAS scans across 71 clusters with 0.3 ≤ z ≤ 0.92 to arrive at an average 60 μm value for cluster emission of 26 ± 5 mJy per cluster, and 46 ± 22 mJy at 100 μm. Dwek et al. (1990) refined models of intra-cluster dust and its interactions and calculated an upper limit of 0.2 MJy/sr for dust-emission from the Coma cluster, consistent with IRAS observations. Then total cluster emission would not be more than a few Jy at the peak wavelength (around 100 μm). They concluded that dust in the cluster centre could not explain the visual extinction, nor could cluster galaxies or their halos. Dust in the outskirts could, if it were un-depleted. But they saw no mechanism for the production of such dust. Wise et al. (1993) analysed 56 clusters at 60 and 100 μm from clusters with a range of X-ray emission, and some without cDs. For the only two clusters (A262 and A2670) showing a far-infrared excess lacking an immediate explanation (in terms of point sources or cirrus) they concluded that the result was likely to be due to discrete sources in the clusters. Averaged over the sample as a whole there was evidence of excess FIR at the 2-σ level. No large FIR excesses associated with cooling flows were found. Bregman et al. (1990) looked for evidence of star formation in 27 cD galaxies. In half of their sample of X-ray-bright clusters they found IR, X-ray and blue luminosities to be comparable, consistent with dust grains heated by the X-ray emitting gas, thereby suggesting that dust cooling can compare with thermal bremsstrahlung as a cooling mechanism for the intra-cluster gas. Cox et al. (1995) studied a much larger sample of 158 Abell clusters, again at 60 and 100 μm, and after making a more rigorous correction for spurious sources due to galactic cirrus, they concluded that only about 10% of cD galaxies in rich clusters have significant FIR emission, but with luminosities 10 times greater than the X-ray luminosities produced in the cores of clusters, a condition which they therefore regarded as transient for any individual cluster. If the FIR emission comes from dust heated by the intra-cluster thermal electrons, significant dust sputtering is expected on timescales of several 108 years. Dust must then be replenished to account for continuous IR emission, presumably through the mechanisms discussed for stripping material from cluster galaxies (see Section 1). By the launch of ISO one could imagine a “life-cycle” of dust in a cluster, tracing the flow of material – gas and dust – out of infalling galaxies, the destruction of dust in the high-temperature intra-cluster medium, and its possible eventual redeposition, through the mechanism of cooling flows, into the cluster-dominant galaxies. But only upper-limits or occasionally, and with marginal significance,

442

L. METCALFE ET AL.

Figure 6. From Stickel et al. (2002): The overall zodiacal-light-subtracted surface brightness ratio I120 μm /I180 μm for A1656 (Coma) averaged over both scan position angles and all detector pixels.

global or average cluster FIR emission, could constrain scenarios for the role of dust in the physics of the clusters as a whole. Stickel et al. (1998, 2002) used ISOPHOT to observe extended FIR emission of six Abell clusters. Strip scanning measurements were performed at 120 and 180 μm. The raw profiles of the I120 μm /I180 μm surface brightness ratio including zodiacal light show a bump towards Abell 1656 (Coma), dips towards Abell 262 and Abell 2670, and are without clear structure towards Abell 400, Abell 496, and Abell 4038. After subtraction of the zodiacal light and allowance for cirrus emission, only the bump towards Abell 1656 (Coma) is still present (Figure 6). This excess of ≈0.2 MJy/sr seen at 120 μm towards Abell 1656 (Coma) is interpreted as thermal emission from intra-cluster dust distributed in the hot X-ray emitting Coma intra-cluster medium. The integrated excess flux within the central region of 10 to 15 diameter is ∼2.8 Jy. Since the dust temperature is poorly constrained only a rough estimate of the associated dust mass of MD ∼ 107 M can be derived. The associated visual extinction is negligible ( AV  0.1 mag) and much smaller than claimed from optical observations. No evidence is found for intra-cluster dust in the other five clusters observed. Quillen et al. (1999) suggested integrated emission from the cluster galaxies as the most likely source for the detected signal at the centre of Coma. Stickel et al. (2002) replied that if this was indeed the case, the same signal should have been detected in all clusters observed. The absence of any signature for intra-cluster dust in five clusters and the rather low inferred dust mass in Abell 1656 indicate that intra-cluster dust is probably not responsible for the excess X-ray absorption reported in cooling flow clusters (White et al., 1991). These observations thus represent a further unsuccessful attempt to detect the presumed final stage of the cooling flow material. This agrees with numerous previous studies at other wavelengths, while spectroscopic observations with ESA’s XMM-Newton X-ray observatory have shown that intra-cluster gas

