Hien D. Tran W. M. Keck Observatory

Hien D. Tran W. M. Keck Observatory
The Unification Model of Active Galaxies: Implications from Spectropolarimetric Surveys Hien D. Tran W. M. Keck Observatory Good morning, and thank you for having me here. In my talk today, I hope to present to you how spectropolarimetric observations of AGNs have helped us better understand the nature of these objects, and in particular, how they help provide some unique insights and implications on the unification model.

Optical Spectra of Various Types of AGNs
As you know AGNs come in various types, and today I will focus only on the seyfert galaxies. Bill Keel

Seyfert 1 & 2 Optical Spectra
Narrow forbidden lines: eg. [O III], [N II] Broad permitted lines: Ha, Hb They come mainly in two types, defined by their optical spectra. Seyferts type 1 show not only the narrow forbidden lines like oiii and nii but also broad balmer permitted lines like halpha and hbeta. Type 2 seyferts on the other hand, show only the narrow emission lines in their spectra. Bill Keel Narrow lines only

Hidden Sey1 Nucleus Inside a Sey2 Galaxy
NGC 1068 Miller & Antonucci (1983) Antonucci & Miller (1985) When Miller & Antonucci looked at the prototypical seyfert 2 galaxy NGC 1068 some three decades ago now, they made a remarkable discovery. They found that in polarized light, the spectrum appeared just like that of a seyfert 1. This strongly suggested that the two types of seyferts are intrinsically the same. Miller, Goodrich & Mathews (1991) Bill Keel

NGC 1068 Hidden Broad-Line Region (HBLR)
Here’s a more recent observation I made from Keck of ngc1068, showing the typical seyfert 2 spectrum in direct light on the top, and the polarized flux spectrum on the bottom, that shows the narrow lines, as well as the broad lines typical of a seyfert 1. These broad lines are easily visible only in polarized light, and thus they’ve come to be known as the hidden broad-line region (or HBLR) Hidden Broad-Line Region (HBLR)

Mrk 348: Another “HBLR Sey2”
HBLR = Hidden Broad-Line Region Typical Seyfert 2 (S2) spectrum Many more examples of such hidden seyfert 1s have been seen since then. This is another example in Mrk 348. Appears like a Seyfert 1 (S1)!

AGN Unification Paradigm: Obscuring Torus
Sey 1 view Sey 2 view Nuclear Engine + Broad-line region (BLR): broad Ha, Hb Obscuring torus Narrow-line region (NLR): [O III], [N II] Thus the unification model of Seyfert was born, which proposed that the two types of seyferts are the same kind of object, and they appear differently only because of their orientation relative to our line of sight. This diagram illustrates the basic components of the model that can elegantly explain the polarimetric observations. In the core is the central agn engine, presumably powered by a supermassive black hole with the ionizing continuum and broad-line region. Surrounding it is an obscuring torus which blocks the light to the central nucleus. Much further out, is the narrow-emission line region which is largely unobscured. When viewed along the pole of the torus, the object would appear as a seyfert 1 galaxy, but a view from the side would show an appearance of a seyfert 2 galaxy. Electrons or dust along the poles, however, could scatter and polarize the light from the obscured central region, and therefore in polarized light the same object would appear as a seyfert 1. Scattering e-, dust Urry & Padovani

NGC 5252 Ionization Cones Morse et al. 1998
This obscuration/reflection unification model has enjoyed enormous popularity and has been very well tested, receiving a lot of support over past decades since its proposal, and I show here an example of just one of the many observations. This is Ngc5252, which shows spectacular ionization cones that are just what you would expect from the unification model. Morse et al. 1998

Puzzling Differences between Sey1 & Sey2
S2s more likely to be in found in hosts with enhanced star formation (Maiolino et al. 1995; Gu et al. 1998). S2s tend to be in hosts with higher frequency of companions (de Robertis, Yee, & Hayhoe 1998; Dultzin-Hacyan et al. 1999). S2 hosts tend to show richer dust features (Malkan, Gorjian, & Tam 1998). Some studies, however, found some puzzling differences between the two types of seyferts that appeared inconsistent with the UM. For example… the hosts of seyfert 2s have been found to have more enhanced star formation, higher frequency of companions, or richer dust features than the hosts of seyfert 1s. These supposedly orientation-independent properties of the hosts are not expected to be different under the um.

Spectropolarimetric Study of Seyfert Galaxies
General applicability of the Unification Model (UM)? What determines the detectability of hidden broad lines? Is there a connection to AGN power? How are the HBLR S2s different from non-HBLR S2s? Correlation of presence of hidden broad lines with observational properties? Spectropolarimetric survey of complete samples of Seyfert 2s to look for hidden broad-line regions (HBLRs) And so…. there are a number of questions with related to the um that we’d like to understand. For example: What is the general applicability of the um ? Do all seyfert 2s contain a hidden seyfert 1 nucleus? Not all seyfert 2s were observed to show polarized hidden broad lines.. Why not? What determines the detectability of the hblr? Is there a connection to agn power? How are the hblr s2s different from the non-bhlr s2s.. we’d also like to see if there are any kinds of correlations between the presence of hblr with other observational properties.. In order to address these questions, we need to have a large survey of seyfert 2s aimed at the search for hblrs.

