chapter 14 black holes as central...

Chapter 14

Black Holes as Central Engines

Black holes imply a fundamental modification of our un-derstanding of space and time. But at a more mundanelevel they also are of great practical importance in astron-omy because they can be extremely efficient sources ofenergy. As we shall now discuss, rotating black holes (ap-proximated by the Kerr solution) are believed to be the en-gine powering a whole set of phenomena associated withquasars, active galaxies, and gamma-ray bursts.

333

334 CHAPTER 14. BLACK HOLES AS CENTRAL ENGINES

14.1 Black Holes as Energy Sources

Since black holes have event horizons that cut their interiors off fromcommunicating with the outside world, they might seem unlikelypower sources.

1. However, the discussion of Penrose processes in the precedingchapter demonstrates conceptually that it is possible to extractenergy from rotating classical black holes.

2. Penrose processes themselves are not likely to lead to large ex-traction of energy from black holes in astrophysical environ-ments.

However, there are two general classes of phenomena that canleadto large energy release for black holes in regions outside the eventhorizon, where the released energy might be accessible to externalprocesses.

1. Accretion onto either Schwarzschild or Kerr black holes canconvert large amounts of gravitational energy into other usableforms of energy before the matter falls through the event hori-zon.

2. If the black hole is rotating, it may be accompanied by strongexternal magnetic fields that can tap the rotational power oftheblack hole to accelerate charged particles to high velocityin rel-ativistic jets along the axes of rotation, outside the eventhorizon.

14.1. BLACK HOLES AS ENERGY SOURCES 335

A consideration of plausible formation mechanisms forblack holes suggests that it is very unlikely that they formwith zero angular momentum, and that they are likely tohave no net charge.

• Thus practically we may focus our attention on Kerrblack holes.

• The Schwarzschild solution may then be viewed asthe limit of the Kerr solution for the special case ofsmall angular momentum.

Let us now turn to an overview of extracting energy fromrotating black holes, either by accretion, or by coupling oftheir rotation to external fields.


14.2 Accretion and Energy Release for Black Holes

In astrophysics, accretion of matter onto a gravitating object can be alarge source of power because it is an efficient mechanism forgravi-tational energy conversion.

• This is particularly true if the accretion is onto a compactobjectsuch as a white dwarf, neutron star, or black hole.

• Accreting particles typically form anaccretion diskof particlesorbiting the object (angular momentum conservation).

• Collisions of the particles in the accretion disk heat it and causeit to radiate energy away, allowing particles to spiral inward andaccrete onto the central object.

• For white dwarfs or neutron stars, the inspiraling material accu-mulates on the surface (which can lead in some cases to violentexplosions).

• For black holes there is no surface but the event horizon clearlysets a boundary for energy extraction.

• No energy can be obtained externally once accreting materialfalls through the event horizon of a black hole.

• However, calculations indicate that large amounts of accretionenergy (that is, gravitational energy) can in principle be extractedfrom in-spiraling matter before it falls through the event horizon,either through emission of

1. Radiation, or of

2. Relativistic jets of matter (which may involve strong mag-netic fields).

14.3. MAXIMUM ENERGY RELEASE IN SPHERICAL ACCRETION 337

Table 14.1: Energy released by accretion onto various objects

Accretion onto Max energy released (erg g−1) Ratio to fusion

Black hole 4.5×1020 75

Neutron star 1.3×1020 20

White dwarf 1.3×1017 0.02

Normal star 1.9×1015 10−4

14.3 Maximum Energy Release in Spherical Accretion

The most spectacular consequence of accretion is that it isan efficient mechanism for extracting gravitational energy.

• The energy released by accretion is approximately

∆Eacc = GMmR

,

whereM is the mass of the object,R is its radius, andm is the mass accreted.

• In Table 14.1 the amount of energy released per gramof hydrogen accreted onto the surface of various ob-jects is summarized (see Exercise).

• From Table 14.1, we see that accretion onto verycompact objects is a much more efficient source ofenergy than is hydrogen fusion.

• But accretion onto normal stars or even white dwarfsis much less efficient than converting the equivalentamount of mass to energy by fusion.


Let us assume for the moment, unrealistically, that all ki-netic energy generated by conversion of gravitational en-ergy in accretion is radiated from the system (we addressthe issue of efficiency for realistic accretion shortly). Thenthe accretion luminosity is

Lacc =GMM

R≃ 1.3×1021

(

M/M⊙

R/km

)(

Mg s−1

)

erg s−1,

if we assume a steady accretion rateM.


Table 14.2: Some Eddington-limited accretion rates

Compact object Radius (km) Max accretion rate (g s−1)

White dwarf ∼ 104 1021

Neutron star ∼ 10 1018

14.3.1 Limits on Accretion Rates

The Eddington luminosity is

Ledd =4πGMmpc

σ,

with σ ithe effective cross section for photon scattering.

• For fully ionized hydrogen, we may approximateσby the Thomson cross section to give

Ledd≃ 1.3×1038(

MM⊙

)

erg s−1.

• If the Eddington luminosity is exceeded (in whichcase we say that the luminosity issuper-Eddington),accretion will be blocked by the radiation pressure,implying that there is a maximum accretion rate oncompact objects.

• EquatingLacc andLedd gives

Mmax≃ 1017(

Rkm

)

g s−1

Eddington-limited accretion rates based on this formulaare given in Table 14.2.


