17. Unified Model: (Antonucci & Miller 1985, Urry & Padovani 1996) Active Galactic Nuclei X-ray: inverse Compton scattering (electron corona) Optical/UV: black body (direct) + emission lines IR: black body (reprocessed by dust)
18.
19. Current SFR exhaust gas in less than dynamical time-scale (i.e. one rotation period).
25. Accretion history (luminosity function of AGN) matches star formation history of the Universe. (Ebrero et al. 2009, Marconi et al. 2006) Sometimes both present SMBH growth formation BULGE Co-evolution! Physical connection between AGN and starburst!
54. From 60 m m and 850 m m surveys -> statistically homogeneous and complete
55. X-ray data for just 6 sources (half undetected) XMM-Newton sample (13 sources) Archive + own AO5 data Spitzer sample (13 sources) Archive data 9 sources in common
79. Templates – Type 1 AGN Lum. independent Richards et al. 2006 Lum. dependent Hopkins et al. 2007
80. Templates – Type 2 AGN Minimal SB contribution (Bianchi et al. 2006) NGC 5506 N H = 3x10 22 cm -2 NGC 4507 N H = 4x10 23 cm -2 Mrk 3 N H = 1.4x10 24 cm -2 NGC 3393 N H > 1x10 25 cm -2
81. Templates - Starburst NGC 5253 Young and dusty NGC 7714 Young and unobscured M82 Old SB IRAS 12112+0305 SB-dom. ULIRG
93. SB emission AGN fraction SED templates Theoretical IR model (RR00, F02, Verma et al. 2002) AGN fraction Theoretical IR model MIR analysis AGN1-L + SB1 AGN1 + SB1
94.
95. Star formation rates Star formation rate (MIR) IR Luminosity Star formation rate (MIR) Star formation rate (SED modelling) Farrah et al. 2002
114. Mergers? Differences hold for F02 complete sample! ¿Young galaxies during maximal episode of star formation? (strong SB and large amounts of gas) Farrah et al. 2002a High luminosity tail of ULIRG population
128. PAH emission at 7.7 mm very uniform (Sargsyan et al. 2009) : log SFR PAH = log ( n L n SB (7.7 m m)) – 42.57 ± 0.2
129. PAH emission very diluted in the spectra -> estimated through our SB template: f l SB (7.7 m m) = f 6 int (1- a 6 ) u l SB (7.7 m m)
130.
Editor's Notes
The “extreme” has been always a fascination for human beings. Science, as any other human activity, reflects this fascination. But the interest of science for extremes is not only a matter of this particular quality of human behaviour. Strange and extreme objects allow scientist to test theories beyond the “normal” conditions where they were established and to learn better how the nature works. In particular, Astronomy is a science of extremes. Most situations studied by astronomers are far beyond our daily experience. Nevertheless we are still interested in extremes. We search for the biggest and brightest objects in the Universe, we call them ultra-this or hyper-that, and all of them allow us a better understanding of the Universe. This thesis is dedicated to one family of these extreme objects: the Hyperluminous IR galaxies.
AGN and SB phenomena are behind HLIRG, so I'll present some brief notes about these two processes before show you the most remarkable properties of ULIRG and HLIRG previously known to this work. Next I'll present the sample we used for this thesis and the methodology employed for its study. Finally I'll show you the most remarkable results we have obtained, what can we conclude from this results and some suggestions on how to improve the results of this thesis.
We name as active galactic nucleus to an energetic phenomenon, not associated to starlight, happening in the central region of some galaxies. The luminosity associated to an AGN can be from ten to one thousand greater than the emission of its host galaxy, and roughly a ten percent of galaxies show AGN activity. There is an enormous number of AGN “flavours” accordingly to their observational properties but, for this work, there are two relevant classifications. We call quasi-stellar objects (or quasars) to the most luminous AGN, while we call seyfert galaxies to the less luminous AGN. This is an arbitrary, but useful, luminosity limit. Usually in quasars the light of the host galaxy is completely outshined. Accordingly to their optical spectra, we named as type 1 AGN those showing broad and narrow emission lines, while we call type 2 AGN to those showing only narrow lines.
One of the most characteristic features of AGN is that they emit significantly over the entire electromagnetic spectrum. In particular, type1 AGN emit equal power per decade of frequency. In this plot you can see also a relevant feature of type 2 AGN: they usually present absorption within the optical to X-ray range.
