Whittle : EXTRAGALACTIC ASTRONOMY

 
       
 

1 : Preliminaries	6 :   Dynamics I	11 : Star Formation	16 : Cosmology
2 : Morphology	7 :   Ellipticals	12 : Interactions	17 : Structure Growth
3 : Surveys	8 :   Dynamics II	13 : Groups & Clusters	18 : Galaxy Formation
4 : Lum. Functions	9 :   Gas & Dust	14 : Nuclei & BHs	19 : Reionization & IGM
5 : Spirals	10 : Populations	15 : AGNs & Quasars	20 : Dark Matter

(1) Introduction

• We now know that ~all galaxy bulges harbour supermassive black holes (SMBH) in their nuclei.
  Most are quiet/silent - starved of gas and detectable only via near-nuclear orbital dynamics
  A few are accreting gas which makes them visible through the release of potential energy.
  Such nuclei are called Active Galactic Nuclei (AGN) and their hosts are called Active Galaxies.

  One definition for AGN might therefore read:

  "A galaxy nucleus which shows evidence for accretion onto a supermassive black hole"

• First note what is not considered an AGN:

1. Nuclear SMBHs are not sufficient to assign an AGN classification:
   e.g. M31 and M32 both have SMBH's but neither show signs activity (accretion power).
2. Accreting stellar mass black holes are not AGN -- we need an SMBH located in a galaxy nucleus.
3. The activity classification explicitly excludes stars or stellar life cycles
   Star formation, HII regions, Supernovae, etc may be present, but they do not constitute an AGN.
4. Luminosity isn't part of the definition: activity can be quite weak
   For this reason, one often qualifies the power of the AGN, for example:
   "QSOs are powerful AGN"; "M81 contains a weak AGN"

• In practice, "evidence for accretion onto an SMBH" means one or more of the following:

1. Emission lines with distinct ratios (defined in 3d below).
   Often, these are strong & imply photoionizaton by a hard spectrum, rather than by a black body.
2. Broad H wings or profile (1000 km/s) - i.e. much more than typical galactic velocities.
3. Broad band unresolved continua, with comparable power in X-ray, UV, optical, IR
   In the optical, they often appear blue; overall, their continua are unlike starlight.
4. Variability of one or more parts of this continuum (or the emission lines)
   this points to a highly compact (1pc) source of emission.
5. Radio source, often bipolar & significantly more luminous than in normal galaxies
   this may also show jet-like morphology; steep spectrum jet & lobes; flat spectrum core.
   

• The study of AGN started ~50 years ago and now comprises a major branch of astrophysics
  A number of factors make AGNs worthy of study:

1. At their heart lie giant black holes, perhaps the most exotic astrophysical animal we know of.
   AGN give us observational access to these otherwise ellusive objects

2. Almost by definition, physical processes near SMBHs are in the relativistic regime
   A host of interesting and extreme physical process occur.
   While often confusingly complex, they provide a laboratory for studying high energy processes.

3. Black hole accretion is very efficient, making AGN amongst the most luminous objects known.
   e.g. for mass m in orbit at R_S = 2GM/c², we have 2KE ~ PE ~ GMm/R_S = ½mc²
     typical accretion efficiencies are ~¼ mc²
   this is huge compared to either fusion (~0.7%) or chemical (~10^-6%) energy sources.
   Accreting just 1 M yr^-1 yields ~10⁴⁶ erg s^-1 4 × 10¹² L 200 L galaxies.

4. In addition to photon luminosity, AGN can provide powerful collimated mechanical luminosity.
   The creation, propagation and interaction of these jets provide yet another important area of study.
   Of course, both accretion flows and jet flows are found in many astrophysical contexts
     AGNs provide excellent windows on both these complex phenomena.

   High AGN luminosities yield other important opportunities:

5. Luminous AGN can be seen to enormous distances (currently, z~7, lookback time ~95%)
   One can therefore study directly their cosmic evolution.
   They show strong evolution, with a "hey day" at z~2 (×1000 more common than today).
   This evolution is closely tied to early galaxy construction -- another topic of great importance.

6. As distant beacons, their spectra provide access to the intervening IGM; for example:
   • the highest z QSOs probe the epoch of reionization,
     
   • the transparency of the Lyman continuum proves the IGM is highly ionized
     
   • the ISM in galaxy halos can be analysed throughout much of cosmic history
     
   • the disks of young spirals can also be studied.

7. The accretion power ultimately emerges in essentially all spectral wavebands
     AGN present observationally very rich targets
     they feature in ~all new satellite & instrument observational programs.

8. AGN influence 11 decades, from accreting SMBH (~AUs) to jet powered radio lobes (~Mpc)
   Of these decades, ~half are observationally unresolvable.
     AGN offer a wonderful window on a huge range of environments and physical processes.

• With this motivation, let's now begin to look at AGN in more detail
  As with all our other Topics, the treatment will be relatively broad and superficial.

(2) Brief Historical Sketch

1909 : Fath used Lick's Crossley reflector to take spectra of "spiral nebulae"
motivation was to test their extragalactic nature: similar to stars, or MW gas clouds?
most had absorption lines (supports extragalactic); NGC 1068 also had nebula-like emission lines

1917-30 : Slipher, Hubble, Humason confirm Fath's observations & find other examples"

NGC 1275, NGC 4051, NGC 4151.

1943 : Carl Seyfert studied 6 of the 12 known examples in more detail. He found:

1. luminous semi-stellar nuclei (image)
2. nebular spectra of high ionization
3. emission lines broader than either the absorption lines, or HII regions from irregular galaxies
4. forbidden lines narrower than Hydrogen lines in some objects.
Not much notice taken at the time.

1974 : Kahchikian & Weedman argue for Doppler origin for the line-widths; define Sy 1 & Sy 2.

1932 : Jansky discovers radio from the sun & MW

1944 : Reber maps sky @ 160 MHz (1.9 m) & sees peak in Cygnus (Cyg A)

Late 1940s : Post WWII groups @ Manchester; Cambridge (UK); & CSIRO (Australia)
simple interferometry (Michelson & cliff) gives positions & sizes (small) to few arcmin.
  Virgo A = M87; Centaurus A = NGC 5128; Cygnus A = ??

Early 1950s : Baade & Minkowski identify Cyg A: "colliding galaxies"
emission lines @ cz=16,830 km/s   high radio luminosity

Late 1950s : small double structures; v.high brightness temperatures.
Burbidge applies synchrotron theory   10⁶⁰ erg ~ 10⁶ M c² stored energy.

1974 : Blandford & Rees provide "twin exhaust model" of jets to power the radio lobes.

1960 : Minkowski identifies 3C 295:   faint galaxy @ z = 0.46 (huge!)
Sandage identifies 3C 48 :   16^m star + faint fuzz;   strong UV & variable.
Greenstein's spectrum shows broad emission lines at odd wavelengths
Several other 3C "stars" identified :   quasi-stellar radio sources, QSRS; QSS; quasars.

1963 : Hazard, Mackey & Shimmins observe lunar radio occulation of 3C 273
  double source; sep 20" (A & B);   B = optical star (f ^0.0);   A = faint jet (f ^-0.9)
Schmidt recognizes H, H, H, H @ z=0.16 and can therefore identify MgII 2800.
Greenstein & Oke : confirm H in 3C 273 & identify MgII in 3C 48 @ z=0.37
Several papers in Nature:


  they prefer cosmological redshift, but this means L(QSO) ~ 10-30 × L(giant elliptical)
  suggest central mass 10⁹ M provides the energy & confines the emission line gas

Mid 1960s : more redshifts @ z~1 ( CIII] 1909; CIV 1550); 3C 9 Ly z=2.012
Sandage finds radio quiet UV bright "QSOs"; estimates more common than radio loud
assuming cosmological z (see below) required high energy sources:
Suggestions: SN chain reaction; collisions in dense star clusters; supermassive stars.
Salpeter (1964) & Zeldovich (1964) independently suggest accretion onto supermassive black hole
Lynden-Bell (1969): developes accretion power; "dead" QSOs may inhabit all nearby galaxies.