ISO’s CONTRIBUTION TO THE STUDY OF CLUSTERS OF GALAXIES

443

does not cool beyond ∼1 keV (see, e.g., Molendi and Pizzolato, 2001), so dust deposition in cooling flows is not actually expected. Not surprisingly, then, further attempts by Hansen et al. (1999, 2000a,b) also failed to detect dust associated with the cooling flows presumed to exist at the centres of most galaxy clusters.

3. Conclusions Building upon the legacy of IRAS, ISO could consolidate a number of important fundamental conclusions about the properties of galaxies in the Virgo, Coma and other nearby galaxy clusters. The correlation between galaxy Hubble type and mid- and far-infrared properties was firmly established. A major new cold dust component was identified by ISOPHOT observations, which could not have been found by IRAS. Moderate resolution FIR spectroscopy was possible for Virgo cluster galaxies using LWS, establishing a correlation between the strength of the [CII] line and the FIR flux and a two order of magnitude difference in [CII] to near-IR ratio between early type galaxies and late (spiral) types. Mid-infrared emission (5– 18 μm) was found to correlate with star formation, but to trace it less faithfully when star formation rates become high enough for UV photons to disrupt the infraredemitting materials. However, the MIR emission traces well the FIR and bolometric emission. In general, the properties of galaxies in nearby clusters (z < 0.1) were found to exhibit little dependence on the cluster environment. ISO, for the first time, could extend mid-infrared observations to clusters beyond z = 0.1, and in so doing has detected over 100 galaxies in clusters in the redshift range 0.17 to almost 0.6. Although the collected observations on seven such clusters were rather heterogeneous, a number of important trends are found in the ISO results. There is a clear tendency for clusters at higher redshift to exhibit higher average rates of star formation in their galaxies, and numerous LIRGs have been observed in such clusters. There is tentative evidence for an association between the infrared luminosities found and galaxy infall from the field, but substantial evidence to link high levels of star-formation in cluster galaxies to the dynamical status of a cluster, and to interactions with other (sub-)clusters. It seems clear that star formation rates deduced for cluster galaxies from optical tracers often fall one to two orders of magnitude below rates derived from MIR emission levels, so that star formation in cluster galaxies may have been seriously underestimated in some cases, in the past. ISO found little evidence for widespread infrared emission from dust in the intra-cluster medium. Clearly, what ISO has not been able to supply has been systematic large area mapping of a substantial sample of galaxy clusters out to high redshift. Observing times to achieve such coverage with ISO would have been prohibitive, given the multi-position and heavily-overlapped rasters that would have been required. If galaxy evolution within clusters is to be further explored and related to galaxy

444

L. METCALFE ET AL.

evolution in the field, clusters of different masses and dynamical status must be studied systematically in the MIR and FIR out to well over 1 virial radius in order to understand how and why the IR properties of galaxies change from cluster to cluster, and from cluster to field. Large area coverage is important since it is known that significant modifications of the galaxy properties already occur in the outskirts of galaxy clusters (Kodama et al., 2001; Lewis et al., 2002; G´omez et al., 2003). Several programmes underway or planned with the Spitzer Space Observatory should thoroughly address these challenges.