Brief History of Spectropolarimetric Seyfert 2 Surveys
Young et al. (1996) sample: 24 “warm” IRAS galaxies & selected Seyfert 2s instrument: AAT 3.9-m Heisler, Lumsden, & Bailey (1997) sample: 16 IRAS-selected Seyfert 2s, S60 > 5 Jy instrument: AAT 3.9-m Moran et al. 2000 sample: 38 objects from Ulvestad & Wilson (1989); distance-limited (cz < 4600 km s–1) instrument: Keck 10-m (“snapshot” mode, ~20min per object) Lumsden et al. (2001) There have been a number of spectropolarimetric surveys of seyfert 2s that were carried out over the past decade or so…. They used different samples selected in different ways. And Most were done on 4-5m class telescope. The moran etal survey used the keck tel, but it was more of a snapshot survey, using only a minimal amount of time per object, and so I think it was comparable in depth to the other surveys. The largest survey was carried out by myself, using the cfa and 12 micron samples at Lick and palomar observatories.. And I will concentrate on this for my talk… sample: 24 IRAS-selected Seyfert 2s, S60 > 3 Jy, LFIR > 1010 L¡, S60/S25 < 8.85 instrument: AAT 3.9-m, WHT 4.2-m Tran (2001, 2003) sample: 49 objects from the CfA and 12 mm samples instrument: Lick 3-m & Palomar 5-m

Summary of Surveys HBLRs detected in  1/2 of Seyfert 2 population
Two possibilities: HBLRs are there but hard to detect Something wrong with the UM Some explanations: Only HBLR S2s are true counterparts to Seyfert 1s Non-HBLR S2s are too weak to possess or sustain any BLRs Broad-line variability Non-HBLRs are narrow-line Seyfert 1s Evolution Obscuration / torus inclination Scattering material The one thing all surveys agreed on is that not all sey2s observed showed hblr in polarized light. In fact, only about half of the s2s showed a detection of hblrs. There are two possibilities. . One.. Is that the hblrs are there, but they’re just hard to detect. Spectropolarimetry is hard, you need a lot of photons to see anything statistically significant…and the polarization levels are often very low, of order 1/2 % or so.., so you throw about 99% of the light away.. We could stop there and go on with our lives, but I find that unsatisfying… Or there’s something wrong with the UM, at least in its simplest form. There are many variables, and some explanations might include…. Evolution may also play a role.. The obscuration and torus inclination may be important,… Or that the scattering material may be missing, or not well placed..

Viewing Angle Hypothesis
For example, in this picture suggested by miller & goodrich and heisler et al, if the scattering takes place very close in near the nucleus, shown in purple here, only objects viewed at a most favourable inclination angles would show HBLRs. The ones viewed more edge-on wouldn’t get any scattered light at all… Miller & Goodrich (1990) C. Heisler (1997)

Diagnostic Diagrams and Luminosity Distributions
Compare various properties among the HBLR, non-HBLR S2s and S1s, illustrating their similarities and differences. These properties include: AGN Strength: Optical: [O III] l5007 luminosity Infrared: Mid-IR 25mm luminosity (L25); IR color f25/f60 Radio: 20cm radio power (S20); S20/f60 X-ray: Hard X-ray (2-10 keV, HX) luminosity; HX/f60 Obscuration: Ha/Hb; NH; HX/[O III]; Fe Ka equivalent width [EW(Fe)] Star formation: Far-IR luminosity LFIR = L60 + L100 Well, with the surveys that were carried, out we could compare the various observational properties among the hblr and non-hblr s2 and see if there are any similarities and differences between them and how they relate to the seyfert 1s. I’ve done that with the samples that I surveyed, and the properties that were compared include indicators of AGN strength, such as oiii, hard xray, radio and mid-infrared luminosities, and their ratios; various probes of obscuration, such as the Balmer decrement, xray to oiii ratios, fe kalpha equivalent widths and column densities. We also compare their far-infrared luminosity, which is a good tracer of the star forming activity.

[O III] Luminosity vs f25/f60
HBLR Sey 2 Non-HBLR Sey 2 Sey 1 HII, LINERs, Starburst stronger f25/f60 is the IR color  dust temp. around AGN [O III] Lum.  AGN strength I will highlight a couple of properties that I think best illustrate the results. Here’s a plot of the oiii luminosity vs the iras 25/60 micron ratio. Solid blue dots represent the hblrs, open circles are the non-hblrs, stars are hii, liners and sbs, which generally don’t have any broad lines, and the seyfert 1s are shown in red. The infrared ratio, f25/f60, can be thought of as the dust temperature around the AGN. OII luminosity is a good indicator of AGN strength. On average, the HBLRs are more luminous and warmer than non-hblrs. They tend to lie among the S1 galaxies while largely avoiding the lower left corner, which is occupied mainly by non-hblrs and HII, liners and SB galaxies.. HBLRs are more luminous and warmer than non-HBLRs weaker hotter cooler

[O III] Luminosity S1 HBLR S2 Non-HBLR S2 [flash this up very quickly]
Here are the comparison of the same properties shown as histograms showing the same results.. (As you can see there’s a clear difference in the distribution of oii lum between hblrs and non-bhlr, but it is quite similar between sey1s and hblr s2s. ) Non-HBLR S2

IR Color: f25/f60 S1 HBLR S2 Non-HBLR S2 [flash up very quickly]
Similar results are also seen for the 25 over 60 micron ratio. Non-HBLR S2

Radio Strength: S20cm/f60 vs f25/f60
HBLR Sey 2 Non-HBLR Sey 2 Sey 1 HII, LINERs, Starburst HBLR S2s are stronger and hotter than non-HBLR S2s. HBLR S2s are similar to S1s. Another property that we can look at is the ratio of the radio to infrared flux at 60 micron ratio. This is another good indicator of the AGN strength. Again it shows that on average (as a group), the hblrs are significantly stronger than non-hblrs, but otherwise similar to the seyfert 1s.