• High measured z suggests very high luminosity
  Rapid variability suggests very small source
    difficult to believe   perhaps z is not cosmological?

• Alternatives: local objects with high gravitational redshifts,
  objects with high Doppler velocities, ejected either from the MW or some nearby galaxy.

• Objections to both these, for example:
  difficult to explain same z for broad and narrow lines with gravitational redshift
  Very high energy for local ejection; why no blueshifts for non-local ejection?

• Evidence grows for cosmological redshift, for example:
  Cluster galaxies with same redshift as QSO
  Fuzz around some QSOs consistent with host galaxy; ultimately same z measured.
  Highest Seyfert luminosities overlap lower QSO luminosities: continuous population
  Today, HST sees host galaxy whenever it "should". Note, it can be difficult to detect because:
  1. QSOs nucleus often outshines surrounding galaxy
  2. Rest frame UV of galaxy is much fainter than in the optical for nearby galaxies
  3. Galaxy surface brightness drops as (1+z)⁴, making detection difficult.

• Today, cosmological redshifts aren't in doubt, but it was important to check them thoroughly.

(3) The AGN Paradigm

• Before encountering the zoo of AGN species, let's outline their basic anatomical components
  Note: not all of these components are apparent/present in any given AGN
  The length-scales below refer to a moderately luminous AGN:   L ~ 10⁴⁵ erg s^-1,   M_BH ~ 10^7.5 M
  these lengths probably depend on luminosity: ×10 for QSOs; ×0.1 for weaker AGN.

• A number of "global" figures help:
  Figure 1:   sketch showing 5 nested scales for Seyferts
  Figure 2:   single spatial axis with various ranges identified
  Figure 3:   Blandford's comic-strip of 12 successive scales differing by a factor 10 in scale.
  Figure 4:   Robson's version of Blandford's comic-strip
  Figure 5:   size & waveband axes identifying various emission regions
  

• Summarizing these, from innermost to outermost scales, we have (plus/minus ½ decade):

Log r pc	Other	Region/Properties
-5	2 AU	RS = 2 MBH,8 AU,   last stable orbit is 3RS   (MBH,8= MBH/108M)
-4	20 AU	relativistic accretion disk (AD); Fe K line; variable X-ray emission (mins-hours)
-3	200 AU	UV accretion disk; radiation supported AD;
-2	2000 AU	optical AD; Broad Line Region (BLR); V~104 km s-1; ~few light days.
-1	0.1 pc	compact, flat-spectrum radio core; VLBI jet; outer BLR
0	1 pc	star cusp with velocities affected by BH; dense gas may block some sight lines.
1	10 pc	Inner Narrow Line Region (NLR); V 103 km s-1; inner bulge.
2	100 pc	bulge; NLR; forbidden emission lines; warm dust; VLA jet.
3	1 kpc	inner disk; disk ISM; Extended-NLR; Bar & inflow; VLA jet.
4	10 kpc	host galaxy; distortions/merger?; ENLR; weak jets blocked.
5	100 kpc	near neighbors; tidal influences; powerful jets.
6	1 Mpc	jets terminate in IGM hotspots; may affect ICM in cluster.

• A related summary that considers causal processes goes like this:

1. On large scales, galaxy disturbances/mergers are efficient at removing angular momentum from orbiting gas and this finds its way down to the nuclear regions, possibly with an associated starburst. Some of this gas may cool and form a dusty thick disk in the central 1-1000 pc. This gas can obscure our view of the central regions and, conversely, block the outgoing radiation from the AGN along certain directions.

2. Ultimately, some of this gas finds its way into an inner accretion disk, whose Keplerian dynamics is dominated by the black hole. Here, magnetic fields create viscosity in the disk, which causes the gas to move inwards, releasing gravitational energy which heats the disk. The disk has a temperature gradient, and generates at least some of the optical/UV/X-ray continuum emission via thermal radiation at its surface. Also likely are non-thermal processes in a disk corona which add to the continuum. Ultimately, gas within a few R_S emits X-rays before crossing the horizon, adding to the mass and angular momentum of the black hole.

3. It is in this innermost region that a jet is somehow generated moving out along the disk axis, probably via magnetic fields tied to the rotating disk. It is not known why this jet is sometimes very powerful (radio loud AGN), and sometimes much weaker (radio quiet AGN). The jets are born with bulk relativistic speeds, and contain magnetic fields and particles with random relativistic speeds -- hence these jets are visible mainly via synchrotron. The jets burrow out through the surrounding galaxy ISM with varying degrees of difficulty, entraining and disturbing the ISM en-route. The most powerful jets can escape the galaxy ISM altogether and traverse a Mpc before slamming into an outward moving "working surface" of IGM material.

4. Returning to the central engine, the accretion process also yields a luminous source of X-ray, UV, and optical photons, with roughly power-law spectral shape. This radiation floods out into the galaxy, although some directions may be blocked by the denser ISM components. Whether or not we get a clear line of sight to the nucleus is thought to explain at least some of the observational variety of AGN.

5. The UV-X-ray radiation is an effective ionizer. It first encounters gas clouds within the central few light days/weeks and the resulting line emission reveals cloud velocities ~10³ - 10⁴ km s^-1 with gas densities ~10¹⁰ cm^-3. This emission defines the so-called "Broad Line Region" (BLR). The BLR velocities are probably gravitational, while the origin of the BLR material is uncertain -- perhaps arising from an accretion disk wind.

6. At larger distances, the ISM velocities and pressures both drop. By 10pc - 1kpc the ionized gas clouds yield velocities ~few ×10² km s^-1 -- roughly equal to typical bulge gravitational speeds -- and gas densities fall from ~10⁶ to ~10² cm^-3. This emission defines the so-called "Narrow Line Region" (NLR). Depending on the geometry of the inner obscuration, the NLR may show bi-conical shape. If jets are blocked in this region, the gas often exhibits significantly higher disturbed velocities.

7. Sometimes the jets and ionized gas can be seen to much larger radii, often in association with eachother. The most powerful jets can travel far from the galaxy and dump their energy in the IGM or, if the galaxy is in a cluster, in the ICM. In that case, large cavities in the X-ray emission can be seen surrounding the expanding radio source.

• With this outline now in place, we can move on to retrace the rather tangled web of historical discovery, which led to an equally tangled web of AGN taxonomy. Hopefully, if you keep the basic anatomy and physiology in mind, you'll be able to clearly navigate this more historical part.

(4) AGN Taxonomy

• Over the years, many different categories of AGN have been identified, each with its own name.
  In hindsight, this apparent diversity has three distinct components:
  • Genuine variation in one or more key properties of AGN
  • Apparent variation arising from our particular viewing angle
  • Variation which arises from observing the object at a particular stage in its evolution.

• There are (at least) two genuine ways in which AGN differ from eachother:
  1. UV/opt/IR luminosity: This probably tracks the accretion rate and black hole mass.
     Three categories:
     
     Quasars   (L_nuc L_gal);
     Strong AGN   (L_nuc L_gal);
     Weak AGN   (L_nuc << L_gal).

  2. Radio Luminosity: determined mainly by the jet power
     The cause of jet power isn't known : possibly black hole spin and/or disk temperature?
     Two categories:
     
     Radio Quiet   (L_R 10^-4 L_opt)
     Radio Loud   (L_R 10^-2 L_opt)

• There are (at least) two differences which seem to depend on viewing angle
  1. Emission Line Shapes: do we get a clear view to the inner (BLR) region, or not.
     Two categories:
     
     Broad + Narrow lines   (can see the BLR)
     Narrow lines only   (can't see the BLR).

  2. High variability & high polarization: this depends whether we look directly down a fast jet.
     Two categories:
     
     High variability & polarization   (look down jet);
     Low variability & polarization   (look at some other angle).