14.3.2 Accretion Efficiencies

• For the gravitational energy released by accretion to be extracted,it must be radiated or matter must be ejected at high kinetic en-ergy (for example, in AGN jets).

• Generally, we expect that such processes are inefficient and thatonly a fraction of the potential energy available from accretioncan be extracted to do external work.

• This issue is particularly critical when black holes are the cen-tral accreting object, since they have no “surface” onto whichaccretion may take place and the event horizon makes energyextraction acutely problematic.

• Let us modify our previous equation for accretion power by in-troducing an efficiency factorη that ranges from 0 to 1:

Lacc = 2ηGMM

R.

• Specializing for the black hole case, it is logical to take theSchwarzschild radius (the radius of the event horizon for a spher-ical black hole), which is given by

Rsc =2GM

c2 = 2.95

(

MM⊙

)

km,

to define the “accretion radius”, since any energy to be extractedfrom accretion must be emitted from outside that radius.


• Then for a spherical black hole

Lbhacc= η

GMMRsc

=2ηGMM

2GMc2 = ηMc2,

andη measures the efficiency for converting rest mass to energyby the accreting black hole power source.

• For hydrogen fusion, the mass to energy conversion efficiency isη ∼ 0.007.

• For compact spherical objects like Schwarzschild black holes orneutron stars, reasonable estimates suggestη ∼ 0.1.

• Shortly we shall see that forKerr black holesit is possible tobe more efficient in energy extraction, basically because thereare more ways to drop mass onto the black hole and have partof it or its emitted radiation escape carrying energy and angu-lar momentum extracted in the interaction than is possible for aspherical black hole.

• For rotating black holes, efficiencies ofη ∼ 0.3–0.4might bepossible.


It is the high efficiency available from accretion onto a su-permassive black hole as a power source that provides themost convincing general argument that AGN and quasarsmust be powered by accretion onto rotating supermassive(M ∼ 109M⊙) black holes.

• For example, in an Exercise you are asked to use ob-served luminosities and temporal luminosity varia-tion to show that a quasar could be powered by ac-cretion of as little as several solar masses per yearonto an object of mass∼ 109M⊙, and that

• this mass must occupy a volume the size of the SolarSystem or smaller.


14.3.3 Innermost Stable Circular Orbit and Binding Energy

The property of particle orbits in the Kerr metric that isprobably of most interest in astrophysics is the bindingenergy of the innermost stable circular orbit, since this isrelated to the energy that can be extracted from accretionon a rotating black hole.

If we denote the radius of the innermost stable orbit byR, for circularorbits we have thatdr/dτ = 0 and from earlier equations of motionin the Kerr metric

ε2−12

=12

(

drdτ

)2

+Veff(r,ε, ℓ) −→ε2−1

2= Veff(R,ε, ℓ).

Furthermore, to remain circular the radial acceleration must vanish,

∂Veff

∂r

∣

∣

∣

∣

r=R= 0,

and to be a stable orbit the potential must be a minimum, whichre-quires the second derivative to be positive,

∂ 2Veff

∂ r2

∣

∣

∣

∣

r=R≥ 0,

with equality holding for the last stable orbit.

This set of equations may be solved to determine the in-nermost stable orbit and its binding energy.


For extremal black holes (a = M), one finds that

1. Co-rotating orbits (accretion orbits revolving in thesame sense as the rotation of the hole) are more stablethan the corresponding counter-rotating orbits.

2. For co-rotating orbits,

ε =1√

3ℓ =

2M√

3R= M.

3. The quantityε is the energy per unit rest mass mea-sured at infinity, so the binding energy per unit restmass,B/M, is given by

BM

= 1− ε.

4. These equations imply that the fractionf of the restmass that could theoretically be extracted is

f = 1− ε = 1−1√

3≃ 0.42.

for a transition from a distant unbound orbit to theinnermost circular bound orbit of a Kerr black hole.


Therefore, in principle 42% of the rest mass of accretedmaterial could be extracted as usable energy by accretiononto an extreme Kerr black hole.

• One expects less efficiency than this for actual situa-tions, but optimistically one might expect as much as20–30% efficiencies in realistic accretion scenarios.

• Compare with the maximum theoretical efficiency of6% for accretion on a spherical black hole.

• Compare with the 0.7% efficiency for hydrogen fu-sion in the conversion of rest mass to usable energy.

• Accretion onto rotating black holes is a very efficientmechanism for converting mass to energy.


14.4 Jets and Magnetic Fields

A magnetic field cannot be anchored to the black hole itself becausethe event horizon pinches off any magnetic field lines that cross it.

• However, the whirling plasma of the accretion disk lies outsidethe event horizon and it could have a strong magnetic field.

• The accretion of matter by the rotating black hole can lead toejection at velocities approaching the speed of light alongthepoles of the black hole rotation for the portion of the matterthatdoes not cross the event horizon.

• The rotating magnetic field of the accreting disk can be carriedaway with the ejected matter, leading to bipolar jets perpendicu-lar to the accretion axis.

• These jets contain charged particles moving at near light velocityand twisting and spiraling magnetic fields.

• The magnetic fields probably are essential to focusing and con-fining the relativistic jets into the narrow cones that are observed,but the details of how this happens are not well understood.

• If the jets are not at right angles to our line-of-sight, we refer tothe one pointed more toward us as thejet and the other as thecounterjet.