It is broadly accepted that the physical mechanism which originates the powerful emission of AGN is the accretion of matter in a supermassive black hole located in the center of the galaxy. The Unified model can explained both the diversity and the extreme luminosities of AGN. The central idea is that differences between AGN are not intrinsic, but mostly are caused by the angle of view of the observer. If the line of view cross the dusty torus, we see an obscured source and the region where broad emission lines are emitted is hidden. We see a type 2 AGN. If we can see directly the central engine of the AGN, we find no obscuration and the broad lines are observed. A radio jet is observed in ten to twenty percent of AGN, taking into account the strong radio emission of some AGN. Actually, we can associate each region of the SED with a particular structure of the unified model: the accretion disk surrounding the black hole is heated by viscous friction and emits in the optical and UV. Part of this radiation excite the gas clouds in the broad and narrow line regions, provoking the emission lines. Another fraction of the optical emission is absorbed by the dusty torus and then re-emited in the IR. The matter of the accretion disk close to the SMBH is highly ionized, forming a corona of high energy electrons which, through inverse compton scattering, convert the optical radiation into X-rays. As I said, we can observe absorption in the optical and X-ray range. In particular, when we cannot observe direct X-ray emission below 10-20 keV, we said that the source is Compton-Thick. Although this an obvious simplification valid a featureless SED, its is usefull to understand the basic physical processes happening in an AGN.
Starburst Galaxies show episodes of intense star formation, which means that, given the current star formation rate, the available gas to form stars will be finished in a relative small amount of time. The overall properties of starburst depend mainly on the age and luminosity of the burst. Metallicity is also a critical parameter for the starburst. SB are observed in several astronomical sources, all of them being strong IR emitters.
The optical and UV emission comes from starlight of the host galaxy and young hot stars formed in the burst. Most of this direct emission is absorbed by the large amounts of gas and dust needed to fuel the starburst and reemited in the IR. Radio emission is also significant, coming from synchrotron radiation emitted by accelerated electrons moving in the galactic magnetic field, and breemstralung from free electrons in regions of ionized hydrogen by the UV emission. X-ray emission, significantly less stronger than AGN, is originated in the last stages of stelar evolution (like supernova remnant or X-ray binaries).
But how AGN and SB are related. Sometimes they are both observed in the same galaxy, but there is a more fundamental link between them. ..... In the last decade, several observations strongly suggest that the growth of SMBH and the galaxy formation are closely related.
In order to study the AGN-galaxy coevolution we need sources where both AGN and SB are present. As we have seen, theoretical and observational results indicates that both phenomena happens in obscured environments, so we need penetrating radiation like X-rays and IR. Both energy ranges are needed to obtain a complete understanding of these processes. Fortunately, we live in an epoch where powerful IR and X-ray telescopes coexist.
We can follow different strategies to exploit the X-ray/IR synergy. The most straightforward method is to perform multiwalength surveys of the sky. Such surveys has been carried out in the last years. Alternatively, we can observe peculiar X-ray sources in the IR, or strong IR emitters in X-rays. We have followed the latter method In this thesis.
Luminous infrared galaxies were one of the most important discoveries of the IRAS surveys. They are galaxies where their bolometric output is strongly dominated by the IR emission. They show an IR luminosity greater than... and they seem to be a numerous constituent of the IR population. Accordingly to their IR luminosity we can define several “families”
First ULIRG were discovered by IRAS. These sources are rare in the local Universe, but they seem to be an important population in high redshift surveys. The mechanism which power these object was discussed since the beginning, but in the last decade a consistent paradigm has emerged, grounded in comprehensive observations from radio to X-rays. They are mergers of gas rich galaxies which trigger both intense star formation and AGN activity. Most are dominated by SB emission, but the presence of AGN and its relative contribution to the bolometric emission rise with the IR luminosity.
This paradigm is not so well-grounded in the case of HLIRG. As in ULIRG, AGN or SB (or a combination of both) must be the engine of these objects. However only one third of HLIRG seem to be in interacting systems, so they cannot be trivially classified as the high luminosity tail of ULIRG population. Since these objects seem to show powerful AGN and strong star formation, they are excellent laboratories to study the relationship between black hole growth an star formation.