The following table loosely identifies a number of categories of AGN.
Click on the links to see example spectra in the image window.

table TBD

Let's define in observational terms some of these names;
(some names are not quite standard, but they will help clarify our discussion of unification)

Name	Radio	Observational Characteristics
Seyfert 1	RQ	moderate luminosity (MB > -23); strong/visible blue optical continuum; host galaxy clearly visible; broad + narrow lines; narrow lines have high ionization; radio quiet
Seyfert 2	RQ	as for Seyfert 1, but weaker/no blue optical continuum & only narrow lines visible
QSO	RQ	Radio quiet quasar; optically luminous (MB < -23); host galaxy barely/not visible; strong optical continuum; broad + narrow lines of high ionization
QSO-2	RQ	same as QSO but missing broad lines; not many currently known (some are IRAS-QSOs)
LINER	RQ/L	Low Ionization Nuclear Emission Line Region; weak/no continuum; narrow lines of low ionization & moderate strength; sometimes weak broad H visible; can be either radio quiet or loud
BLRG	RL	Broad Line Radio Galaxy; similar to Seyfert 1 but radio loud
NLRG	RL	Narrow Line Radio Galaxy; similar to Seyfert 2 but radio loud
PRG-II	RL	Powerful Radio Galaxy of Fanaroff-Riley class II (edge brightened, powerful jet); unspecified optical spectrum (could be BLRG/NLRG/LINER)
RG-I	RL	Similar to PRG-II except lower radio luminosity & Fanaroff-Riley class I (edge darkened, lower power jet)
LD-QSR	RL	Lobe Dominated (steep spectrum) Radio Loud Quasar; usually FR-II radio morphology; optically similar to QSO
CD-QSR	RL	Core Dominated (flat spectrum) Radio Loud Quasar; optically similar to QSO
BL-Lac	RL	Strong featureless continuum, no/weak lines, little starlight; highly variable; high polarization; radio loud flat spectrum core
OVV-QSR	RL	Optically Violently Variable Quasar; similar to BL Lac but normal QSO spectrum

Notes addressing some details:

1. Whether QSO-2s exist (optically luminous AGN with no visible BLR) isn't yet clear
   A few examples exist, but they seem to be rare.

2. The Seyfert 1 - 2 division is continuous; there are intermediates with weak BLRs
   In practice, four intermedate designations have been used:
   • Seyfert 1.2 : [OIII]5007 / H(broad) ~ 1
   • Seyfert 1.5 : [OIII]5007 / H(broad) ~ 0.3
   • Seyfert 1.8 : H(broad) just visible
   • Seyfert 1.9 : H(broad) not visible, but H(broad) just visible

3. Whether all Seyfert 2s have hidden BLRs is still unclear (some certainly do)
   There is likely to be an intrinsic range in BLR strength, possibly going down to zero.
   NLRs (ionized gas on kpc scales), on the other hand, seem to be ubiquitous.

4. The physical nature of LINERs is still under review (see below, in 3d)
   There may be (at least) two types, "weak [OI]" and "strong [OI]" LINERs ([OI] 6300).
   Weak [OI] LINERs may be powered/photoionized by stars (possibly post-AGB stars)
   Strong [OI] LINERs usually have weak broad H wings, suggesting a true AGN.
   In that case, low radiation parameter photoionization by an AGN matches the spectra well.

5. The association of Elliptical/Spiral with radio loud/quiet is still uncertain.
   While this seems the case for intermediate/low AGN power, it is less clear for QSOs

• Galaxy nuclei can be both active and have star formation.
  What we see (and classify) depends on the relative strength of these contributions
  • strong activity can mask weak/moderate star formation
  • comparable contributions can lead to mixed classification
  • star formation can mask weak activity.

• Regarding mixed objects, there are two contexts worth mentioning.
  1. At modest luminosity, emission line ratios can lie between those of AGN & HII region
     often the two components have different kinematics (line profiles; kpc-scale velocities)
  2. At high luminosity, many luminous infra-red galaxies (e.g. ULIRGs) may hide a buried QSO.
     such objects may be evolving from merger/starburst, to blowout, to exposed QSO (Topic 7).

• Given the various possible types of galactic nuclei, is there a quick/easy way to tell them apart?
    in most cases, the answer is yes, by using the relative strengths of the narrow lines

• Ultimately, different objects contain different sources of ionization with different spectral form.
  The resulting ionized regions have different structure and different emission line strengths.
  For example, here are four classes of emission line region:

  Object	Ionizing source	Spectral form
  PN	Post-AGB star	Black Body, T ~ 35,000K
  HII region	OB stars	Black Body, T ~ 10,000K
  Seyfert	strong AGN	Power Law, High U
  LINER	weak AGN	Power Law, Low U

• Two important papers introduced ratio-ratio diagrams to separate these classes :
    Baldwin, Phillips & Terlevich (1981 PASP 93, 5)
    Veilleux & Osterbrock (1987 ApJS 63, 295)
  the diagrams use only strong lines, and have come to be known as BPT (or VO) diagrams.
  

  BPT use : [OIII]5007 / H;   [NII]6584 / H;   [OIII]5007 / [OII]3727;   [OIII]5007 / [OI]6300.
  VO use : [OIII]5007 / H;   [NII]6584 / H;   [OI]6300 / H;   [SII]6717+6731 / H

  Note that some BPT ratios need reddening correction while the VO ratios don't (being close in ).

• Here's some example spectra (image) and BPT diagrams (image)
  In general:
  • Seyferts have strong [OIII] / H ; strong [NII] / H; strong [OI] / H
    i.e. a wide range of ionization degree
  • LINERs have lower ionization degree: weaker [OIII] / H but strong [NII] / H & [OI] / H
  • HII regions have much weaker [OI] / H and somewhat weaker [NII] / H.

  Some comments:

  • HII regions lie along a sequence in metallicity : low Z have higher ionization.
    The sequence can reach [OIII] / H ~ 10, so high [OIII]/H is not unique to Seyferts.
  • AGNs are different because their ionizing spectra are broad, with a high energy component.
    Behind the fully ionized region lie large partially ionized regions kept hot by x-ray heating.
    It is in these regions that the [OI], [NII] and [SII] lines are generated efficiently.
  • One can think of the Seyfert/LINER group as a single sequence in radiation parameter, U.
    e.g. Seyferts: intense hard radiation field     LINERs: weak hard radiation field
  • Very high ionization lines can be unambiguous indicators of AGN:
    e.g. [NeV]3426, since Ne³⁺ + h     Ne⁴⁺ + e^-   requires 90 eV photons.
  • Some non-AGN emission regions can mimick LINER spectra :
    e.g. cooling flows; shocks in starburst driven winds; bulge ISM ioinzed by post-AGB stars.
  • Fast (500 km/s) shocks can also mimick Seyfert spectra, though no clear cases are known
  • Through the 1980s [OIII] linewidth was used in defining AGN (FWHM > 300 km/s)
    no more: FWHM tracks bulge mass; many AGN now known with FWHM < 300 km/s.

(5) AGN Detection and Identification

• AGN, particularly luminous ones, are rare - how does one find them?
    search for objects with, e.g., strong radio; UV; emission lines; X-ray.
  Since AGN differ in their properties,
  no method finds all AGN; and
  all samples are biassed to some degree.
  However, each selection strategy has led to a deeper understanding of the AGN phenomenon.

Historically, most early AGNs were discovered from radio surveys at Cambridge & Parkes: e.g.

• 3C (1959) & 3CR (Revised; 1962) cover > -5^o with S > 9 Jy @ 178 MHz (328 sources)
  Note: 1 Jansky = 10^-26 W m^-2 Hz^-1 = 10^-23 erg s^-1 cm^-2 Hz^-1
  1C & 2C were found to be badly confusion limited, and are never referred to.
  4C (1965) had S > 2 Jy; PKS (1965) survey covered the southern sky @ 408 MHz.