• Because of relativistic beaming effects (such as those discussedin Box 14.1 and Exercises 4.8 and 14.3), which are a commonoccurrence when relativistic jets are oriented near the line ofsight, the counterjet may be faint and difficult to see relative tothe jet.

14.4. JETS AND MAGNETIC FIELDS 347

Box 14.1 Relativistic Jets and Apparent Superluminal Velocities

BA

B'

θ∆ϕ

~ d

d v (t - t )12

t 1

t 2

A distant source at B moves at a velocityv ∼ c toward B′. At time t1 the sourceat B emits a light signal that is detected at timet ′1 by an observer at A. When thesource reaches B′ at timet2, it emits another light signal that is detected by observerA at time t ′2. In Exercise 4.8, Lorentz invariance is used to show that theapparenttransverse velocityβT = vT/c is related to the actual velocityβ = v/c by

βT ≡vT

c=

β sinθ1−β cosθ

.

Thus for actual velocityβ < 1, the apparent transverse velocity can exceedc byany amount, since the maximum valueβ max

T = β/(1−β 2) is unbounded asβ → 1.This illusion of an apparent velocity exceeding that of light is observed frequentlyfor jets in radio astronomy where it is calledsuperluminal motion.The figure below

1992

1993

1994

1995

0 1 2 3 4Milli-Arcseconds

0 20Light Years

Tim

e

40 60 80 100

3C 279

Radio Jet

shows a radio jet in the blazar/quasar 3C 279 (redshiftz= 0.534) exhibiting appar-ent superluminal motion, with an inferred transverse velocity ∼ 7c (Exercise 14.1).


The charged particles in the jet, which are primarily elec-trons since they are more easily accelerated than moremassive particles, spiral around the field lines at relativis-tic velocities and emitsynchrotron radiationbecause ofthe accelerated spiral motion (see Box 14.2).

14.4. JETS AND MAGNETIC FIELDS 349

Box 14.2 Nonthermal Emission

The Planck law describesthermal emission,which is characterized by the emissionof radiation from a hot gas that is in approximate thermal equilibrium; the resultingspectrum is ablackbody spectrum.The characteristic Planck law curves for ther-mal emission peak at some wavelength, and fall off rapidly atlonger and shorterwavelengths, with the position of the peak moving to shorterwavelength as thetemperature of the gas is increased (the Wien law). Light from most stars, and lightfrom normal galaxies, is dominantly thermal in character.

In some cases we may observe emission ofnonthermalradiation, which has aspectrum that increases in intensity at very long wavelengths. The most commonform of nonthermal emission in astronomy issynchrotron radiation,where high-velocity electrons (or other charged particles) in strong magnetic fields follow a spi-ral path around the field lines, radiating their energy in theform of highly-beamedand highly-polarized light.a The figure below left contrasts a thermal (blackbody)spectrum characteristic of 6000 K and nonthermal emission,and the figure belowright illustrates the synchrotron radiation mechanism.

Spiralelectron

path

Beamedradiation

Beamedradiation

Magnetic field line

Synchrotron Radiation

Wavelength

Log R

ela

tive F

lux

Blackbody

(6000 K)

Nonthermal

RF

IRVis

The wavelength of the emitted synchrotron radiation is related to how fast thecharged particle spirals in the magnetic field. Thus, as the particle emits radiation,it slows and emits longer wavelength radiation. This explains the broad distributionin wavelength of synchrotron radiation as compared with thermal radiation.

Nonthermal emission is less common than thermal emission inastronomy, butthe presence of a nonthermal component in a spectrum typically signals violent pro-cesses and large accelerations of charged particles. High-frequency synchrotron ra-diation also implies the presence of very strong magnetic fields, since the frequencyincreases with tighter electron spirals, which are characteristic of strong fields.

aSynchrotron radiation is strongly polarized because it is emitted in a narrow beam inthe local plane of the electron’s spiral path, and the motionof the electrons as viewed fromthe side of the spiral is almost side to side in a straight line; see the figure above.


The resulting synchrotron radiation has a nonthermal spectrum and ispartially polarized.

• It is strongly focused in the forward direction by relativistic beam-ing.

• Fluctuations of the jet in time will also be compressed intoshorterapparent periods by relativistic effects.

• For an observer in the general direction of a jet, these effects willexaggerate both the apparent intensity and the time variation ofthe nonthermal emission.

• Thus, the nonthermal part of the continuum emission originateslargely in the synchrotron radiation produced in the jets .

• The thermal continuum is typically produced in the accretiondisk and the surrounding matter that it heats.

14.5. QUASARS 351

14.5 Quasars

In the 1950s astronomers began to catalog systematically objects inthe sky that emitted radio waves.

• As these radio sources were cataloged, an effort was made tocorrelate the objects emitting radio waves with sources visiblein optical telescopes.

• The resolution of the single-dish radio telescopes in use at thetime was much poorer than that of large optical telescopes, sothere often were many possible optical sources that might poten-tially be correlated with the fairly uncertain position of aradiosource.

• Nevertheless, some progress was made in these identifications.


14.5.1 “Radio Stars” and a Spectrum in Disguise

Most radio sources were found to be correlated with certain galax-ies but a few of the optical sources identified with the radio sourcesappeared to be points with no obvious spatial extension, just as onewould expect for stars.

• These were often called“radio stars” , but they had very strangecharacteristics for stars.

• In 1963, the first two of these radio stars were associated withthe radio sources 3C48 and 3C273, respectively.