CfA + 12mm samples: HX/f60 vs f25/f60
Blue: NH < 1023 cm-2 Green: 1023 < NH < 1024 cm-2 Red: NH > 1024 cm-2 Yellow: Unknown NH strongest Solid curves are iso-NH models of Risaliti et al. (2000). Decreased obscuration shifts points vertically up. decreasing obscuration For a same range of NH, stronger AGN lie to the upper right. HBLR S2s tend to be stronger, but not necessarily less obscured than non-HBLRs. weakest

Comparison between S1, HBLR S2 & non-HBLR S2
Mean, standard deviation (s) and K-S statistical test results for various properties of the S1, HBLR S2s and non-HBLR S2s in the CFA and 12mm samples. Here’s a table showing the full statistical results of the comparisons. I’m not going to go through all of the them…. Note: L in solar luminosity, EW(Fe) in eV, NH in cm-2 K-S Pnull is the probability for the null hypothesis that the two distributions are drawn at random from the same parent population.

Comparison between S1s, HBLR S2s & non-HBLR S2s
S1 = Sey S3 = HBLR Sey S2 = non-HBLR Sey 2 I will just summarize here that for the AGN strength, all indications are that s1 and hblr s2 are the same, and they are significantly stronger than non-hblrs s2s. In terms of obscuration, I found that the hblrs and non-hblrs are about similar, so the lack of hblr detections cannot be explained by obscuration, nor can it be explained by any kind of elevated star forming activities, as the tracer for that, the far infrared luminosity, seems indisinguishable amongst the three types of seyferts.

Large-scale Properties of Seyfert Galaxies
De Robertis et al environment study: Very significant difference found in mean environment between S1s and S2s. Sample contains few known HBLR S2s (NGC 4388). Schmitt et al study of a far-IR selected sample (Keel et al): No statistically significant difference between S1s and S2s in terms of host galaxy morphology and freq. of companions. However, sample was selected for warm IR color, and hence HBLR S2s. 70% of S2s in this sample are HBLRs, compared to ~ 50% or less for others. Clavel et al ISO study: Mid-IR ISO spectra of HBLR S2s are similar to S1s. 7 mm continuum, PAH luminosities and EW. Those of non-HBLR S2s are different (indistinguishable from SBs). Smaller 7 mm conti. L and higher PAH EW, but similar PAH lum. So if we suppose for the moment that non-hblrs are different from hblrs, which in turns are the same the S1s seen from a different direction, then we can examine previous studies in light of this idea. I’ve done this exercise and we can see that for example. In the study of de robertis et al 98 that found different mean environment between s1s and s2s, it’s perhaps because the sample contains few known hblrs. On the other hand, The schmitt et al study found no sig. difference using a different sample. However, that sample was selected for warm IR color, which means that non-hblrs were selected against Clavel etal study found that the mid-ir spectra of hblr s2s are similar to s1s, whereas those of non-hblrs s2 are quite different and similar to starburst galaxies. And finally, if we group the malkan etal 98 hst snapshot survey into the hblr s2s and non-hblrs, we found that the differences in dust morphology that were found between s1s and s2s tend to disappear Malkan et al HST snapshot survey: S2s richer in dust than S1s. However, when grouped into HBLR and non-HBLR S2s, differences in dust morphology and incidence disappear. non-HBLR S2: 55%, HBLR S2: 27%, S1: 23%

Mid-IR ISO Spectra of Seyfert Gals
HBLR S2 PAH features Here are some examples of mid-ir spectra of iso spectra from clavel etal Note that the s1 and hblr s2 spectra are very similar to each other, whereas the non-hblr s2 spectrum is markedly different, being dominated by a strong PAH feature. Non-HBLR S2 From Clavel et al. 2000

HBLR Detection Rates Percentage of detection for different classes of AGNs More powerful type of AGN has more HBLR detections  Evolutionary connection between HBLR and non-HBLR S2s ? SB/LINER  non-HBLR S2  HBLR S2/S1

Two types of Seyfert 2s? Not only are HBLR S2s stronger than non-HBLR S2s, they are comparable in a lot of ways to S1s. Non-HBLR S2 properties, on the other hand, generally do not match well with those of HBLR S2s or S1s. Proposal: HBLRs are visible largely because their AGNs are intrinsically more powerful Many non-HBLR S2s maybe too weak to represent “real” hidden S1s, and HBLR S2s may be the only true counterparts to normal S1s. So in summary,… So I proposed that…

Mid-Infrared line ratios
Since the study came out about 10 yrs ago, there’s been a number of studies that attempt to compare other properties… Wu etal 2011 found that the hblrs tend to show stronger mid-ir NeV and OIV lines than non-hblrs. These lines are good indicators of AGN activity, as they are high ionization lines in the mid-IR that are little affected by dust extinction. Again they confirm the idea that HBLRs are more powerful than non-hblrs. Wu et al. (2011)