• Optical IDs found ~75% radio galaxies ~25% QSOs;
  High-z (>2) AGN quite rare, but present; most have steep spectrum (F ^-0.8)
  Until ~1970 it was thought all AGN were radio loud;
  Then optical surveys find many radio quiet AGN:
  Now think only 10% of AGN are radio loud

• Since these early surveys, there have been many others : e.g.
  • Green Bank: Northern sky with S > 0.8 Jy @ 1.4 & 5 GHz (1970s)
  • Parkes: Southern sky with S > 0.7 Jy @ 2.7 GHz (1970s)
  • FIRST: ~¼ full-sky @ 1.4 GHz & 5" beam (VLA-B) S > 1 mJy (~10⁶ sources; 1990s)
  • NVSS: ~ 0.8 full-sky @ 1.4 GHz & 45" beam (VLA-C) S > 2 mJy (2 ×10⁶ sources; 1990s)
  • 5C: ~15 areas of 10 deg² with S > 10 mJy @ 408 MHz & S > 2 mJy @ 1.4 GHz (1990s)
  • VLA: deep pencil beam surveys: 0.05 deg² with S > 15 µJy (early 1990s)

• In general, from these surveys one finds:
  • The highest flux sources are all powerful AGN.
  • The intermediate flux sources are also AGN but of lower luminosity
  • The fainter sources include many more normal or starburst galaxies
  • At lower frequency (100s MHz) one tends to find large steep spectrum AGN
  • At higher frequency (GHz) one tends to find compact flat spectrum AGN

(i) Single Color

• Bare AGN have much "broader" spectra than stars/galaxies     UV & IR excesses w.r.t. visible.
  In surveys for QSOs, we need to discriminate against stars
  From a U-B vs B-V color-color plot (peterson fig 1.6) a good choice for QSOs is U-B < -0.4
  The main contaminants are hot sub-dwarfs & white dwarfs (O & B stars are too rare to matter).
  Here are two well known surveys that used this method:

• BQS : Palomar Bright Quasar Survey (early 1980s; Schmidt & Green).
  Used Palomar 48" Schmidt to survey 11,000 deg² to m_B < 16^m; select U-B < -0.44
  1700 candidates, of which 114 (7%) are AGN (90 have M_B < -23, so are classified as QSOs).
  These are often referred to as "PG quasars" (Palomar Green).

• Boyle; Shanks; Peterson (~1990) make a similar survey, but much deeper & smaller area
  Use UK 48" Schmidt with U & J plates for 7   5^o×5^o areas to B_J < 21; select U-B < -0.4
  Automated plate scanning finds 1400 UVX candidates; fiber spectroscopy identifies 420 QSOs.
  They estimate ~90-95% complete for 0.6 < z < 2.2

• The UVX selection method breaks down for QSOs with z above 2.2
  Why? Because Ly enters the B filter, so U-B appears red     confuse with stars
  At even higher z, Ly forest and LyC IGM absorption suppress both U & B     ~invisible
  Notice how N(z) for all known QSOs (c. 1993) closely reflects emission-line influence on color (pet: 10.3; K&N 6.2)

• However, at z > 4.2 the IGM absorption makes B-R much redder than ~all stars
    The APM Survey : (Irwin; McMahon; Hazard; 1991) search 2000 deg² for B_J-R > 3
  Proved to be quite efficient at finding very high-z QSOs.

  (ii) Multi-color

• Not surprisingly, ambiguities in U-B for z > 2.2 are removed by including additional colors
  A number of such multi-color surveys have been done, especially aiming for high-z QSOs.
  Here are just four of them:

• UK Schmidt Survey: (Warren; Hewett; Osmer; 1991) take 6 bands: U,B_J,V,OR,R,I of 43 deg²
  Use the APM to scan the plates for stellar objects with 16 < m_or < 20     5 colors.
  Stars define a "tube" in 5-d space, while QSOs lie outside this space (see pet 10.4).
  Ultimately find 85 QSOs with z > 2.2.

• POSS II quasars: (Kennefick et al 1995) use g,r,i plates of 680 deg²
  look for Ly in r giving very red g-r for 4.0 < z < 4.8
  find ~20 QSOs with z > 4   (needs updating)

• Sloan DSS QSOs: TBD

• 2dF QSS : TBD

• Since AGNs have distinctive optical spectra, is there a way to take spectra of regions of the sky?
  The answer is yes!, and the technique is called slitless spectroscopy
  The method is simple, and has a venerable history (1920s HD spectral types for ~10⁵ stars):
  • Attach a large prism in front of the entire telescope
    
  • Use Schmidt telescopes + photographic plates     ~small-ish aperture + large field of view.
    
  • The prism apex angle is small ( few degrees)     dispersions ~10²-10³ A/mm at the plate.
  Variations in the method use gratings/grisms/grenses, sometimes placed nearer the focal plane
  These also give zero-order images     helps measure positions; magnitudes; redshifts
  The plates contain thousands of small spectra, to be scrutinized by eye (or machine).
  For AGN: look for strong blue/UV continuum &/or emission lines (bright patches)

• Main limitations of slitless surveys are:
  • brighter limiting magnitude (since light is dispersed)
  • difficulty of completeness estimates, since detection depends on several factors
  • bias in favour of strong high equivalent width emission lines
  • there are barren redshift ranges with few lines (e.g. z~1, has only MgII 2800)
  • spectra may be overlapping in crowded (e.g. low b) fields.

  (i) Markarian Survey

• 1967-1981: Byurakan Observatory (Armenia) 1.3m Schmidt surveyed 10,000 deg²
  Look for galaxies m_B 15.5^m with strong UV continuum (term UVX = UV-excess).
  15 lists published containing 1500 "Markarian Galaxies"
  Follow-up spectroscopy (mainly by Osterbrock at Lick) found ~10% (~150) Seyferts
  Since all have strong UVX, Seyferts divide ~50/50 for types 1/2, with quite luminous Seyfert 2s.
  The other 90% are mainly starburst galaxies (strong blue/UV light from OB stars).

• These Markarian Seyferts were a primary target list for studies in the 1970s-80s.
  Their virtue is that they provide a sample of relatively nearby (z 0.05) luminous Seyferts.

  (ii) Other Slitless Surveys

• Since color searches for QSOs fail for z 2 (see above), slitless spectra can help.
  This is astrophysically important, to go beyond the "quasar era" at z ~ 2
  Look for rest frame UV lines: Ly 1215; CIV 1550; CIII] 1909 (eg Ly @ 5500A z~3.5)

• LBQS : Large Bright Quasar Survey (1986 - 95; Foltz; Hewett; Chaffee).
  Use UK Schmidt & 2^o prism & IIIaJ plates to survey 450 deg²
  Select for blue continuum (z < 2.2) or spectral features (z > 2.2; limit z~3.4 @ 5400A from IIIaJ)
  Find ~1000 QSOs with 16 < B_J < 18.5 and 0.2 < z < 3.4
  Sample is ~complete in this brightness & redshift range good for demographic studies.

• PTGS : Palomar Transit Grism Survey (mid-late 1990s; Schneider; Schmidt; Gunn).
  Use 200" & grism & CCDs in drift-scan mode; gives 4400 - 7500A for Ly at 2.5 < z < 5)
  Survey 62 deg² in six stripes     90 QSOs with 2.75 < z < 4.75
    use to show drop in QSO co-moving space density at z > 3.

• The previous methods find high levels of activity which dominate the optical light.
  What about weaker activity? How does one hunt for that?
  One of the most sentitive diagnostics of activity are emission line ratios, for which one needs:
    high S/N ratio spectroscopy of galaxy nuclei, through small apertures.
  A number of such surveys have been done, some with AGN detection as goal, some not.
  For well defined samples these give the most unbiassed window on AGN demographics

• There have been (at least) four spectroscopic surveys of the nuclei of nearby bright galaxies:
  • 1980 Heckman: 100 early type spirals & ellipticals (check)
  • 1983 Keel : 100 spirals (check)
  • 1984 Stauffer : 80 Spirals (check)
  • 1985-95 Ho; Fillipenko; Sargent: 486 galaxies from the RSA (B_T < 12.5)
  To access weak emission lines, galaxy continuum spectra must often be subtracted (figs)
  Emission line ratios are used to establish nucleus type: Seyfert; HII region; LINER.

• From his survey, Heckman identifiedLINERs = Low Ionization Nuclear Emission Line Regions
  While unclear at the time, it is now thought that many of these are weakly active nuclei.