References Altieri, B., Metcalfe, L., Kneib, J.-P., et al.: 1999, A&A 343, 65. Andersen, V., and Owen, F. N.: 1995, AJ 109, 1582. Barnes, J. E., and Hernquist, L. E.: 1991, ApJ 370, L65. Barvainis, R., Antonucci, R., and Helou, G.: 1999, AJ 118, 645. Bekki, K.: 2001, ApJ 546, 189. Bicay, M., and Giovanelli, R.: 1987, ApJ 321, 645. Biviano, A., and Katgert, P.: 2004, A&A 424, 779. Biviano, A., Metcalfe, L., McBreen, B., et al.: 2004, A&A 425, 33. B¨ohringer, H., Matsushita, K., Churazov, E., Ikebe, Y., and Chen, Y.: 2002, A&A 382, 804. Boselli, A., Lequeux, J., Contursi, A., et al.: 1997a, A&A 324, L13. Boselli, A., Tuffs, R., Gavazzi, G., Hippelein, H., and Pierini, D.: 1997b, A&A 121, 507. Boselli, A., Lequeux, J., Sauvage, M., et al.: 1998, A&A 335, 53. Boselli, A., Gavazzi, G., and Sanvito, G.: 2003a, A&A 402, 37. Boselli, A., Sauvage, M., Lequeux, J., Donati, A., and Gavazzi, G.: 2003b, A&A 406, 867. Boselli, A., Lequeux, J., and Gavazzi, G.: 2004, A&A 428, 409. Bregman, J. N., McNamara, B. R., and O’Connell, R. W.: 1990, ApJ 351, 406. Butcher, H., and Oemler, A.: 1978, ApJ 219, 18. Butcher, H., and Oemler, A.: 1984, ApJ 285, 426. Cayatte, V., van Gorkom, J. H., Balkowski, C., and Kotanyi, C.: 1990, AJ 100, 604. Cesarsky, C., Abergel, A., Agnese, P., et al.: 1996, A&A 315, L309. Clegg., P. E., Ade, P. A. R., Armand, C., et al.: 1996, A&A 315, L38. Coia, D., Metcalfe, L., McBreen, B., et al.: 2005a, A&A 430, 59. Coia, D., McBreen, B., Metcalfe, L., et al.: 2005b, A&A 431, 433. Contursi, A., Boselli, A., Gavazzi, G., Bertagna, E., Tuffs, R., and Lequeux, J.: 2001, AA 365, 11. Cox, C. V., Bregman, J. N., and Schombert, J. M.: 1995, ApJS 99, 405. Doyon, R., and Joseph, R. D.: 1989, MNRAS 239, 347. Dressler, A.: 1980, ApJ 236, 351. Dressler, A.: 2004, in Mulchaey, J. S., Dressler, A., and Oemler, A. (eds.), Clusters of Galaxies: Probes of Cosmological Structure and Galaxy Evolution, Carnegie Obs, Pasadena, CA. Dressler, A., and Gunn, J. E.: 1983, ApJ 270, 7. Dressler, A., Lynden-Bell, D., Burstein, D., et al.: 1987, ApJ 313, 42. Dressler, A., Oemler, A., Couch, W., et al.: 1997, ApJ 490, 577. Dressler, A., Smail, I., Poggianti, B. M., et al.: 1999, ApJS 122, 51. Duc, P.-A., Poggianti, B. M., Fadda, D., et al.: 2002, A&A 382, 60. Duc, P.-A., Fadda, D., Poggianti, B., et al.: 2004, astro-ph/0404183.