Obscuration Luminous S2s: More likely to find HBLRs in less obscured objects L [O III] > 1041 ergs/s L [O III] > 1041 ergs/s more obscured L [O III] < 1041 ergs/s L [O III] < 1041 ergs/s less obscured In terms of obscuration, the picture is a little more more complex. Wu etal separated the non-hblrs objects into the more luminous and less luminous groups, and found that for the more luminous sample, it appears that obscuration may play a role, in the sense that hblrs are more likely to be detected in less obscured objects.Bbut for less luminous objects, there is no statistical difference. Wu et al. (2011) Obscuration plays a role for high-luminosity AGNs

Eddington Ratios Marinucci et al. (2012)
Only objects with good quality data and MBH estimated from stellar velocity dispersion. Clear separation between HBLRs and non-HBLRs, especially in Compton-thin sources. Inclusion of Compton-thick sources blurs out the separation, but no HBLR falls below Lbol/Ledd threshold. Recently, marinucci et al compared the eddington ratios for a sample of seyferts with good quality data, and with black hole masses estimated from stellar velocity dispersion only. Here’s a plot of the bolometric luminosity vs. eddington ratios. The top panel shows compton-thin objects only, so there is ambiguity on their luminosities since no extinction corrections were applied. The black points are hblr s2s and the red points are non-hblr s2s. You can see that there is a clear separation between the two types. If you include the compton-thick sources, shown on the bottom here, the separation is less sharp, but there is no hblr that falls below a certain threshold, indicated by the dotted line here… This is in line with predictions of theoretical models that suggest that below a certain luminosity or accretion rate, no outflows could be supported from the accretion disk, and broad line regions may not be able to form. Theoretical models: BLR unable to form when AGN luminosities or accretion rates are too low to support outflows from accretion disks (Nicastro 2000; Elitzur & Shlosman 2006; Cao 2010)

Two Key Questions Non-detections real or due to limited depth of surveys? Lower AGN luminosity of non-HBLRs leads to intrinsic differences? Now, we still like to address two key (lingering) questions: First, Are the non-detections real, or are they due to the limited depth of the surveys? And even if we accept that the luminosity of non-hblrs are lower compared to hblrs, does that mean there are any intrinsic physical differences between them? Does anything happen that low luminosities?

HBLR Detection: Flux Comparison
HBLR Sey 2 Non-HBLR Sey 2 HII, LINERs, Starburst There is no separation in the observed fluxes for the most common AGN indicators between the two S2 types. HBLR detection reaches to very low flux level, below those of many non-HBLR S2s. We can address the first question by looking at the sources in flux space.. If the detection of hblrs is limited by how bright the object is, then we would expect that hblrs would be preferentially detected in brighter sources. However, we see that there is no separation in the observed fluxes between the two types. Furthermore, the hblr detection reaches to very low flux level, below that of many non-blrs.. The flux distributions are the same for the two S2 types, indicating that there is no preferential selection for HBLRs, and the non-detections cannot simply all be due to the detection limit of the survey! Non-detections of HBLRs cannot simply all be due to the detection limit of the survey.

Deep Keck Spectropolarimetric Survey
Same sample as Tran (2003) CfA and 12μm samples Target mainly non-HBLR Seyfert 2s ~ 25 objects LRIS + polarimeter Typical exposure times: min per object Multiple epochs X deeper than previous surveys Six new southern objects observed at CTIO As you know it is impossible to prove that something is not there, but we want to push as hard as possile.. So to answer the question with even better confidence, over the past several year, I’ve re-observed the non-hblr targets at Keck with much higher sensitivity. Repeated observations were made at longer exposure times, that essentially probe to a depth of about 10X that of previous surveys. I also added to the sample 6 new southern objects observed at CTIO to expand the sample.

Results of the Deep Keck Survey
Six new HBLRs Non-HBLR  HBLR: Mrk 573, NGC 3892, NGC 5929, UGC 6100 (Keck) ESO 541-IG12, NGC 1125 (CTIO) Five new non-HBLRs HBLR  non-HBLR: NGC 5347 (called “HBLR” by Moran et al.) ESO 33-G2, ESO 253-G3, NGC 3147, NGC 4968 The main result was that the new deep survey uncovered only a handful of new HBLRs. There are also a few more non-hblr objects added.

New HBLRs: NGC 3982 & UGC 6100 NGC 3982 UGC 6100
Here are the new hblrs.. Ngc3982 and ugc The top panel is the polarization spectrum, showing the higher polarization in the broad line, which is a classic signature of hblr, due to dilution from unpolarized host galaxy light.

A non-HBLR: NGC 5347 Here is an example of a non-hblr.