• Using the Palomar survey it is found that (see below for details):
  •   ~86% of galaxy nuclei have detectable (1 A EW) emission lines
    
  •   ~13% are dominated by Seyfert emission (cf. 2-5% in the Markarian lists)
    
  •   ~30% are dominated by LINER emission
    
  •   ~40% are dominated by HII emission
    activity is much more common than originally thought, if one includes weak AGN.

• Recently, the Sloan & 2dF galaxy redshift surveys have found thousands of new AGN :
  These allow a much more definitive analysis of the demographics of galaxy nuclei.

• As discussed in Topic 9.11bv, AGN tend to have "warmer" FIR colors than normal galaxies.
  Rather than peaking near 60µm (30K dust) they peak closer to 12-25µm (~100 - 300K)
  Moderate luminosity AGN can be selected quite efficiently in the IRAS color-color plot
  For example, de Grijp et al (1992) find .... TBD.

• X-ray emission is obviously an efficient diagnostic for AGNs, especially away from the MW plane
  Compared to other wavebands, X-ray surveys have lagged behind somewhat, but are catching up
  It is useful to divide X-rays into two bands : soft (0.2 - 2 keV) & hard (2 - 10 keV)
  Typical columns to galaxy nuclei may absorb soft, but not hard, X-rays
    soft surveys may miss a significant fraction of the high-column AGN (e.g. Seyfert 2s).
  

• There are a number of X-ray surveys which are commonly referenced:
  • HEAO-1 all sky (8.2 Sr) 2-10 keV; S > 3×10^-11 erg s^-1 cm^-2 (Piccinotti et al 1982)
    85 sources detected; ~60 extragalactic; ~30 clusters; ~30 AGN (Seyfert 1s)
  • EMSS (Einstein Extended Medium Sensitivity Survey; Gioia et al 1990)
    hunt in 1435 0.3-3.5 keV IPC fields for serendipitous sources
    Ultimately, ~800 deg²; varying flux limits; 835 sources; 51% Seyferts & QSOs; 4.3% BL Lacs.
  • ROSAT Deep Survey (Hassinger et al 1993) of the "Lockman Hole" (low N_H) at 0.5 - 2 keV
    ~660 sources detected (+ "sky" fluctuations give information on fainter sources)
  • ROSAT all sky survey (TBD)
  • Chandra Deep Field (TBD)

• There has been one all sky gamma ray survey (Thompson et al 1995)
  EGRET on Compton-GRO found 130 sources with S > 10^-11 erg s^-1 cm^-2 at 30 - 3000 MeV;
  of these, 40% (~50) were AGN.

• It transpires that essentially all AGN are variable.
    variability may provide a relatively unbiassed selection method.
  Also, of course, AGN should have zero proper motion (!)
  Hence, a combined criteria will eliminate most variable star contaminators
  This method has been investigated (Majewski et al 1991), but not yet used in earnest.

(6) AGN Demographics

• Overall, how common are the various types of AGN?
  As with all astrophysical objects, it depends on luminosity: many more low luminosity AGN.
  Our ability to analyse AGN populations depends, to some extent, on the range in luminosity.

• First consider the spectroscopic surveys of complete samples of nearby galaxies.
  The most recent (prior to SDSS) is the Palomar survey by Ho, Fillipenko & Sargent.
    486 galaxies of all Hubble types, brighter than B_T = 12.5, north of = -2^o
  Of their sample, 86% have emission lines (EW(H) 1A); (100% of the spirals)
  Expressed in terms of emission types:
  • 40% have HII emission regions due to star formation
  • 30% are LINERs
  • 13% are Seyferts (with #Sy1 ~ #Sy 2)

  Combining LINERs & Seyferts:
    43% of all local galaxies could be considered active !
  This is much more than originally thought (e.g. in the 1970s)

• This may, however, be an overestimate:
    close examination of the LINERs suggests two types:
  • Strong [OI] LINERs ([OI]6300/H 0.1) are probably genuine weak AGN
  • Weak [OI] LINERs ([OI]6300/H 0.1) are probably photoionized by stars
  their ratio is ~2:1, making the total fraction of AGN closer to ~30% (check)

• A conservative estimate comes from the incidence of detectable broad H.
  • 40% of all Seyferts
  • 15% of all LINERs (xx% of strong [OI] LINERs; 0% of weak [OI] LINERs)
    10% of all galaxies have a broad H component
  This latter number is a strict lower bound to AGN frequency

• The high frequency of weak AGN is consistent with the widespread presence of nuclear SMBH.

• The spectroscopic surveys also find a link between emission type & Hubble type.
  As seen in these histograms: [image]
  • HII emission tends to be in late type Spirals, including lower luminosity galaxies;
    
  • LINERs & Seyferts tend inhabit early type Spirals, & avoid low luminosity galaxies
    Activity favours galaxies with relatively luminous bulges.
  Similarly, the most luminous early type galaxies (M_B -22) are ~all active.

• A very rough census of the local populations might go like this:
  (Note: 1 Gpc³ sphere radius 620 Mpc out to z ~ 0.20h 2 Gyr lookback time)

  Type	# Gpc-3
  Galaxies (excl. dwarfs)	108
  Luminous Spirals	107
  Weak AGN	106.5
  Markarian Seyferts	105
  Radio Galaxies	1000
  Radio Quiet QSOs	100
  Radio Loud QSRs	1

  Clearly, powerful AGN are locally very rare (e.g. nearest QSR = 3C273 @ 480h^-1 Mpc)
  At redshift ~2 (10 Gyr ago), luminous AGN were much more common, by ~×1000 (see sec 16)

• More precise discriptions are local AGN luminosity functions
  These are usually defined for a given band (radio; optical; X-ray; rarely bolometric)
  It is often necessary to combine different samples (e.g. faint & bright), and
  correct for various types of bias (see Topic4.4)

  Here is one version of the optical luminosity function (L,z=0): [figure]
  Of course, it is the change of with redshift that that helps define AGN evolution (sec 16)

• An alternative demographic is the number of AGN per square degree of sky (~5 full moons)
  This depends on AGN brightness & waveband, and in part measures a survey's relative efficiency.

  Here is N(<B) deg^-2 for QSOs, divided into z <2.2 and z > 2.2 [Fig 6.7 K&N + K fig3.5&3.6]
  For example, above 21^m in B, there are ~30 QSOs with z < 2.2 and ~8 with z > 2.2
  @ 22.5^m those numbers are 130 & 30, while @ 16^m there is ~1 QSO per constellation (10^o square).
  @ 13^m we have one QSO in the whole sky: it is 3C 273.

• Comparing with the other two bands at their approximate current limits :
  In X-rays, the ROSAT/EMSS survey finds ~100 AGN deg^-2 above 10^-14 erg/s/cm² (0.3-3.5 keV)
  In radio, one finds ~25 deg^-2 at 5 GHz above 1 mJy (where most sources are still AGN)
  While many of the X-ray sources will also be optical QSOs, the radio sources will not.
  Surveys in the three main wavebands yield comparable numbers of AGN at their current limits

(7) AGN Unified Schemes

• The huge diversity of AGN can be greatly simplified by first recognizing two intrinsic differences:
  1. a continuous range of luminosity, from very weak to exceedingly powerful
  2. a clear division into radio loud (L_GHz/L_opt ~ 10^-2) and radio quiet (L_GHz/L_opt ~ 10^-4)

• Can we gain further simplification by asking how appearance changes with viewing angle?
  The answer is yes...., to some extent.
  Below we identify various AGN pairings which may differ only by viewing angle
  Since viewing angle is essentially arbitrary, this "Unifies" the types involved
  

• Please note: no-one is suggesting that all AGN are identical !
  Just that there may not be quite as many sub-types as originally thought.

• It is important to realise that there are two different phenomena invoked in unification schemes

  (i) Angle Dependent Obscuration

• Near nuclear regions may have dense gas which blocks some lines of sight into/out of the nucleus.
    the ionizing radiation field may not emerge isotropically
    we may or may not see the active nucleus.