ISO’s CONTRIBUTION TO THE STUDY OF CLUSTERS OF GALAXIES

445

Dwek, E., Rephaeli, Y., and Mather, J. C.: 1990, ApJ 350, 104. Edge, A., Ivison, R., Smail, I., Blain, A., and Kneib, J.-P.: 1999, MNRAS 306, 599. Elbaz, D., Cesarsky, C. J., Chanial, P., et al., 1999, A&A 351, 37. Elbaz, D., Cesarsky, C. J., Chanial, P., et al., 2002, A&A 384, 848. Ellingson, E., Lin, H., Yee, H. K. C., and Carlberg, R. G.: 2001, ApJ 547, 609. Ellis, R., Smail, I., Dressler, A., et al.: 1997, ApJ 483, 582. Fadda, D., Elbaz, D., Duc, P.-A., et al.: 2000, A&A 361, 827. Fasano, G., Poggianti, B. M., Couch, W. J., et al.: 2000, ApJ 542, 673. Franceschini, A., Aussel, H., Cesarsky, C. J., Elbaz, D., and Fadda, D.: 2001, A&A 378, 1. Girardi, M., and Mezzetti, M.: 2001, ApJ 548, 79. G´omez, P. L., Nichol, R., Miller, et al.: 2003, ApJ 584, 210. Gruppioni, C., Lari, C., Pozzi, F., et al.: 2002, MNRAS 335, 831. Gunn, J. E., and Gott, J. R.: 1972, ApJ 176, 1. Hansen, L., Jørgensen, H., Nørgaard-Nielsen, H., Pedersen, K., Goudfrooij, P., and Linden-Vornle, M.: 1999, AA, 349, 406. Hansen, L., Jørgensen, H., Nørgaard-Nielsen, H., et al.: 2000a, A&A, 356, 83. Hansen, L., Jørgensen, H., Nørgaard-Nielsen, H., et al.: 2000b, A&A, 362, 133. H´eraudeau, Ph., Oliver, S., del Burgo, C., et al.: 2004, MNRAS 354, 924. Hu, E. M., Cowie, L. L., and Wang, Z.: 1985, ApJS 59, 447. Kawara, K., Matsuhara, H., Okuda, H., et al.: 2004, A&A 413, 843. Kelly, D. M., and Rieke, G. H.: 1990, ApJ 361, 354. Kenney, J. D., and Young, J. S.: 1986, ApJ 301, 13. Kenney, J. D., and Young, J. S.: 1989, ApJ 344, 171. Kessler, M. F., Steinz, J. A., Anderegg, M. E., et al.: 1996, A&A, 315, 27. King, L. J., Clowe, D. I., and Schneider, P.: 2002, A&A 383, 118K. Kodama, T., Smail, I., Nakata, F., Okamura, S., and Bower, R.: 2001, ApJ 562, 9. Lari, C., Pozzi, F., Gruppioni, C., et al.: 2001, MNRAS 325, 1173L. Larson, R. B., Tinsley, B. M., and Caldwell, C. N.: 1980, ApJ 237, 692. Lavery, R. J., and Henry, J. P.: 1988, ApJ 330, 596. Leech, K. J., Volk, H. J., Heinrichsen, I., et al.: 1999, MNRAS 310, 317. Leggett, S. K., Clowes, R. G., Kalafi, M., et al.: 1987, MNRAS 227, 563. Lemke, D., Klaas, U., Abolins, J., et al.: 1996, A&A 315, L64. L´emonon, L., Pierre, M., Cesarsky, C. J., et al.: 1998, A&A 334, L21. Lewis, I. J., Balogh, M., De Propris, R., et al.: 2002, MNRAS 334, 673. Margoniner, V., de Carvalho, R., Gal, R., and Djorgovski, S.: 2001, ApJ 548, L143. Martin, C. L.: 1999, ApJ 513, 156. Metcalfe, L., Kneib, J.-P., McBreen, B., et al.: 2003, A&A 407, 791. Molendi, S., and Pizzolato, F.: 2001, ApJ 560, 194. Moore, B., Katz, N., Lake, G., Dressler, A., and Oemler, A.: 1996, Nature 379, 613. Niklas, S., Klein, U., and Wielebinski, R.: 1995, A&A 293, 56. Odenwald, S. F.: 1986, ApJ 310, 86. Ostriker, J., and Silk, J.: 1973, ApJ 184, 113. Pierre, M., Aussel, H., Altieri, B., et al.: 1996, A&A 315, L297. Poggianti, B. M., Smail, I., Dressler, A., et al.: 1999, ApJ 518, 576. Poggianti, B. M., Bridges, T. J., Mobasher, B., et al.: 2001, ApJ 562, 689. Popescu, C. C., and Tuffs, R. J.: 2002, MNRAS 335, 41. Popescu, C. C., Tuffs, R. J., Volk, H. J., Pierini, D., and Madore, B. F.: 2002, ApJ 567, 221. Pustilnik, S. A.: 1975, SvAL 1, 49. Quillen, A. C., Rieke, G. H., Rieke, M. J., Caldwell, N., and Engelbracht, C. W.: 1999, ApJ 518, 632. Rodighiero, G., Lari, C., Franceschini, A., Gregnanin, A., and Fadda, D.: 2003, MNRAS 343, 1155.

