Presentation is loading. Please wait.

Presentation is loading. Please wait.

The Evolution of AGN Obscuration

Similar presentations


Presentation on theme: "The Evolution of AGN Obscuration"— Presentation transcript:

1 The Evolution of AGN Obscuration
Ezequiel Treister (ESO) Meg Urry (Yale) Julian Krolik (JHU) Shanil Virani (Yale) Eric Gawiser (Rutgers)

2 Active Galactic Nucleus (AGN)
supermassive black hole actively accreting matter x galaxy luminosity from few lt-hrs Host Galaxy AGN are explained by accretion of matter into a super-massive black holes. Masses of the black hole are 10^6-10^9 and the luminosity of the central parsec can be larger than the whole galaxy

3 The AGN Unified Model blazars, Type 1 Sy/QSO broad lines
AGN unification paradigm can explain the observed differences between type 1 and type 2 AGN. Urry & Padovani, 1995

4 The AGN Unified Model radio galaxies, Type 2 Sy/QSO narrow lines
Urry & Padovani, 1995

5 Supermassive Black Holes
Many obscured by gas and dust How do we know that? Local AGN Unification Explain Extragalactic X-ray “Background” How do we know that there are many obscured BHs? Locally the obscured/unobscured ratio is ~3:1 More importantly, this is the case at all redshifts because of the X-ray background Credit: ESO/NASA, the AVO project and Paolo Padovani

6 Observed X-ray “Background”
We know now that the X-ray background is just collective emission from previously unresolved sources, 90% of them AGN. The X-ray background spectrum is very steep, with a peak at ~30 keV. This is good indication that obscured AGN should outnumber the unobscured ones. Frontera et al. (2006)

7 AGN in X-rays Increasing NH Photoelectric absorption
affect mostly low energy emission making the observed spectrum look harder. The reason is that obscuration makes the X-ray spectrum appear harder. In these units, an unobscured AGN is roughly flat, while as we add more and more obscuration (parameterized as a neutral hydrogen column density) the observed spectrum appears harder and harder, thus closer to the observed X-ray background. Increasing NH

8 X-ray Background XRB well explained using a combination of obscured and unobscured AGN. # w NH > 1023 cm-2 still uncertain. With this idea in mind, for almost 20 years now people tried to use a combination of obscured and unobscured AGN to fit the spectral shape and intensity of the XRB. The most important parameter is of course the assumed fraction of obscured AGN. Here I show an example for our later work, in this case assuming a fraction of obscured to unobscured AGN of ~3:1. Setti & Woltjer 1989 Madau et al. 1994 Comastri et al. 1995 Gilli et al. 1999,2001 And others… Treister & Urry, 2005

9 Multiwavelength Surveys
Hard X-rays penetrate most obscuration Energy re-radiated in infrared High resolution optical separates host galaxy, constrains redshifts E-CDF-S Chandra Coverage 0.25 deg2 FX=~10-16erg cm-2s-1 ~800 X-ray sources COSMOS XMM+Chandra Coverage 2 deg2 FX=~10-15erg cm-2s-1 ~1100 X-ray sources

10 Extended Chandra Deep Field South (optical)

11 Extended Chandra Deep Field South (x-rays)

12 Hardness Ratio vs. Luminosity
More unobscured AGN at high X-ray luminosity? Treister et al in prep.

13 NH vs. Luminosity In general good agreement between optical and X-ray
classification. Treister et al in prep.

14 NH vs. Redshift Obscuration increases with redshift?
Or selection effect due to X-rays K correction? Treister et al in prep.

15 HST ACS color image (0.3% of GOODS)
This is a small fraction of the GOODS field, a deep multiwavelength survey with Chandra, Spitzer and HST Showing three examples of X-ray sources, one faint in the optical and another completely invisible.

16 HST+Spitzer color image (0.3% of GOODS)
However, all these sources are bright in the IR, as expected since most of the absorbed energy is re-emitted at these wavelengths.