[O III] Luminosity – IR Color Diagram
f25/f60 is the IR color  dust temp. around AGN stronger [O III] Lum.  AGN strength Here’s an updated oiii lum vs ir color plot I showed earlier, that now includes the new survey results. Note that It’s not covered wall to wall with hblr objects, but remain pretty much the same as before, and the two types are still significantly different. So previous conclusions largely remain unaffected. weaker warmer cooler

Receding Torus: Evolution of AGN Obscuration
Obscured AGN fraction decreases with luminosity Consistent with receding torus model: larger opening angle for higher luminosity AGN The weaker nature of the non-hblr S2s may lead to other physical effects that could impact their broad-line detectability. For example, at lower luminosities the opening angle of the torus could become narrower, allowing less scattered light to get through or supporting a smaller scattering region, such that hblrs would be hard to detect. This plot from hasinger shows that the obscured agn fraction decreases strongly with luminosity, providing some evidence for this receding torus model. Hasinger (2008)

Evolution? Yu & Hwang (2011) Wu et al. (2011)
HBLR Non-HBLR Wu et al. (2011) solar 5x solar Significant difference in [N II]/Ha ratio Can be explained by an increase in nitrogen abundance in non-HBLR S2s Stellar evolution  overabundance of N/O  evolutionary connection between HBLR and non-HBLR S2s Perhaps evolution plays a role…. Yu & hwang and Wu et al independently plotted these objects on the BPT diagrams and found that there is a significant difference in the NII/Halpha ratio, in the sense that the non-hblr S2s display a higher ratio. Modeling suggests that the ratio can be reproduce by a 5X increase in nitrogen abundance And this may imply that stellar evolution may cause the N/O relative abundance to change and perhaps a evolution connection between the two S2 types

Evolution? (2) S1s HBLR S2s Non-HBLR S2s Starbursts Wu et al. (2011)
Another sign of evolutionary connection is suggested by Wu et al, who showed this plot of the mid-infrared line ratios as a function of the PAH feature equivalent width. Again, the HBLR and non-HBLR S2s are well separated in this diagram, and S1s and hblr S2s stay together. In addition, there is a continuous sequence between the subtypes, shown here by the dotted line, which represents the “mixing line” between pure AGN in the upper left and pure starburst in the lower right. Now, if the non-hblrs represent those in which the broad line region is very weak, or absent, then one would expect some seyfert 2s whose appearance is unchanged regardless of orientation. This is the so called genuine or “true” type 2 seyferts. If we can find objects in which we believe our view to the nucleus is unobstructed, then this idea can be tested.. Wu et al. (2011)

HBLR S2s Non-HBLR S2s Wu et al. (2011) Starbursts

“Naked” View of Type 2 Seyferts
NGC 3147, NGC 4698, 1ES Other candidates in Shi et al (2011) and Bianchi et al (2012) (IRAS , NGC 4594, Q …) All are Compton-thin, consistent with little or no intrinsic X-ray absorption above Galactic column density (~ cm-2 ) Rapid, persistent, and strong X-ray variability observed over 12 year time scale in 1ES X-ray variability also observed in NGC 3147 High hard X-ray to [O III] ratios (1-100) indicate little obscuration Inferred nuclear optical extinction is less than ~ 1 mag. All classified as type 2 AGNs (no broad lines observed)  contrary to expectation from the AGN unification model There are a number of such candidates in the literature, and I’m showing here some of the best candidates. All of these objects have one thing in common in that they appear to show no intrinsic x-ray absorption. They typically show rapid xray variablitiy, or high hard xay to oiii ratios, indication little or no obscuration. So by all indications, we have a seyfert 1, or unobscured, view to these objects, and yet they all have a type 2 spectrum with no apparent broad emission lines.

Spectropolarimetric Results
Deep, repeated observations to probe weak lines. In each case, a small amount of polarization is detected but no polarized broad lines indicative of a hidden broad-line region are seen in the polarized flux spectra. I’ve observed three candidates with deep keck spectropolarimatry in order to look for any faint scattered broad line in case they are heavily obscured. Although a polarized continuum is detected, no broad lines are seen in the polarized flux spectra.

Conclusions Conclusions Conclusions
Near-Infrared Spectroscopy Results The spectra are dominated by galactic starlight, and we do not detect any emission in Paβ or Brγ. No direct broad emission lines are present. A search for direct broad permitted hydrogen lines in the near infrared through any obscuring dust also turned up empty. The continuum is dominated by galactic starlight. If typical broad lines were present, their non-detections would indicate an extinction of AV ~ 11-26 Conclusions If typical broad lines were present, their non-detections would indicate an extinction of AV ~ 11-26, inconsistent with the ``naked'' nature of these galaxies. While the obscuration may be due to different material for X-ray and optical light, it appears that the BLRs in these objects are anemically small or absent, due to the weakness of their active central engines. Conclusions If typical broad lines were present, their non-detections would indicate an extinction of AV ~ 11-26, inconsistent with the ``naked'' nature of these galaxies. While the obscuration may be due to different material for X-ray and optical light, it appears that the BLRs in these objects are anemically small or absent, due to the weakness of their active central engines. Conclusions If typical broad lines were present, their non-detections would indicate an extinction of AV ~ 11-26, inconsistent with the ``naked'' nature of these galaxies. While the obscuration may be due to different material for X-ray and optical light, it appears that the BLRs in these objects are anemically small or absent, due to the weakness of their active central engines.

Why don’t we see any broad lines, given the naked nature of these AGNs?
Misclassified Compton-thin AGNs? - High spatial resolution Chandra and XMM-Newton observations rule out confusion from external sources - Temporal variation in X-ray flux implies X-ray is likely not scattered Variable broad emission lines? - Multi-epoch spectropolarimetric observations designed to search for variability failed to find any - Available spectra over timescales of years do not show any evidence of broad-line appearance Hidden narrow-line Seyfert 1 (NLS1) galaxies? - No emission lines of any kind, broad or narrow, are seen in polarized flux spectra - No polarized FeII emission X-ray unobscured, but optically highly obscured? - Narrow-line Balmer decrements are fairly normal, giving AV < 1.6 mag - Heavy obscuration in BLR itself? High AV/NH ratio? So why don’t we see the broad lines given the seemingly unobstructed view of these AGNs? I list here are a number of possibilities….most of which are not very promising for the reasons shown here. A possible explanation is that the obscuration in the x-ray and optical is vastly different, perhaps there is very heavy obscuration in the blr itself, or these objects represent extreme cases, where the dust to has ratios are abnormally high.