• Although these objects had the appearance of stars in opticaltelescopes, they had spectra unlike any stars that had ever beenobserved.

1. There is a very strong continuum across a broad range ofwavelengths that is non-thermal in character.

2. Sitting on top of this continuum are emission lines, but theyare very broad, and their wavelengths do not correspond tothe lines for any known spectra.

14.5. QUASARS 353

Later in 1963, Dutch astronomer Maarten Schmidt found whilestudy-ing the spectrum of 3C273 that the strange emission lines were reallyvery familiar spectral lines in disguise.

• They were lines of the Balmer series of hydrogen (and a line inionized magnesium), but redshifted by a very large amount.

• The redshift of 3C273 corresponds to a recessional velocity about15 percent of the speed of light, if interpreted as a Doppler shift(later we shall see that such cosmological redshifts shouldnotbe interpreted as Doppler shifts, though).

• Once this was realized for 3C273, it quickly became apparentthat the spectrum of 3C48 could be interpreted in the same way,but with a redshift that was even larger.

• The reason that it took some time to arrive at this interpretationof the spectra for these objects is that

1. They were thought initially to be relatively nearby starsbe-cause they seemed to be pointlike.

2. Thus no one had any reason to believe they should be stronglyredshifted.

These objects were namedquasistellar radio sources,which was quickly contracted toquasars.


14.5.2 Quasar Characteristics

Soon other quasars were discovered and it became apparent that theyrepresented a class of objects with the following characteristics.

• Quasars appeared to be star-like in the initial images, butmorecareful study shows fuzziness and faint jets associated with somequasars.

• Because quasars emit strongly in the ultraviolet, they aredis-tinctly blue at optical wavelengths.

• Many of the first quasars discovered were also radio sources, butwe know now that most quasars are not strong radio emitters.

• They exhibit a non-thermal continuum spectrum that is strongerthan most other sources at all wavelengths, varies substantiallyin time, and exhibits the basic characteristics of synchrotron ra-diation (nonthermal and polarized).

• They usually exhibit emission lines that are very broad. Ifthebroadening of the spectral lines is attributed to random motion ofsources in the emitting regions, velocities in the range of 10,000km/s are indicated by the widths of the emission lines.

• Many quasars exhibit large redshifts, implying by the Hubblelaw (see Ch. 16) that they are at great distances and that the lightthat we see from them was emitted when the Universe was muchyounger than it is now.

• Quasar counts as a function of redshift indicate that they weremore abundant in the early Universe than in the later Universe(Fig. 14.1).

14.5. QUASARS 355

Redshift54 3 2 1.5 1 0.5

0

10

20

30

40

Density o

f Q

uasars

(to

day =

1)

Age of Universe (today = 1)

0 0.2 0.4 0.6 0.8 1

Figure 14.1:Observed quasar densities as a function of the fractional age of theUniverse. We see that quasars were much more abundant earlier in the historyof the Universe than they are today. The decrease of quasar abundances at thevery earliest times (redshifts larger than about 3) may represent partially the finiteamount of time for quasars to begin forming after the formation of the Universe andpartially observational bias, since it is more difficult to detect objects at the largestdistances and therefore the earliest times.

The huge energy output, large redshifts, and jets and otherstructure make it clear that quasars are not stars. Now infavorable cases we have images showing the host galaxyof the quasar. Thus, it is clear that quasars are enormouslyenergetic phenomena that occur in certain galaxies.


5"

3C 273

Jet

N

E

Quasar 3C 273 and jet Optical jet + radio contours

3C 273

Figure 14.2:The quasar 3C273, which has a redshift of 0.158, meaning thatthelight that we now see coming from it was emitted about 2 billion years ago andit is about 660 Mpc away, by the Hubble law. The sharp radial lines from thequasar are optical spike artifacts common in quasar images because of their star-like brightness.

Optical and radio images of the quasar 3C273 are illustratedin Fig. 14.2.

• The left image of 3C 273 in Fig. 14.2 shows the quasar and a jet.

• The right image, which has been rotated and enlarged relativeto the left image, superposes on this optical image contoursofradio frequency intensity.

• Despite its distance, 3C 273 has an apparent visual magnitude of+12.9, which implies that it must be incredibly luminous.

When summed over all wavelengths, one finds that an av-erage quasar like 3C 273 is typically some 1000 timesmore luminous than bright normal galaxies.

14.5. QUASARS 357

14.5.3 An Implied Enormous and Compact Energy Source

Quasars are extremely luminous at all wavelengths, whichimplies an enormous energy source.

• Many also exhibit variability in this luminosity ontimescales as little as months, weeks, or even hours.

• As discussed in Box 14.3 (next page), this variabilityimplies that their energy output originates in a verycompact source.


Box 14.3 The Speed of Light and Maximum Size of an Energy Source

If an energy source of a certain size is to exhibit a well-defined period in its lumi-nosity, some signal must travel through the object to tell itto vary. Since the signalcan travel no faster than light velocity, the maximum sizeD of an object varyingwith some characteristic timeP is the distance that light could have traveled duringthat time,D ∼ cP, as illustrated in the figure below.

P

D ~c P

Energy Source

Time

Inte

nsity

For example, if a source is observed to vary its light output substantially over aperiod of a week, the spatial extent of the energy-producingregion can be no largerthan a "light week", which is 1.82× 1011 km. The distances covered by light forvarious fixed times are summarized in the following table.