17 IR Fraction vs Flux AGN are bright IR sources! Treister et al 2006
Indeed, by studying the fraction of sources classified as AGN in X-rays we found that this fraction depends strongly on the IR flux, going from ~50% to 5%. This results holds even correcting for the AGN missed in X-rays. Treister et al 2006

18 IR Luminosity Distribution
On average, AGN are ~10x brighter than normal galaxies For fainter AGN, the host galaxy makes a significant contribution A similar result is found while studying the IR luminosity of X-ray selected AGN. However, by computing the luminosity distribution for a random sample of normal Galaxies with the same redshift distribution we found significant overlap. This indicates that in particular for the fainter AGN the host galaxy makes a significant contribution to the IR light. Treister et al 2006

19 X-Ray to mid-IR Ratio Significant separation between obscured and unobscured sources. Unobscured QSO Template Smaller separation at shorter wavelength (host galaxy) and largest at longer wavelength (self-absorption). Obscured QSO Template A similar result is found while studying the IR luminosity of X-ray selected AGN. However, by computing the luminosity distribution for a random sample of normal Galaxies with the same redshift distribution we found significant overlap. This indicates that in particular for the fainter AGN the host galaxy makes a significant contribution to the IR light. Treister et al in prep.

20 X-Ray to mid-IR Ratio More separation at lower luminosities.
Change in the opening angle with luminosity? Larger opening angle, ie less self-absorption. A similar result is found while studying the IR luminosity of X-ray selected AGN. However, by computing the luminosity distribution for a random sample of normal Galaxies with the same redshift distribution we found significant overlap. This indicates that in particular for the fainter AGN the host galaxy makes a significant contribution to the IR light. Treister et al in prep.

21 Infrared Background AGN (+ host galaxy) contribute ~3-10% of the total extragalactic background light Even though AGN are bright IR sources, they only contribute ~3-10% of the IR background depending on wavelength. So, in contrast to the X-ray background, AGN are a small fraction of the total extragalactic IR light. Treister et al 2006

22 NH Distribution: What do we know so far?
More obscured AGN at low luminosity (Steffen et al. 2003, Ueda et al. 2003, Barger et al. 2005, Akylas et al. 2006) More obscured AGN at high-z? (Ueda et al. 2003: No, La Franca et al. 2005: yes, Ballantyne et al. 2006: yes) Problems: Low number of sources Selection effects: - X-ray selection (missed obscured sources) - Optical incompleteness (no redshifts) - X-ray classification: “K correction” But, what do we really know about the fraction of obscured AGN? Thanks to X-ray surveys, it is now well established that there should be relatively more obscured AGN at lower luminosities It is possible that there is also evolution in the fraction of obscured AGN, however even X-ray surveys have several problems: Because of these problems, we decided to generate the largest possible sample of X-ray selected AGN in order to study the fraction of obscured AGN.

23 Meta-Survey 7 Surveys, 2341 AGN, 1229 w Ids 631 Obscured
(no broad lines) 1042<Lx<1046, 0<z<5 This sample is ~4X larger than previous studies and includes both large area/shallow sources in order to properly sample the high luminosity sources and deep surveys for the low luminosity/high redshift AGN. This is the X-ray sensitivity function, which only tells you part of the story, because since redshifts are required to calculate luminosities and identify sources. Hence, the optical flux should be considered as well. Treister & Urry, 2006

24 Total effective area of meta-survey
This is our sensitivity plot that we will use to correct for the effects of incompleteness. As you can see the area covered is now a function of both the X-ray flux and the optical magnitude. Treister & Urry, 2006

25 Ratio vs Redshift Treister & Urry, 2006
First, we studied the dependence of the fraction of obscured AGN on redshift. Our full sample has a completeness of ~50%. Data points show the “raw” data, not corrected for any selection effects, while the solid lines shows the expectation for a intrinsically-constant fraction if incompleteness is taken into account. Treister & Urry, 2006

26 Ratio vs Redshift Treister & Urry, 2006
Same plot but increasing the completeness by restricting our sample. As you could expect, the observed fraction gets flatter with higher completeness. Same with the expected curve. Treister & Urry, 2006

27 Ratio vs Redshift Treister & Urry, 2006
This is now for a very high completeness of ~90%. Here it starts to appear like the observed fraction actually increases with redshift. Treister & Urry, 2006

28 Ratio vs Redshift Key input: Luminosity
dependence of obscured AGN fraction. In order to correct for the effects of incompleteness, we divide the observed data points by the expectation for a constant fraction, so that the bottom panel shows the fraction of obscured AGN relative to a constant value. 3 reasons for decline: Lose absorbed X-ray sources (K correction helps you) Lose redshifts because of optical magnitude most important effect: luminosity dependence. See also: La Franca et al. 2005 Ballantyne et al. 2006 Akylas et al. 2006 Treister & Urry, 2006

29 Ratio vs Luminosity Treister & Urry, 2006
Hence, it makes sense to study the dependence of the fraction of luminosity. As you can see, a simple linear parameterization on log L does a very good job explaining the observational results. Treister & Urry, 2006