Most Likely Explanation
Low-powered AGNs with weak or absence of BLRs - All three objects are low-luminosity AGNs - Observed Eddington ratios are consistent with being below minimum threshold needed to support BLRs (L/LEdd ≲ 10-3) (e.g.: Nicastro 2000) More recent X-ray & optical analysis of NGC 3147 confirms it as a best candidate for a “true” S2 (Matt et al 2012, Bianchi et al 2012) However, NGC 4698 is likely Compton-thick (Bianchi et al 2012) But I think the most favored explanation is that these represent low-powered AGNs with very weak or absent BLRs. … Their estimated eddington ratios all fall below the threshold needed to support BLRs… More recent analysis by Matt et all and Bianchi etal also confirm that NGC 3147 remains the best candidate for a true S2. However, Bianchi etal also suggest that ngc 4698 may be compton-thick, rather than compton-thin, based on their analysis of simultaneous xray and optical observations.

Correlation with Broad Hα Luminosity
Stern & Laor (2012) Recently Stern & Laor found that there is a tight relationship between the xray luminosity, shown on the left here, or the oiii luminoslty, shown on the right, versus the broad halpha luminosity for a large sample of AGNs from the sdss… There are, I think, nearly 4000 objects in this sample. They estimated the upper limits for the broad halpha luminosities for all the true S2 candidates, based on some theoretical assumptions, and placed them on the diagram. They found that for some candidates, they lie well off the relationship in the Xray luminosity diagram, suggesting that their true type-2 nature may be confirmed. However, on the OIII lumiosity diagram, all of the candidates appear to lie on the extrapolation of the relationship to low luminosities. Thus, the lack of blrs in these objects may NOT be significant. There are a couple of things I’d like to say about this… First, the broad halpha luminosities for these candidates, are of course, all upper limits. So they could well lie to the left of the diagrams, and so could still be consistent with the idea that their broad lines are under-luminous. The assumptions could be wrong, and the upper limits could be overestimated. In fact, Bianchi etal recently found that the OBSERVSED upper limit for the broad Halpha in ngc 3147 is about 2 orders of magnitudes smaller than what’s plotted here. I show these as the blue arrows. I also note that all of the broad halpha upper limits are either about the same or higher than the oiii luminosities! This seems pretty crazy to me, because oiii is perhaps the strongest emission line in these objects, and if the broad halpha is as strong as that, we should have no trouble seeing them. So I think the upper limits here are perhaps too generous. For NGC 3147, Bianchi et al (2012) found upper limit of broad Ha ~ two orders of magnitude smaller!

Seyfert 2 Spectropolarimetry
Some Surprises In the remaining time I will spend a few minutes talking about some surprises that were found in the spectropolarimetric observations of seyfert 2s….

NGC 2110: A Hidden Double-Peaked Emitter
Moran et al, Dec. 2005 February 2007 November 2007 The left panel here shows what was observed in Dec by Moran et al. for NGC Note the high polarization in the blue wing of Halpha, and the clear blue peak in the polarized Halpha emission line. Note that this line profile is asymmetric and very broad, very similar to that of the prototypical double-peaked emitter Arp102b. This, I think, represents the type 2, or hidden counterpart to the normal, directly viewed double-peaked emission-line AGNs. When I observed it in February 2007, this blue peak has disappeared, and the polarization there is also much smaller. By November 2007, the blue peak is starting to come back, with a correspondingly higher polarization there. Clearly, the polarized broad Halpha is changing both in shape and intensity on time scales of months. This is something that has never been observed before in other classical hidden broad lines. In general, variability is not expected in reflected, polarized light, since the scattering process tends to smear out any intrinsic variations. FWHM ~ 15,000 km/s FWZI ~ 28,000 km/s

Double-Peaked Emission-Line AGN

Hidden (Polarized) Hα Profile Variability
NGC 2110 NGC 5252 Similar behavior was also found for NGC And this figure shows the dramatic changes both in strength and profile of the polarized broad Halpha emission line for both objects over a period of a few years. The timescales of variability are of order less than a year, similar to the dynamical timescales of the accretion disks. Profile variability of the broad Halpha emission line has also been a very common property of the normal, directly viewed double-peaked emitters. Timescales ≲ 1 yr (similar to dynamical timescales of accretion disks)

HDPE: Observed Polarization Variability
NGC 2110: Blue continuum and broad-line polarization PAs are the same  similar polarization mechanism and geometry polarization PA does not change with time polarization PA (~70º) perpendicular to ionization cone axes  scattering is the source of polarization NGC 5252: Green This shows the observed polarizations of NGC 2110 in blue and NGC 5252 in green in the q-u plane. Continuum polarizations are shown as solid dots and the broad Halpha polarizations are shown as solid squares. Solid triangles denote the narrow-line polarizations. Solid lines connect adjacent epochs. If you assume that the narrow-line polarization represents the interstellar polarization in the host galaxy for NGC 5252, and make a correction for that, the polarization changes are all along the a radial vector on the q-u plane, indicating that the polarization PA is non-varying and the same between the continuum and broad line. This means that they have the same polarization mechanism and geometry. Furthermore, the polarization PA is found to be closely perpendicular to the radio and ionization cone axes in both galaxies, suggesting that scattering is the source polarization.