Table 14.3: Distances covered by light in a fixed timeTime km AU Parsecs

Year 9.46×1012 63,240 3.07×10−1

Month 7.88×1011 5270 2.58×10−2

Week 1.82×1011 1216 5.90×10−3

Day 2.59×1010 173 8.41×10−4

Hour 1.08×109 7.21 3.50×10−5

Minute 1.80×107 0.120 5.84×10−7

Second 3.00×105 0.002 9.73×10−9

Millisecond 3.00×102 0.000002 9.73×10−12

These arguments place only an upper limit on the sizes of variable sources. Signalscausing the variation may travel at less than light speed andthe size of the energy-producing region may be less than the upper limit imposed by these arguments. Butthis is a very powerful argument because it depends only on a general principle(finite speed of light) and not on the internal details of the source.

14.5. QUASARS 359

For example, if the source exhibits variability over a pe-riod of an hour, from Table 14.3 we see that the maximumsize of the emitting region is about 7 AU.

When it was first realized that quasars possess extremely powerfulenergy sources and that their variability argues that this energy is pro-duced in a region that can be no larger than the Solar System, it raisedserious issues about whether known physical processes thatcould ac-count for the energy characteristics of quasars.

• Since the discovery of quasars we have achieved a much deeperunderstanding of black holes and a likely mechanism to produceenergy on this scale in such a compact region has emerged.

• There is relatively uniform agreement that rotating blackholescontaining of order a billion solar masses (which would haveevent horizons smaller in size than the Solar System) are themost plausible candidates for quasar energy production.

However, before turning to a more detailed discussion of this idea, letus examine another class of observational phenomena that appears tohave much in common with quasars.


14.6 Active Galactic Nuclei

A collection of billions of ordinary stars is expected to produce aspectrum that is approximately blackbody, because the spectrum ofthe individual stars is blackbody.

• Such a spectrum is dominated by a continuum that often peaksat visible wavelengths.

• For example, the Milky Way emits radio waves, but its radioluminosity is about a million times smaller than its visiblelumi-nosity.

• In addition, the spectral lines observed for normal galaxies (asfor its stars) are mostly absorption lines, with few emission lines.

Thus, spectra for normal galaxies, as for the stars that theycontain, are

• typically a continuum peaking at visible-light wave-lengths representing thermal emission, with

• absorption lines superposed on the continuum andfew emission lines.

14.6. ACTIVE GALACTIC NUCLEI 361

However, some galaxies differ from this norm,

• They exhibit nonthermal emission from the RF to X-ray regionof the spectrum and/or jets and unusual structure associated withthe visual appearance of the galaxy.

• We refer to these asactive galaxies.

• Since the source of the activity is normally concentrated in thenucleus of the galaxy, they are also calledactive galactic nucleior AGN.

Generally, active galaxies exhibit some combination of thefollowing characteristics:

• Unusual appearance, particularly of the nucleus.

• Jets emanating from the nucleus.

• High luminosities relative to normal galaxies, butgenerally smaller than for quasars.

• Excess radiation at RF, IR, UV, and X-ray wave-lengths.

• Nonthermal continuum emission, often polarized,perhaps with broad and/or narrow emission lines.

• Rapid variability from a compact energy source inthe galactic nucleus.


Many galaxies with active nuclei are relatively nearby but

• They are more likely to be found at larger distances,implying that they were more common earlier in theUniverse’s history.

• Many of these characteristics are also exhibited byquasars, and we shall see that quasars may be closelyrelated to the nuclei of active galaxies.

• In fact, we shall conclude below that quasars es-sentially are a particularly luminous form of activegalactic nucleus.

There are several general classes of AGN that we nowsummarize.


14.6.1 Radio Galaxies

Radio galaxies are AGN associated with a large nonthermal emissionof radio waves.

Strong radio sources are often named by the constellationfollowed by a capital letter designating the order of dis-covery in the constellation. For example, the relativelynearby radio galaxy M87 (in the Virgo Cluster) is associ-ated with the powerful radio source Virgo A.

• Powerful radio galaxies are elliptical and often exhibit jet struc-ture from a compact nucleus.

• Weaker radio sources are found associated with smaller jets insome spiral galaxies.

• There are two broad classes of radio galaxies.

1. Core-halo radio galaxiesexhibit radio emission from a re-gion concentrated around the nucleus of the galaxy that iscomparable in size to the optically visible galaxy.

2. Lobed radio galaxiesdisplay great lobes of radio emissionextending in some cases for millions of light years beyondthe optical part of the galaxy.

• A radio galaxy can be a spectacular sight at RF wavelengths,but even powerful radio galaxies may appear to be rather normalelliptical galaxies at optical wavelengths.

• Often abnormalities at optical wavelengths become obvious onlyif the very core of the galaxy is resolved.


Jet

Disk

400 LY

Ground Radio + Optical Hubble Space Telescope

88,000 LY

RadioLobes

Figure 14.3:The radio galaxy NGC4261. On the left the false-color radio-wavemap is superposed on the ground-based optical image of the galaxy. The imageon the right is a high-resolution Hubble Space Telescope image of the core of thegalaxy, showing a dusty accretion disk thought to surround acentral black-holeengine that is driving the jets producing the radio lobes. The black hole engine isthought to lie inside the bright dot at the center of the disk.