• What we can/can't see depends on the location and opacity of the obscuring material
  Perhaps surprisingly, dense molecular gas & dust is often found in active nuclei
  It is thought that this dense phase provides the necessary UV-optical opacity
  

• The obscuring region is thought to lie somewhere between the BLR and NLR : 1 - few pc out.
  Its geometry is uncertain, though it is often assumed to be in a thick disk/torus
  This naturally yields an axis, possibly aligned with the inner AGN axis.
  There may, of course, be other obscuring material further out, e.g. within nuclear dust lanes

• In the simplest case, three components are unaffected by the obscuring region
  1. Narrow Line Emission -- which originates outside the obscuring region
  2. Hard X-rays -- which penetrate the region
  3. Far IR -- which also penetrates the region (though probably not Near-IR or Mid-IR)
    we expect these to be emitted isotropically
  AGN selected using these properties should yield unbiassed samples with regards orientation
  Such samples are important in exploring the Unified schemes.

  (ii) Doppler Boosting of a Relativistic Jet

• Radio loud AGN are thought to have a relativistic jet.
  Emission from this jet will be Doppler boosted by the bulk flow in the forward direction.
  Depending on your viewing angle, you may see this emission brighter/unchanged/dimmer

• Boosting the jet base can increases the prominance of the flat spectrum core &/or VLBI jet
  Further out, the more extended (VLA, steeper spectrum) jet may also be boosted/suppressed

• However, the outermost (steep spectrum) radio lobes are not boosted (they move too slowly)
  Their emission is therefore thought to be isotropic
    samples selcted by this extended flux should be orientation independent.

• Obviously, Doppler boosting is only invoked for unification within radio loud AGN

  Note: Doppler boosting should not be confused with two typographically related terms:

  • Doppler bootsing -- a very fast style of tap dancing
    
  • Doppler boosing -- drinking so fast you turn blue

• This provides, perhaps, the clearest example of unification :
    at least some Seyfert 2s are in fact Seyfert 1s seen from the "wrong" direction.
  

• The model invokes angle dependent obscuration: a ~pc scale (equatorial) dense region
  This region blocks the nuclear radiation, which enters the NLR/ENLR anisotropically
  For Seyfert 2s, it is argued, we are in the blocked (equatorial) sight line
  There are a number of (~independent) lines of evidence for this scenario:

  (i) NLR/ENLR shapes

• NLR/ENLR shapes are studied by imaging AGN using narrow filters centered on emission lines.
  In fact, dividing the [OIII]5007 image by the H image reveals just the high ionization gas
  Here are some examples, most taken using HST (figs)

• For Seyfert 2s the emission regions are very often elongated, a number having (bi-)conical shape
  Conversely, Seyfert 1s show such geometry much less frequently
  This is exactly what you expect for an equatorially blocked ionizing radiation field.

• One needs to be careful: bipolar radio (weak) jets are often associated with NLR emission
    the ionizing radiation field is best probed using ENLR emission, beyond the radio source

• When cone geometry is observed, the cone opening angles are ~40-80^o
    ~25% of the sky witnesses the central source
  this is consistent with the ratio of Sy1/Sy2 in orientation independent samples (see below)

  (ii) NLR vs nuclear luminosity comparison

• Our model assumes that, for Seyfert 2s, the NLR sees a much brighter nucleus than we do.
    the NLR should be more luminous than we would guess from our view of the nucleus

• How do we estimate the (ionizing) luminosity entering the NLR ?   By using H
    one ionizing photon makes one recombination which makes _H / _B = 0.118 H photons.
  (the 's are recombination coefficients)
  So, if Q_ion ionizing photons emerge isotropically from the nucleus, & a fraction C_i enter the NLR:
    L_H = C_i Q_ion × 0.118 × 4.08 × 10^-12 = 5.11 × 10^-13 C_ion Q_ion erg s^-1

• Let's model the central source as a power law: L ^-, or equivalently L = L_ref (/_ref)^-2
  where L_ref is the luminosity in erg s^-1 A^-1 at a reference wavelength _ref
  Choosing _ref = 912A and integrating from 912 to
    the total number of ionizing photons is Q_ion = 4.18× 10¹³ × (L₉₁₂ / )
  Hence, we expect L_H = 21.4 × C_i × (L₉₁₂ / )   erg s^-1

• Is this in fact what we find?
  The comparison is tricky for a couple of reasons:
  the UV continuum is not easily accessible
  the optical continuum is confused by starlight, especially in Seyfert 2s
  C_i, the NLR covering factor, is uncertain: maybe ~25% from cone angle? (but not > 1 !)
  The few UV observations available suggest: Seyfert 1s obey the relation while Seyfert 2s do not.

• Extrapolating the continuum into the optical, we have: L₄₈₀₀ = L₉₁₂ (4800/912)^-2
    the H equivalent width: EW(H) = L_H / L₄₈₀₀ = 21.3 × 5.3^2- × C_i   Angstrom
    EW(H) 113 C_i   Ang. ( 1; and EW is w.r.t. the non-stellar optical continuum)
  This is, in fact, true for the BLR emission in Seyfert 1s, but is too low for Seyfert 2s
    since C_i cannot be > 1, we conclude we don't see the nuclear continuum in Seyfert 2s.

• An alternative estimate of the total nuclear UV output is the reprocessed FIR emission:
    L_UV L_FIR 4d² × 1.8 × 10^-11 [13.5S₁₂ + 5.2S₂₅ + 2.6S₆₀ + S₁₀₀] erg s^-1   (Topic 11.2)
  As before, the IRAS luminosities imply much larger UV luminosities than we in fact see.

  (iii) Q(nuc) from ENLR Radiation Parameter

• A similar analysis uses ENLR emission lines to estimate the UV flux impinging on it.
  The [SII]6717/6731 line ratio gives the electron/gas density in the ENLR
  other lines (eg [OIII]5007/[OII]3727 vs [OIII]5007/H) give the radiation parameter, U,
  where U = Q_ion / (4R² c n_e), and R is the separation between ENLR and the nucleus:
    knowing R and n_e, we find Q_ion, which can be compared with the observed nuclear UV flux.
    Again, it is clear we do not see the full UV radiation field in Seyfert 2s.

• the virtue of this analysis is that it is independent of covering factor
  the drawback is that it uses a (potentially uncertain) link between line ratios and ionizing flux.

  (iv) Periscopic (Polarization) View of the Nucleus

• This is the most impressive demonstration that some Seyfert 2s hide a Seyfert 1 nucleus.
  light emerging along the polar axis can be scattered into our line of sight
  if we could see just the scattered light, we would be looking down the polar axis, like a periscope.
  Since scattering introduces linear polarization (E vector perp to radial vector):
    do spectropolarimetry and extract just the polarized (scattered) light.

• Famous example: NGC 1068, a classic Seyfert 2:
  The spectrum of polarized light shows a Seyfert 1 spectrum ! (image)
  A centro-symmetric polarization pattern pinpoints the location of the hidden AGN
  The amount of polarization is independent of wavelength :
    scattered by free electrons (Thomson cross-section independent of in optical & UV).

  In NGC 1068 there is a second mirror located ~300pc out: an HII region containing dust
  This also yields a Seyfert 1 spectrum in polarized light, but with slight differences:

  it rises strongly into the blue, confirming dust as the scatterer
  the broad H profile is slightly narrower & blueshifted by ~400 km/s w.r.t. the first mirror
    the electrons have T ~ 3×10⁵ K which adds thermal broadening to the first mirror
    the electrons in the first mirror are outflowing in a 400 km/s wind.

• So far ~half the Seyfert 2s studied show broad H in polarized light
  The fraction could increase with better data (difficult work, often only ~1-few% net polarization).
  Still unclear whether there are any pure Seyfert 2s with no BLR or strong blue nucleus

• In general, polarization angles in Seyfert 2s are perpendicular to their radio axes.
    consistent with radiation emerging along radio axis, some of which is scattered.
    obscuration & black hole (jet) axes may be the same.
    supports obscuration as thick equatorial disk, aligned with inner accretion disk.