30 Ratio vs Luminosity Treister & Urry, 2006
For example, the latest X-ray background fitting of Gilli et al. which includes a more complex parameterization fails to explain the observed data. Treister & Urry, 2006

31 Ratio vs Luminosity Treister & Urry, 2006 Hasinger et al.
Even if we add recent data from a similar sample by Gunther Hasinger there is a clear discrepancy at high luminosities. So, it is clear that there is a luminosity dependence of the obscured AGN fraction. Now, we would like to understand why this is the case. Hasinger et al. Treister & Urry, 2006

32 The AGN Unified Model Urry & Padovani, 1995
The simplest unified model tells you that the near and mid IR emission comes from the torus regions closer to the BH. In this model, the relative fraction of IR to bolometric emission is a direct measure of the fraction of sky covered by the torus, and hence of the fraction of obscured AGN. Urry & Padovani, 1995

33 A Changing Torus? Granato & Danese Torus model ~2x
A change in the IR/NUV flux ratio may indicate a change in torus geometry. Granato & Danese Torus model ~2x If the change in the fraction of obscured AGN is due to a change in the opening angle, then most IR torus models (in this case from the model of Granato & Danese) predict that the IR emission is scaled proportionally.

34 Torus Structure Sample
Completely unobscured AGN Narrow redshift range, 0.8<z<1.2 Wide range in luminosity Data at 24 µm from Spitzer High L SDSS DR5 Quasar sample 11938 quasars, 0.8<z<1.2 192 with Spitzer 24 µm photometry 157 of them with GALEX UV data Low L GOODS: North+South fields 10 unobscured AGN All Spitzer 24 µm photometry 8 with GALEX UV data In order to study this we constructed a large sample of unobscured AGN in a narrow redshift range in order to look for a dependence of the fraction of IR emission on luminosity. Mid L COSMOS 19 unobscured AGN 14 Spitzer 24 µm photometry All with GALEX UV data

35 Torus Structure Bolometric luminosity constructed from NUV to mid-IR.
No change in NUV/Bol ratio with luminosity! The first result is that at least the fraction of light at NUV wavelengths, coming directly from the accretion disk does not depend on luminosity. Treister et al. 2008

36 Torus Structure Change in 24 µm/Bol ratio with luminosity!
Lower ratio at high L Consistent with larger opening angles at higher luminosities. However, we found a clear change, by a factor of ~3x in the fraction of IR light measured at rest-frame 12 microns as a function of luminosity. This change goes in the right direction, namely low luminosity AGN have a higher fraction of IR to bolometric luminosity, which can be interpreted as a smaller opening angle. Now, using torus models we can convert this fraction of IR light into a obscured fraction and compare with the results from X-ray surveys. Treister et al. 2008

37 Fraction of Obscured AGN
Similar luminosity dependence as found on X-ray surveys. Higher values from fIR/fbol method. Obscured AGN missed by X-ray surveys. This is our main result from this study. In general there is a good agreement between the obscured AGN fraction from X-ray surveys and from our method. However, our results are systematically higher. Can this further indicate that there is a large fraction of obscured AGN missed by X-ray surveys? Can these sources be detected in the mid-IR? Treister et al. 2008

38 Compton Thick AGN Defined as obscured sources with NH>1024 cm-2.
Very hard to find (even in X-rays). Observed locally and needed to explain the X-ray background. Number density highly uncertain. High energy (E>10 keV) observations are required to find them. There is a population that was completely missed in all these studies. Those are the most obscured sources. So obscured that are not detected even in X-rays. Those are the CT AGN. Even though we don’t know the exact density of these sources, they are required in large numbers by XRB synthesis models and are observed locally.

39 INTEGRAL Survey PIs: Meg Urry, Shanil Virani, Ezequiel Treister
Exp. Time (Msec): 0.7 (archive)+1.5 (2005)+ 1 (2008). Deepest extragalactic INTEGRAL survey Field: XMM-LSS (largest XMM field) Flux limit: ~4x10-12 ergs cm-2 s-1 (20-40 keV) Area: ~1,000 deg2 Sources: ~20 Obscured AGN: ~15 (~5 Compton-thick) In order to study directly this population, at least at low redshifts, three years ago we started a deep survey at high energies with INTEGRAL.