The left side of Fig. 14.3 shows enormous and powerful radio emis-sion lobes superposed on the optical image of the ellipticalgalaxyNGC4261.

• The optical image of NGC4261 on the left side suggests a rathernormal looking elliptical galaxy.

• However, the high-resolution image of the core of the galaxyshown on the right indicates that something very unusual is hap-pening at the center of the galaxy.

• The suggestion is that there is a supermassive black hole atthecenter and that this is the energy source powering the radio jetsassociated with the galaxy.


Figure 14.4: The Seyfert 2 galaxy NGC7742, which lies about 22 Mpc awayin the constellation Pegasus. The nucleus is very compact and bright at opticalwavelengths, which is characteristic of Seyfert spirals.

14.6.2 Seyfert Galaxies

Seyfert galaxies are usually spirals with very bright, verycompactnuclei. An example of a Seyfert galaxy is displayed in Fig. 14.4.

• Seyfert galaxies are the most commonly observed active galaxies(about 1 percent of observed spirals are Seyferts).

• Seyfert galaxies exhibit a strong nonthermal continuum from IRthrough X-ray regions of the spectrum, with emission lines ofhighly-ionized atoms that are sometimes variable.

• Emission lines suggest a low-density gas source, while high ion-ization implies a hot source responsible for the ionization.

• Some Seyferts appear to have jets leading to continuum RF emis-sion, but these are modest on the scale of the jets and RF emis-sion observed for active elliptical galaxies.


The thousand or so Seyfert galaxies that are now known can be clas-sified into two subgroups:

• Seyfert 1 galaxies have both broad and narrow emission linesand are very luminous at UV and X-ray wavelengths.

1. Broad emission lines are associated with allowed transitionsin elements like hydrogen, helium, and iron.

2. Their width is caused by Doppler broadening that suggeststhe lines are produced in clouds with average velocities ashigh as∼10,000 km/s.

3. Narrow lines are associated with forbidden transitions,forexample in twice-ionized oxygen (O III); their widths implythat the source velocities are less than∼1000 km/s.

4. These differences suggest that broad and narrow spectrallines originate in different physical regions of a Seyfert 1.

• Seyfert 2 galaxies have relatively narrow emission lines suggest-ing source velocities less that∼1000 km/s.

1. They are weak at X-ray and UV wavelengths but very strongin the IR.

2. Generally, their continuum emission is weaker than for Seyfert1 galaxies.

3. In Seyfert 2s both the forbidden and allowed lines are nar-row, suggesting that both kinds of lines originate in the sameregion with relatively low source velocities.


Days

Rela

tive X

-Ray Inte

nsity

IRAS 13224-3809

(Seyfert 1)ROSAT

X-Ray

0 5 10 15 20 25 30

0

100

200

300

Figure 14.5:X-ray variability of the Seyfert 1 galaxy IRAS 13224-3809.

Seyfert galaxies can exhibit 50-percent changes in opticalbrightnessin a matter of weeks and even larger changes over periods of months.

• As illustrated in Fig. 14.5, the brightness variation at X-ray wave-lengths can be even greater.

• By the arguments given previously for brightness variation inquasars, the rapid fluctuation on a timescale of approximately aday exhibited in Fig. 14.5 implies a very compact energy sourcefor the X-rays.

We conclude that they energy source for the X-rays is nolarger than light days in diameter (comparable in size withthe Solar System, which is about half a light day in diam-eter).


14.6.3 BL Lac Objects

BL Lacertae objects(BL Lac objectsor justBL Lacs,forshort)

• Exhibit either no emission lines or only very weakones, but have a strong nonthermal continuumstretching from RF through X-ray frequencies.

• They are generally radio-loud, and exhibit strong po-larization of their emitted light (typically several per-cent, as compared with 1 percent or less for mostother AGN).

• BL Lacs are members of a more general class ofAGN calledblazars. (We shall be somewhat loosein our terminology and use the terms "blazars" and"BL Lacs" interchangeably.)

• Blazars are relatively rare among AGN. For example,there are about 100 times more quasars and 10 timesmore Seyfert galaxies known than blazars.

• However, they are extremely powerful, typically be-ing 10,000 times more luminous than the Milky Wayand therefore some 1000 times more luminous thana Seyfert galaxy.


By masking the bright core that is responsible for pro-ducing the intense and rather featureless continuum, it ispossible to acquire a spectrum of the “fuzz” seen faintlyaround many BL Lacs.

• This spectrum and the variation of the light intensityof the fuzz with distance from the core indicate thatthe fuzz is the outer part of a giant elliptical galaxyin most of the cases that have been analyzed.

• This suggests that most BL Lacs are the active coresof giant elliptical galaxies.

• From the redshifts of faint spectral lines it is foundthat blazars often correspond to more distant objectsthan Seyfert galaxies or radio galaxies, but they arecloser to us on average than quasars.


1980 1985 1990 1995

Years

0

5

10

RF

In

ten

sit

y (

3.8

cm

)1975

BL Lacertae

Figure 14.6:Variability of BL Lacertae.

• As illustrated in Fig. 14.6, blazars can exhibit dra-matic variability in their light output (there is a cor-responding strong variability in their polarization).

• Although Fig. 14.6 emphasizes the longer-term vari-ation of the light output from a BL Lac, closer exam-ination of the lightcurves indicates that they can varysignificantly on timescales as short as days.

• This implies a power source the size of the Solar Sys-tem or smaller.