The main goal of this work is to characterize the properties of HLIRG in several energy ranges and offer a consistent view of them. We will try to identify and unravel the AGN and SB emission using different methods, and we will try to reproduce the broadband emission of HLIRG using simple qualitative models. Finally we will check if HLIRG are an homogeneous population or can divided in different classes accordingly to their physical properties.
First of all, we need to assemble a proper sample of HLIRG in order to study their properties.
The sample assembled by Rowan-Robinson was the largest sample of HLIRG when we started this work. However it is largely inhomogeneus, because the sources come from very different surveys. Moreover, since the only criterion to select the sources is the luminosity limit, we expect some contamination of pure QSO. Starting from this sample, Farrah and colaborators built a new sample of HLIRG in a way independent of inclination, obscuration or AGN content. They selected HLIRG from flux limited IR surveys. Given the completeness and homogeinity of these surveys, the sample of Farrah is suitable to extract conclusions from the global population of HLIRG. However there were not available enough X-ray data to perform a proper X-ray study of the Farrah sample, so we assambled instead two different samples. The XMM sample includes sources with XMM-Newton observations (archival and private data), while the Spitzer sample includes sources with data from th IR spectrograph on-board Spitzer.
This diagram shows how the samples overlap between them. Note that the sample of Farrah is completely covered by our two samples.
Our multiwavelength estudy of HLIRG is divided in three steps: an X-ray spectral analysis, a MIR spectral analysis and finally, to obtain a global view of these sources, we studied their broadband spectral energy distributions, from radio to X-rays.
X-ray data were reduced following the standard procedure and using the standard software. Ten out of 13 sources were detected. These spectra are heterogeneous, both in quality and in properties. We tried to model both the AGN and SB emission: The main component of the AGN X-ray emission can be modelled as a powerlaw with a cut-off longward ~100 keV. This direct emission can be either absorbed or reflected (forming the compton hump and the 6.4 keV iron emission line). Frequently is also observed an excess in the soft part over the powerlaw. The exact origin of this soft excess is still discussed, but given the resolution of our spectra, we can just model this component with a thermal emission model. SB emission can be modelled either by a flat powerlaw or through a thermal emission model.
We extracted the IRS low resolution spectra starting from pipeline products. We followed the standard procedure using the standard reduction software for Spitzer. Given the low dispersion of pure SB and AGN emission within the 5-8 microns range, we can apply a simple model to disentangle the AGN and SB emission in HLIRG. This method has been already successfully used in ULIRG in several works.
The model is composed by an AGN component (an absorbed powerlaw) and an SB template. The obscuration for this component is already included in the template.
The SED we used to this analysis were built using data from several astronomical databases and from the literature. In addition, we included the X-ray and MIR data already presented. Since we just need a qualitative characterization of the AGN and SB emission, we modelled the SED with observational templates. In some sources the SED cannot be modelled with the addition of AGN and SB components, so we used instead composite templates (from sources where AGN and SB are both present).
We used to type 1 AGN templates: a luminosity independent template, and a luminosity dependent one, where the optical-to-X-ray ratio increases with the bolometric luminosity.
As type 2 AGN template we used the SED of four well observed sources, selected from a sample with minimal starburst contribution. The main difference between them is the obscuration level.
To model the SB emission we use 4 templates, coming from sources with different ages, luminosities and content of dust.
We selected four well observed sources where both AGN and SB are present as composite templates. They have different AGN relative contribution.
This plot shows the hard X-ray versus the FIR luminosity. This line represents the expected X-ray luminosity of a starburst galaxy given its FIR luminosity. And this area is the same for AGNs, taking account the intrinsic dispersion in the spectral energy distribution of AGNs. We have included a comparison sample of high redshift HLIRGs, and a sample of ULIRG studied in X-rays. These are AGN dominated objects, and these are SB dominated. Most of our HLIRGs, and the high-z sample, lie in the AGN-zone, although they are systematically underluminous in X-rays with respect to the mean value. The non detected sources are consistent with starburst luminosity. There are also two composite objects, that is, nor SB or AGN alone can explain their X-ray luminosites. Among these SB dominated HLIRGs are the sources optically classified as starburst, but also two type 1 QSO. They are quasars extremely underluminous in X-rays.