What Causes the Rapid Variations?
Variability timescales ≲ 1 yr  very compact scatterers Discrete clouds ≲ 1 ly, NOT filled cones with large filling factor Dramatic line profile changes in polarized flux NOT due to “light echo” or “search light” effect Non-changing polarization PA  changes in line-emitting flux, not scattering medium What could cause such rapid variations in polarizations? The fact that we detect any variations in polarized broad line at all indicates that the scatterers are physically very compact, with size scales of order a few light months, similar to the dynamical time scales of double-peaked emitters. This suggests that the scattering is probably done by discrete clouds rather than in a filled cone with a large covering factor. Secondly, the presence of dramatic line profile changes in polarized flux suggests that they are not caused by simple light echo or search light effects…. And.. because the continuum polarization and PA do not change, with the PA being the same as in the line, these variations are most likely due to changes in the line emitting flux, perhaps because of changes in the structure of the line emitting region, and NOT scattering medium.

Constraining the Properties of the Scatterers
No “smearing” of line profile  Te ≲ 106 K Polarization PA ⊥ to cone axis  polar distribution Scattering spherical “blob” model: f = Lsc/Lin ≈ σT ne 2r ΔΩ assume = σT ne 2r ~ 1 (optically thin) ΔΩ ≈ ¼ (2r/d)2 = r2/d2 2r ~ 1 ly f ~ 1%  ne ~ 107 cm-3; d ≲ 10 pc r τe d ΔΩ With some reasonable assumptions we can try to constrain the properties of the scatterers… Electron scattering is likely since the polarized continuum is fairly flat and not significantly bluened. Assuming a spherical scattering blob with radius r located some distance d from the nucleus, we can write down the scattered fraction of light as this expression here… where delta omega is the fraction of solid angle subtended by the scattering cloud. If the scattering is optically thin, and from the observed variability timescale and observed scattered fraction of about 1%, we can derive that the density of the clouds is about 10^7 per cubic centimeter, and the distance is within about 10pc. Thus, the scatterrers lie just outside the obscuring torus between the BLR and NLR, and they are much more compact and close-in than previously envisioned for the classical HBLR Seyfert 2s, where the scattering region is thought to be the size of the extended NLR, or of order hundreds of parsecs. Scatterers lie just outside the obscuring torus between BLR and NLR Much more compact and close-in than previously thought

Compact Scattering Region, Close to Nucleus
Line-emitting “ejectiles” from the nucleus Ejection (bi-polar) outflow model Simple: same material for both scatterer and line-emitting gas Well-defined bi-cones Radio hot spots or material entrained in the base of jets Preference of DPEs in radio-loud AGNs Problematic for accretion disk model as scattering angle must be small (< 15°) Material from outskirts of obscuring torus itself Clumpy obscuring torus model (Elitzur & Schlosman 2006; Nenkova et al 2008) Properties similar to scattering clumps Viewing angle (i ~ 65°) more consistent with extended cone morphology What could be these compact scattering regions? Well I suggest three possibilities…. They could be line-emitting ejectiles from the nucleus, which would be consistent with the bi-polar outflow model for the double-peaked emitters. Or they might be radio hot spots or material entrained in the base of the radio jet, which would naturally explain the preference of double-peaked emitters in radio loud AGNs Or they could be material from the outskirts of the obscuring torus itself! I note that the physical properties derived for the scattering blobs are similar to those of the clouds in the clumpy torus models of elitzur & schlosman and nenkova et al…

(unobscured/unabsorbed)
Unified Model of AGN Type 2 characteristics (narrow lines) Type 1 characteristics (broad lines) Type 2 View (obscured/absorbed) √ Type 1 View (unobscured/unabsorbed) In summary, let me show this table that conveys what the unified model tries to explain.

(unobscured/unabsorbed)
Unified Model of AGN Type 2 characteristics (narrow lines) Type 1 characteristics (broad lines) Type 2 View (obscured/absorbed) √ - Very red quasars Type 1 View (unobscured/unabsorbed) - Naked (true) S2s - Host-obscured type 2 QSOs - XBONGs?

Summary Unified Model of AGN is undoubtedly correct
Orientation, evolution, luminosity, and obscuration may all play important roles Scatterers can be very compact, (≲ 1 ly), and located close (< 10 pc) to central nucleus, giving rise to variability of polarized light So… In summary, I think there is no question that the um is correct. I think it’s been proven beyond reasonable doubts…. But I believe other factors, such as orientation, evolution, luminosity and obscuration may all play important roles. Finally, scatters can be very compact and located very close to the nucleus, giving rise to variability of polarized light. I’ll stop there and thank you for your attention. ==== So from these observations we can draw the following conclusions: The fact that we detect any variations in polarized broad line at all indicates that either the scatterers are physcially very compact, with size scales of order a few light months, similar to the dynamical time scales of double-peaked emitters. Either case, this suggests that the scattering is probably done by discrete clouds rather than in a filled cone with a large covering factor. Secondly, because the continuum polarization and PA do not change, with the PA being the same as in the line, these variations are most likely due to changes in the line emitting flux, perhaps because of changes in the structure of the line emitting region, and NOT scattering medium. Finally, because they share many characteristics similar to the double-peaked emitters, as listed here, Ngc 5252 and 2110 are the type-2 or hidden counterparts of this class of objects, which has not been found until now. Thank you and I’ll stop there. 56