108101010121014

Frequency (Hz)

Flu

x (

Rela

tive U

nits)

102

104

106

108

1010

1012

1

SeyfertNGC 1068

SeyfertNGC 4151

Quasar 3C 273

RFFarIR

NearIRV

is

Normal galaxy

Figure 14.7:Comparison of some representative spectra for normal galaxies, ac-tive galaxies, and quasars. The spectrum of the galaxy is approximately thermal butthe spectra of the active galaxies and quasars are highly nonthermal. Because elec-trons must radiate energy continuously as they spiral in magnetic fields, the energydriving the huge sustained synchrotron emission from thesenonthermal sourcesmust be replenished constantly. Strong, polarized, nonthermal emission is an indi-rect sign not only of strong magnetic fields, but also of a verylarge energy sourcefor the quasars and active galaxies.

14.6.4 The Unified Model of AGN and Quasars

Some representative spectra of normal galaxies, quasars, and activegalaxies are shown in Fig. 14.7.

• These spectra imply that active galaxies and quasars are some-thing very different from normal galaxies.

• On the other hand, the similarity of the spectra for quasarsandSeyfert galaxies implies that there might be a relationshipbe-tween quasars and active galaxies.


• That relationship is suggested by the huge nonthermal emissionfrom AGN and quasars, which implies a strong and very com-pact energy source of enormous strength.

• The only plausible candidate for the engine driving these phe-nomena is a rotating, supermassive black hole of mass millionsto billions of solar masses at their center.

The high efficiency for gravitational energy conversion inblack hole accretion provides the most convincing gen-eral argument that active galactic nuclei and quasars mustbe powered by accretion onto rotating supermassive (M ∼109M⊙) black holes. For example, in Exercise 15.3 ob-served luminosities and temporal luminosity variations areused to show that

– A quasar could be powered by accretion of as little asseveral solar masses per year onto an object of mass∼ 109M⊙.

– This mass must occupy a volume the size of the SolarSystem or smaller to be consistent with observations.

• In the past several decades a large amount of observationaldataand theoretical understanding supporting this point of view haveemerged.


The present belief is that (despite significant observationaldifferences)

• AGN such as Seyfert galaxies, blazars, and radiogalaxies are quite closely related, and that

• Quasars in turn are just a particularly energetic formof AGN.

• All are now thought to be active galaxies with brightnuclei powered by rotating supermassive black holes.

In the unified model that we now discuss, theobservational differences among quasars andvarious AGN mostly reduces to a matter of

1. How rapidly matter is accreting onto theblack hole(feeding the monster).

2. Whether the region of the central engineis masked from our view by dust(hidingthe monster).

3. How far away the AGN or quasar is fromus(proximity to the monster).


We are thus led to an hypothesis:Perhaps all large galaxies havesupermassive black holes at their centers.

• The question of whether the galaxy exhibits an active nucleus isthen primarily one of "feeding the monster" at the center.

• In a quasar, the black hole is accreting matter at a higher rate,leading to very high luminosity.

• In a normal galaxy like the Milky Way, the black hole is ratherquiet because it is presently accreting little matter.

• Seyfert and other active galaxies are somewhere in between: theblack hole is active because it is accreting matter, but at a ratelower than that for a quasar.

This picture is supported by the finding that

1. Seyfert galaxies are more abundant than average ininteracting pairs of galaxies, and that

2. Approximately a quarter of observed Seyferts exhibitevidence for tidal distortion.

Both of these observations suggest that interactions withother galaxies may “turn on” an AGN by increasing thefuel flow to the central black hole.


The AGN Central-Engine Model

We now discuss a unified AGN model where all activegalaxies and quasars are powered by central supermassiveblack holes.

1. In this unified model, the differences among AGNarise primarily from differences in the orientation an-gle and local environment for the central engine.

2. The first determines whether the view of the centralengine is blocked by dust; the second determines therate at which fuel flows to the central engine.


Black Hole Engine

DustyTorus

DustyTorus

Thin Hot Accretion Disk

Jet

Jet

Clouds inNarrow-LineRegion (NLR)

Clouds inBroad-LineRegion (BLR)

To Radio Lobe

To Radio Lobe

Figure 14.8:The standard AGN black-hole central engine paradigm.

Fig. 14.8 illustrates the standard black-hole central engineparadigm for AGN and quasars.

• The black hole itself occupies a tiny region in thecenter (its size is exaggerated in this figure).

• It is surrounded by a dense torus of matter revolvingaround the hole and a flattened accretion disk insideof that where the matter whirls even faster.

• On the polar axes of the rotating black hole there arejets of matter being ejected as part of the matter inthe accretion disk is sucked into the black hole andpart is flung out at high velocity in the jets.


Relativistic Jet

Counterjet

SpinningAccretion Disk

Rotating BlackHole Engine

Rotating Magnetic FieldAnchored to Spinning Accretion Disk

SynchrotronAcceleration of Charged Particles

SynchrotronRadiation

Ionizing Radiationfrom Hot Disk (UV)

Figure 14.9:The central engine of an active galactic nucleus.

Figure 14.9 illustrates schematically what we believe thecentral engine of an AGN or quasar would look like if wecould turn the radiation off and clear away the gas anddust.

• The central engine consists of a rotating, supermas-sive black hole surrounded by a thin accretion disk.

• The accretion disk is very hot because of collisionsin the rapidly swirling gas that it contains.

• Because it is hot, it radiates strongly in the UV.