  (v) Nuclear Absorption: X-ray spectra

• Is there any direct evidence for a near-nuclear obscuring region?
  In fact there is, and it comes from the X-ray spectra of Seyfert 1s and Seyfert 2s
  Recall that with columns 10²² cm^-2 soft X-rays (0.2-4 keV) suffer significant absorption
  Seyfert 2s tend to have X-ray spectra with soft absorption, while Seyfert 1s don't.
  The derived columns are large: 10²² - 10²⁴ cm^-2 corresponding to A_V ~ 5 - 500.

• Hard X-rays (2-10 keV) lie above the photoelectric edges, and aren't absorbed at these columns
  Do the hard X-ray spectra of Seyfert 2s resemble those of Seyfert 1s ?
  Yes, they have the same slope and their luminosities approach those of Seyfert 1s
  Also L(hard X)/L(UV-opt) is high in Seyfert 2s, because we aren't seeing the UV-opt emission.

• A couple of interesting details:
  
  • The soft X-rays are often thought to be scattered; showing strong FeK periscopically
    
  • In NGC 1068 the column is ~10²⁵ and even the hard X-rays are blocked (Thomson thick)
      it is anomalously week in both soft and hard X-rays
      it shows no soft absorption
      it is thought to be all scattered
      at 1% scattering efficiency, we recover a Seyfert 1 luminosity ~10⁴⁴ erg s^-1

• AGN samples selected by hard X-rays are thought to be ~unbiased w.r.t. orientation

  (vi) Nuclear Absorption: Molecular Masers

• Conditions for Masers: long low v path through dense gas exposed to strong pumping radiation
  These conditions are met in many AGN nuclei; the most famous is NGC 4258 [Topic 13.4e]
  The path lengths will be at their largest for edge on dense gas disks.

• Surveys of Seyferts find many H₂O masers, but only Seyfert 2s
    consisent with expectations for obscuring tori
  Unfortunately, few/none have the clean geometry of NGC 4248 allowing a BH mass determination.

  (vii) Opening Angle

• A further check of the anisotropic absorption picture is the following:
  Cone angles seen in ENLR's (~40-60^o) predict relative numbers of Sy2 to Sy1 of ~2 to 1
  Is this true for orientation independent samples?

• Broadly speaking, yes, though reliable samples are hard to come by.
  • Selected by Far-IR, gives #Sy2/#Sy1 ~ 2 (Kinney et al 1996 check)
  • Selected by [OIII]5007, gives #Sy2/#Sy1 ~ 5   (Salzer 1989)
  • Selected by host galaxy (ie spectroscopic surveys), gives #Sy2/#Sy1 ~ 1 (Ho et al)
  One concludes that the opening angle is significant, though perhaps less than 50% of the sky.

• Can this estimate be repeated at QSO luminosities? Not really, finding QSO-2s is very difficult.
  It isn't yet clear whether the apparent lack of QSO-2s is because the opening angle widens to ~4
  or because obscured QSOs would fail to be found in, e.g., color & slitless surveys
  FIR surveys do find buried AGN, though often the story is muddled by strong star formation.

  (viii) Do All Seyfert 2s have Seyfert 1 Nuclei?

• dunno

(8) Accretion Power

• The combination of high luminosity and variability places strong limits on AGN energy sources
    they must be very compact and very efficient

• Hoyle & Fowler (1963) first recognized that gravitational collapse might power AGN engines.
  Zeldovich and, separately, Salpter (1964) proposed accretion onto a massive black hole
  This became "the AGN paradigm", especially after a nice paper by Lynden-Bell (1974).
  

• Real accretion is likely to be complex and time varying, with many associated physical processes.
  Our discussion will be somewhat idealized -- but it provides a likely overall framework.

• There are a few inter-related themes:
  1. A natural maximum accretion rate limits AGN luminosity and black hole growth rate
  2. Even a little angular momentum will ensure the accreting gas forms an accretion disk
  3. Turbulent viscosity allows gas to lose angular momentum, spiral in, & release energy.
  4. Black hole spin sets the disk's inner radius which in turn sets the disk's net luminosity.
  5. Finally, it seems that initial ("seed") black holes will inevitably form in galaxy nuclei.

• Like people, black holes cannot gulp down food with arbitrary speed.
  One limitation, which primarily applies to spherical accretion, is due to Eddington (date).

• There is a feedback valve which regulates the accretion flow.
    accretion luminosity generates an outward radiation pressure on the surrounding matter
    this counters the inward gravitational force.
    if the two are equal, the accretion flow is shut down

• The standard (oversimplified) scenario considers spherical accretion:
  A radiation field of energy flux F (erg s^-1 cm^-2) has momentum flux (ie pressure) F/c
  [note units of F/c: erg cm^-3 energy density (U_rad) pressure (P_rad)]
  For a single electron, the outward radiation force is:
  
  F_rad = _T F/c = _T L/4r²c
  where _T = 6.65 × 10^-25 cm^-2 is the Thomson cross section (~independent of E up to ~½MeV)

• The electrons are strongly tied to the protons, on which gravity principally acts:
  
  F_grav = GMm_p/r²
  Hence the limiting Eddington luminosity, below which gravity "wins" (ie F_grav = F_rad) is:
  L_Edd = (4Gcm_p)/_T × M   =   1.26 × 10³⁸ M/M erg s^-1   =   1.26 × 10⁴⁶ M₈ erg s^-1

    Spherically accreting black holes of mass M cannot have luminosities above L_Edd
    Conversely, accretion powered QSOs demand massive (~10⁸ M) central objects

• We may tie this to a maximum (Eddington) accretion rate via the accretion efficiency,
  A source with efficiency radiating at L_Edd, accretes matter at a rate:
  
  (dM/dt)_Edd = ^-1 L_Edd/c²   =   1.4 × 10²⁶ _0.1^-1 M₈ gm s^-1   =   2.2 _0.1^-1 M₈   M yr^-1

• Do black holes accrete at these maximum rates ?
  Estimates of M_BH typically find sub-Eddington luminosities: L 0.1 L_Edd

• The accretion rate provides a framework for studying black hole growth times
  Allowing for sub-Eddington accretion rates, L/L_Edd, we have:
  dM_BH/dt = 2.2 _0.1^-1 (L/L_Edd) × 10^-8 M₈   M yr^-1
  yielding exponential growth with e-folding growth times:
  t_growth   4.5 × 10⁷ _0.1 (L_Edd/L) yr

  Thus, growth from ~10³ to 10⁸ M is ~10 e-foldings
    0.5 - 5 Gyr for L/L_Edd = 1 - 0.1   which seems reasonable for most AGN

• Some high-z high luminosity QSOs may demand super-Eddington accretion
  This is not too problematic.
  1. Radiate above L_Edd by breaking spherical symmetry: equatorial accretion & polar emission.
  2. Accrete above Mdot_Edd: optically thick flow traps radiation which is also swallowed.

• Because the gravitational potential is so steep at small r, most energy is released in the inner disk.
  For this reason, the location of the innermost radius of the disk is crucially important
  For finite sized central objects, this is usually the object's surface radius
  For black holes, the inner disk radius depends on the black hole's spin (it is NOT just the horizon)

  (i) Newtonian Point Mass Potential

• First consider a Newtonian point mass potential:
  Let's define the gravitational radius: r_g = GM/c² = distance where PE_grav = -mc² = -(rest mass)
    one would need a total conversion of rest mass to energy to escape     not possible.
    r_g is very small: r_g = 1.5 km M/M = 1.0 AU M/10⁸M = 5 mm M/M_E

• Escape velocity v_esc = (2GM/r)     v_esc = c at a radius of 2GM/c² = 2r_g
  Michel (1750) recognizes this as an important radius: light can't escape from inside 2r_g
  By 1920s, this re-emerges as an event horizon for Schwarzchild black holes: r_s = 2r_g.

• Consider planar orbits in this Newtonian potential. (c.f. discussion in Topic 6.2c)
  The radial acceleration equation is (centrifugal - gravity):

  (15.1)

  Using the angular momentum per unit mass, h = r×v = r v, we can rewrite this as:

  (15.2)

  The effective potential, _eff, allows us to describe the radial motion in 1-D form [image]

• For Newtonian, _N,eff is zero at large r, slowly drops to a minimum before rising steeply at small r
  Stable circular orbits are found at the minimum in _eff (lowest energy for given h)
  The inner steep term imposes an unsurmountable angular momentum (or centrifugal) barrier
  Only when h=0 (a radial orbit) can the trajectory strike the origin.