Backup Slides

Summary Unified Model of AGN is undoubtedly correct
Orientation, evolution, luminosity, and obscuration may all play important roles In summary, I think there is no question that the um is correct. I think it’s it’s been proven beyond reasonable doubts…. But I believe other factors, such as orientation, evolution, luminosity and obscuration may all play important roles. Finally, scatters can be very compact and located very close to the nucleus, giving rise to variability of polarized light. I’ll stop there and thank you for your attention. 58

Laor: Origin of Radio Emission

Eddington ratios Bian & Gu (2007)
HBLR Bian & Gu (2007) Non-HBLR Compton-thin S2s Lack of HBLRs below some accretion threshold

AGN Unification Model Type 1 AGN (direct light)
Type 1 AGN (scattered/ polarized light) Type 1 AGN (direct light)

Mid-IR Luminosity: 25mm S1 HBLR S2 Non-HBLR S2

CfA + 12mm samples: HX/f60 vs f25/f60
Blue: NH < 1023 cm-2 Green: 1023 < NH < 1024 cm-2 Red: NH > 1024 cm-2 Yellow: Unknown NH Solid curves are iso-NH models of Risaliti et al. (2000). HX/f60: mainly a measure of obscuration, but also indicates AGN strength.

Central AGN: Double Absorber Model
Risaliti et al. (2000) X rays from the nucleus are heavily absorbed, while the 25mm emission, originating from the warm dust of the torus, is much less obscured.

Far-IR Luminosity: (60mm+100mm)
LFIR: indicator of star-forming (non-AGN) activity in host galaxy. Indistinguishable distributions for all 3 Seyfert types. HBLR S2 Non-HBLR S2

AGN and Normal Galaxy Optical Spectra
Hb, [O III] Ha, [N II] Intensity Typical “active” gal spectrum emission lines Typical normal gal spectrum: integrated light of stars absorption lines Wavelength (A)

CfA & 12mm Seys: [O III]/Hb vs f25/f60
Sey 1 (narrow Hb comp. only) HBLR Sey 2 Non-HBLR Sey 2 HII, LINERs, Starburst

Correlation of HX/[O III] with EW(Fe) & NH
 HBLR Sey 2,  Non-HBLR Sey 2,  HII, LINERs, Starburst There is a good anticorrelation between HX/[O III] and both NH and EW(Fe), indicating that these parameters can be used as measures of the nuclear obscuration.

Obscuration: Hard X-ray/[O III] & NH
Similar obscuration between HBLR and non-HBLR S2s HBLR S2 Non-HBLR S2

Hard X-ray vs. [O III] Luminosities
HBLR Sey 2 Non-HBLR Sey 2 HII, LINERs, Starburst HBLRs and non-HBLRs can be well separated by their HX and [O III] luminosities.

Figure 4.- Hard X-ray vs. [O III] Luminosities
The diagram can be divided into four quadrants with the dividing lines roughly at L(HX) = erg/s and L([O III]) = erg/s. There is a good positive correlation between these two quantities, as would be expected. In the upper right quadrant lie exclusively the HBLRs; these are the strong AGNs with genuine hidden S1 nuclei. The lower right quadrant is occupied by similarly powerful AGN with HBLRs, but these suffer from high obscuration; they are the so-called Compton-thick AGNs. All of the four labeled HBLR occupants in this quadrant (IC3639, N424, N1068, N7674) have NH > 1024 cm-2. The vast majority of the non-HBLRs lie in the lower left quadrant; these are the intrinsically weak AGNs. The lack of objects in the the upper left quadrant is real: hard X-ray luminous AGNs are not expected to show weak [O III]. This diagram shows that HBLRs and non-HBLRs can be well separated by their HX and [O III] luminosities.

Spectropolarimetric Study of Seyfert Galaxies
Spectropolarimetric survey of complete samples of Seyfert 2s (CfA and 12mm samples) to look for hidden broad-line regions (HBLRs) Observations: Lick & Palomar Observatories 59 Seyfert 2 galaxies Results: HBLRs detected in  1/2 of Seyfert 2 population How are the HBLR S2s different from non-HBLR S2s? Is the Unification Model (UM) applicable to all AGN types? Generality of UM for Seyfert galaxies? Correlation of presence of hidden broad lines with observational properties Extension to other types of AGN: Is there a connection to AGN power?

Figure 9 from Parsec-scale Dust Emission from the Polar Region in the Type 2 Nucleus of NGC 424
S. F. Hönig et al ApJ doi: / X/755/2/149

Hien D. Tran W. M. Keck Observatory

Similar presentations

Presentation on theme: "Hien D. Tran W. M. Keck Observatory"— Presentation transcript:

Similar presentations

About project

Feedback

Log in

Auth with social network:

Hien D. Tran W. M. Keck Observatory

Similar presentations

Presentation on theme: "Hien D. Tran W. M. Keck Observatory"— Presentation transcript:

Similar presentations

About project

Feedback