446

L. METCALFE ET AL.

Rosati, P.: 2004, in Mulchaey, J. S., Dressler, A., and Oemler, A. (eds.), Clusters of Galaxies: Probes of Cosmological Structure and Galaxy Evolution, Carnegie Obs., Pasadena, CA, http://www.ociw.edu/ociw/symposia/series/. Rowan-Robinson, M., Lari, C., Perez-Fournon, I., et al.: 2004, MNRAS 351, 1290. Sato, Y., Kawara, K., Cowie, L.L., et al.: 2003, A&A 405, 833. Scoville, N. Z., Becklin, E. E., Young, J. S., and Capps, R. W.: 1983, ApJ 271, 512. Serjeant, S., Oliver, S., Rowan-Robinson, M., et al.: 2000, MNRAS 316, 768. Serjeant, S., Carramiana, A., Gonzles-Solares, E., et al.: 2004, MNRAS.tmp, 475. Silk, J., and Burke, J. R.: 1974, ApJ 190, 11. Silva, L., Granato, G. L., Bressan, A., and Danese, L.: 1998, ApJ 509, 103. Soucail, G., Kneib, J.-P., Bzecourt, J., et al.: 1999, A&A 343, L70. Stickel, M., Lemke, D., Mattila, K., Haikala, L. K., and Haas, M.: 1998, A&A 329, 55. Stickel, M., Klaas, U., Lemke, D., and Mattila, K.: 2002, A&A 383, 367. Tuffs, R. J., Popescu, C. C., Pierini, D., et al.: 2002, ApJS 139, 37. Tully, R. B., and Shaya, E. J.: 1984, ApJ 281, 31. van Dokkum, P. G., and Stanford, S. A.: 2003, ApJ 585, 78. van Dokkum, P. G., Franx, M., Fabricant, D., Illingworth, G. D., and Kelson, D. D.: 2000, ApJ 541, 95. van Dokkum, P. G., Stanford, S. A., Holden, B. P., et al.: 2001, ApJ 552, L101. Visvanathan, N., and Sandage, A.: 1977, ApJ 216, 214. Voshchinnikov, N. V., and Khersonskii, V. K.: 1984, AdSpR 3, 443. Voshchinnikov, N. V., and Khersonskii, V. K.: 1984, ApSS 103, 301. Wang, G., Leggett, S. K., Clowes, R. G., MacGillivray, H. T., and Savage, A.: 1991, MNRAS 248, 112. White, D. A., Fabian, A. C., Johnstone, R. M., Mushotzky, R. F., and Arnaud, K. A.: 1991, MNRAS 252, 72. Wise, M., O’Connell, R., Bregman, J., and Roberts, M.: 1993, ApJ 405, 94. Yahil, A., and Ostriker, J.: 1973, ApJ 185, 787. Young, E., Low, F. J., Soifer, B. T., et al.: 1984, ApJ 278, L75.