• Anticipating the relativistic regime, let's re-express 14.2:
  • include rest mass, so at large r the potential energy per unit mass c² (not zero)
  • divide by c² to scale all energies relative to rest mass
  • express angular momenta relative to r_gc, using H = h/r_gc per unit mass

  (15.3)

  We are now ready to compare this with the effective potential for black holes.

  (ii) Schwarzchild (Non-Rotating) Potential

• Schwarzchild (1916) derived the metric for the space-time surrounding a point mass:

  (15.4)

  Notice the discontinuity at r = 2r_g, and sign change of A when r < 2r_g
  This radius is called the Schwarzchild radius: r_S = 2r_g
  It marks an event horizon: inside r_S all light cones point inwards     a "boundary of no return"

• The relativistic effective potential per unit mass is given by:

  (14.5)

  Try expanding this in the limit r >> r_g and you recover 14.4, the relativistic version of _N,eff.

• Comparing _S,eff with _N,eff we notice several important differences: [see image]
  1. The -(r_g/r)³ term ensures an inner turnover     the central AM barrier is finite
       below a certain peri-center distance an orbit is pulled in.
  2. For H² > 12, _S,eff has a minimum just like _N,eff, allowing stable circular orbits
       their radii are: r_co = ½r_g [H² + (H⁴ -12H²)]
  3. For nearly circular orbits, the radial oscillation frequency is: ² = (d²_eff/dr²)_rco   [ Topic 6.2c]
     Unlike the Newtonian case, this does not equal the orbital frequency
       the orbits do not close, but instead precess
       this is the origin of the famous precession of the perihelion of Mercury's orbit.
  4. For H² < 12, there are no minima or maxima, so no persistent orbits exist
       at H_crit = 23, the last circular orbit has radius r_lco = ½r_gH² = 6r_g = 3r_S
       circular orbits with r < 3r_S simply don't exist, they quickly spiral inwards.

• In an astrophysical context, gas enters the black hole via an accretion disk
  Viscous forces cause gas to slowly migrate to smaller circular orbits, releasing energy on the way
  Finally, at r = 3r_S the gas plunges into the black hole without dissipation of much further energy.
    the binding energy at 3r_S defines the total net accretion efficiency (not at r_S)
    inserting r = 3r_S and H = 23 into 14.5 gives _S,eff /c² = 0.943
    the amount lost is: _S,eff() - _S,eff(3r_S) = (1.0 - 0.943)c² = 0.057c²
    simple accretion onto a Schwarzchild black hole is ~5.7% efficient.
  (The simple Newtonian estimate at 3r_S is ½ GM/3r_s = 8.3% )

  (iii) Rotating Black Holes

• Perhaps surprisingly, black holes can have angular momentum: investigated by Kerr (1963)
  In this case the metric shows "frame dragging"     a "whirlpool" of space
    retrograde pointing rockets must fire to keep "stationary" as seen from afar
  

• In addition to the horizon, there is a new "surface": the stationary limit
  Inside the stationary limit even light cannot move in a retrograde direction ("flow" of space > c)
  The surface has an oblate axisymmetric shape:
  r_sl = r_g × [1 + (1 - a²cos²)]     [see image]
  where a measures the BH AM per unit mass: a = (J_BH/M_BH)/(cr_g)   scaled w.r.t.   c r_g
  (note: a is sometimes defined as a length (Peacock) or mass (Krolik); here it is dimensionless)

• Meanwhile, the true horizon shrinks from the Schwarzchild value of 2r_g to:
  r_h = r_g × [1 + (1 - a²)]
  Note that r_h is spherical, and meets r_sl at the poles.

• Plots of _K,eff (not shown) for a rotating BH are similar to those of the Schwarzchild case:
    circular orbits are allowed down to a last minimum radius, r_lco
    as AM increases, r_lco decreases (increases for retrograde).
  There is a hard upper limit to AM: a = 1, for which r_lco = r_g   (9 r_g for retrograde)
  Inserting a = 1 and r = r_g into _K,eff yields a binding energy of 42% mc²
  (the simple Newtonian efficiency is ½GM/r_g = 50%)
  In reality, optimum conditions can only create a = 0.998, which drops the yield to ~30%.

  Thus, the range a = 0 0.998 yields accretion efficiencies 6 - 30%
  Hence, the often used value, 10%

• These efficiencies can yield enormous luminosities with only modest accretion rates:
  consider:   1 M yr^-1 = 2 × 10³³ / 3.17 × 10⁷ = 6.3 × 10²⁵ gm s^-1   (M_E every 2 minutes)
  giving:   5.7 × 10⁴⁵ _0.1 M/M yr^-1   erg s^-1   ( 1.4 × 10¹² L     100 L galaxies)
    easily enough to power QSOs.

  Whether black holes can actually gorge themselves so fast is discussed below.

(i) Nuclear ISM

• TBD

  (ii) Tidally Disrupted Stars

• Black hole sizes grow in proportion to their mass: r_h     M_BH
  This is unusual : constant density solids have r     M^1/3,   degenerate matter has r     M^-1/3
    black holes get "bigger" quickly

• Their mean density drops: <_BH>     M / (4/3r_h³)     1.7 M₈^-2 gm cm^-3 (Schwarzchild hole).
  Obviously, for 1-few M the mean density is huge, but SMBH are about like water !

• Tidal stretching forces behave in a similar way:
  Newtonian analysis: T = dF/dr = 2GM/r³   10^-6 M₈^-2   N m^-1   at the horizon
  Again, for 1-few M, T ~ few million tons per meter; while for an SMBH it is imperceptible
    you could happily fall into an SMBH and survive much of the journey to the singularity.

• Of more astrophysical interest: are stars tidally destroyed before or after entering the horizon?
  
  Before:   the material will enter the accretion disk and liberate energy on the way in
  After:     the star is already lost and its infall goes essentially unnoticed.

• The Roche criteria for tidal breakup of a self-gravitating object is discussed in Topic 12.3bi
  Breakup occurs if the mean density within the orbit is greater than the mean density of the orbiter
  Now, the mean black hole density within r_h is 1.7 14 M₈^-2   gm cm^-3 (Schwarzchild Kerr).
  Whereas A0 M0 main sequence stars have <>     0.2 2     0.3 3 gm cm^-3
    Black holes with M few × 10⁸ will start swallowing main sequence stars whole.

  However, giant stars are 10 - 100 times R, with <>   ~ 10^-4 - 10^-5
    Giant stars, while much rarer, are essentially always disrupted

• Previous estimates of engine efficiency, , express this relative to the infalling mass.
  One factor which limits is the final KE of the accreted mass: it is lost to the black hole.
  Is there a way to recover any of this energy, perhaps by "slowing" down the hole?

• First let's see how much energy is actually available.
  The mass/energy of a rotating hole is in two forms: rotational energy & irreducible mass
  For a maximal Kerr hole (a = 1) it is found that M_irr = M_BH / 2 0.71 M_BH
  Hence, the total available for extraction is 29% of M_BH: a significant amount!

• A purely dynamical process was suggested by Penrose (1969)
  Between r_h and r_sl lies the ergosphere: "ergo" = work/energy
  Within this region, imagine throwing an object "upstream" in a retrograde direction
    space pushes everything prograde: the object falls in & you recoil and escape
    you slowed the BH (a bit) and extracted some of its rotational energy!
  Unfortunately, the high "throw speed" required by this process is unlikely to occur naturally.

• An MHD process was suggested by Blandford & Znajek (1977)
  TBD

TBD : Rees diagram:

(9) Continuum SEDs

(10) Emission Line and Ionized Cloud Physics

(11) The Broad Line Region

(12) The Narrow Line Region

(13) Intrinsic Absorption Lines

(14) AGN Radio Properties : Cores, Jets, Lobes

(15) AGN Host Galaxies and Environment

(16) AGN Evolution