40 INTEGRAL Mosaic (2.2 Ms) Significance Image, 20-50 keV MCG-02-08-014
These are the first results. We found a total of 10 sources, including the “famous” CT AGN NGC1068. But, we recently got a nice surprise when we got the first 2 Mseconds of data. We detected in INTEGRAL a source not detected in ROSAT, MCG Significance Image, keV

41 MCG-02-08-014 z=0.0168 Optical class: Galaxy IR source (IRAS)
Radio source (NVSS) Narrow line AGN (based on OIII emission) No soft X-rays (ROSAT) Good CT AGN candidate This source was only classified spectroscopically as an AGN by the SDSS because narrow OIII was detected. It is also a radio source, but not detected in soft X-rays, hence is a very good CT AGN candidate in the nearby Universe.

42 INTEGRAL AGN logN-logS
In order to compare our results with the expectations from XRB models, The first tool for that is the logN-logS plot. Here we show it including the results from large area surveys (almost all-sky) at these energies. In general there is a good agreement. Data points from Beckmann et al. 2006 Treister et al, submitted

43 Space Density of CT AGN X-ray background does not constrain density of CT AGN We can also construct the logN-logS for CT AGN only. Here we found that the CT AGN fraction is ~4x lower than what expected by models. This is because the XRB does not constrain the density of CT AGN. Treister et al, submitted

44 CT AGN and the XRB XRB Intensity Treister & Urry, 2005 HEAO-1 +40%
Indeed, for a given assumed XRB intensity, the CT AGN density is degenerated with the normalization of the Compton reflection component, which contributes most of the X-ray emission on CT AGN. Treister et al, submitted

45 CT AGN and the XRB XRB Intensity HEAO-1 Original Treister & Urry, 2005
Gilli et al. 2007 Treister & Urry, 2005 Different models assumed different normalizations for the XRB, but now thanks to the recent INTEGRAL and Swift results, it is clear that the original HEAO-1 intensity, with a 10% uncertainty was correct, as assumed by the models of Gilli et al published this year. Treister et al, submitted

46 CT AGN and the XRB XRB Intensity HEAO-1 Original Treister & Urry, 2005
Gilli et al. 2007 Treister & Urry, 2005 Most likely solution However, now we can constrain the density of CT AGN directly from the INTEGRAL observations, finding that the most likely solution has a CT AGN fraction ~4x lower than previously expected. CT AGN Space Density Treister et al, submitted

47 X-ray Background Synthesis
So, we can now construct XRB population synthesis models completely constrained by observations, finding that most of the emission comes from sources with relatively low luminosities, 10^43-44 which corresponds to bright seyferts or faint quasars.

48 NH Distribution So, we can now construct XRB population synthesis models completely constrained by observations, finding that most of the emission comes from sources with relatively low luminosities, 10^43-44 which corresponds to bright seyferts or faint quasars.

49 Summary AGN unification can account well for the observed properties of the X-ray background. AGN are luminous infrared sources, but contribute ~5% to extragalactic infrared background The obscured AGN fraction decreases with increasing luminosity. Ratio of IR to Bolometric luminosity in unobscured AGN suggest this is due to a change in opening angle. The obscured AGN fraction increases with redshift as (1+z)0.4. Survey at high energies starts to constrain the spatial density of CT AGN. Most important conclusions: Fraction of obscured AGN depends on both luminosity and redshift Luminosity dependence due to a change in opening angle AGN are luminous IR sources but contribute only ~5% of the IR background. We are starting to constrain the density of CT AGN using observations at high energies.

50

51 SBH Spatial Density Natarajan & Treister, in prep

52 XRB Intensity

53 Ratio vs Luminosity Treister & Urry, 2006

54 Luminosity vs Redshift
Treister & Urry, 2006

55 Incompleteness Effects
Treister & Urry, 2006

56 redshifts of Chandra deep X-ray sources
GOODS-N Model R<24 R<24 redshifts of Chandra deep X-ray sources GOODS-N Barger et al. 2002,3, Hasinger et al. 2002, Szokoly et al. 2004

57 Dust emission models from Nenkova et al. 2002, Elitzur et al. 2003
Simplest dust distribution that satisfies NH = 1020 – 1025 cm-2 3:1 ratio (divide at 1022 cm-2) Random angles  NH distribution NH=1020cm-2 NH=1022cm-2 NH=1025cm-2

58 Space Density of CT AGN Strong degeneracy between reflection
component and number of CT AGN. Treister et al, submitted


Download ppt "The Evolution of AGN Obscuration"

Similar presentations


Ads by Google