Radiation from the accretion disk is expected to be domi-nated by thermal processes.

• Typical estimates of the temperature for the accretiondisk of a 109M⊙ black hole are about 10,000 K;

• by the Wien law this corresponds to a peak wave-length of 3000 Angstroms, which lies in the UV re-gion of the spectrum.

• The accretion disk temperature depends inversely onthe mass of the black hole, so smaller ones have hot-ter accretion disks.

Thus stellar-size black holes have accretiondisks that are even hotter than the ones con-sidered here, and they radiate more in the X-ray region of the spectrum.


Black Hole Engine

DustyTorus

DustyTorus

Thin Hot Accretion Disk

Jet

Jet

Clouds inNarrow-LineRegion (NLR)

Clouds inBroad-LineRegion (BLR)

To Radio Lobe

To Radio Lobe

Photons from the accretion disk are responsible for much of the con-tinuum observed from AGN, either directly, or by heating surroundingmatter which then re-radiates at longer wavelengths.

1. Photons emitted by the accretion disk also ionize atoms inthenearby clouds of gas where velocities are very high and producethe broad line emission spectrum of the AGN(the Broad-LineRegion or BLR).

2. Finally, they ionize clouds further away from the centralenginewhere velocities are lower and this produces the narrow lines ofthe emission spectrum(the Narrow-Line Region or NLR).

Whether both broad and narrow emission lines are visible to adistantobserver will depend on the location of the observer relative to theplane of the accretion disk and the torus that may surround it.


DustyTorus

DustyTorus

Hole Disk

NLR

NLR

IonizationCone

IonizationCone

Ionizing (UV)

Radiation

Figure 14.10:Ionization cones of an AGN. The regions shown in violet can seedirectly the UV radiation radiated by the accretion disk. Other regions cannot seethe accretion disk directly because it is blocked by the dusty torus. Clouds in thebroad-line region are labeled BLR and clouds in the narrow-line region are labeledNLR.

If this picture of the central engine is correct, there should be evidencein AGN for anisotropic ionization.

• That is, we should see ionization in the central region of thegalaxy concentrated in particular directions from the center.

• In the simplest case, there should be cone-shaped regions of ion-ization corresponding to the directions from the hot accretiondisk that are not blocked off by the dust torus.

• This ionization cone is displayed schematically in Fig. 14.10.

• All of the region shaded in violet can "see" the hot central accre-tion disk and the ionizing radiation that it is emitting.

Various observations suggest the presence of such anisotropic ioniza-tion zones near AGN central engines.


A Unified AGN and Quasar Model

Let us now pull together the various threads of this dis-cussion and summarize the basic unified model and ac-tive galactic nuclei and quasars. We hypothesize that ac-tive galaxies are powered by rotating, supermassive blackholes in their center having all or most of the followingfeatures:

• Accretion disks.

• Possibly a dusty torus surrounding the accretion disk.

• Clouds near the black hole that move at high velocityand produce broad emission lines.

• Lower-density and slower clouds further from theblack hole that produce narrower emission lines.

• Possible relativistic jet outflow perpendicular to theplane of the accretion disk.


Observer

Observer

Seyfert 2Galaxy

Lobe RadioGalaxy

Seyfert 1Galaxy

BL LacObject

In Spiral Galaxies In Elliptical Galaxies

Figure 14.11:Unified model of active galactic nuclei. In this picture all are pow-ered by rotating supermassive black holes and the observational differences arelargely determined by viewing angles relative to the plane of the black hole accre-tion disk and the jets. Quasars are then hypothesized to be very similar to AGN,but more powerful because the black hole is accreting at a higher rate.

The cartoon in Fig. 14.11 illustrates this unified AGNmodel.

• If the orientation of the system relative to the ob-server is as displayed in the top row of Fig. 14.11,

1. We see a Seyfert 2 galaxy if the active galacticnucleus is in a spiral galaxy.

2. We see a lobed radio source if the AGN is in anelliptical galaxy with strong jets.


Observer

Observer

Seyfert 2Galaxy

Lobe RadioGalaxy

Seyfert 1Galaxy

BL LacObject

In Spiral Galaxies In Elliptical Galaxies

• On the other hand, if the orientation of the activegalactic nucleus central engine is as in the bottomrow of the above figure,

1. We see a Seyfert 1 galaxy if the host is a spiralgalaxy.

2. We see a blazar (BL Lac) if the host is ellipticaland the jet is strong.

3. We see perhaps a core-halo radio galaxy if thejets are weaker in an elliptical system.


We then conjecture that quasars are just particularly ener-getic forms of these active galaxies and are described bythe same unified model.

1. The primary difference is that quasars are very lumi-nous because their black holes are being fed matterat a higher rate than for moderately luminous AGN.

2. We surmise that quasars are more abundant at largerredshift because large redshift corresponds to ear-lier in the Universe’s history, when matter was moredense and collisions between galaxies more frequent.

Therefore, quasars may be just "better fed" active galacticnuclei from a time when more fuel was available to powerthe black hole engines.

As a corollary, we may conjecture that many nearby nor-mal galaxies once sported brilliant quasars in their cores,but have since used up the available fuel.

• Their massive black holes lie dormant, ready to blazeback to life should tidal interactions with anothergalaxy divert matter into the black hole.

• This may be true of our own galaxy, which appears tohave a∼3 million solar mass black hole at its center,but presently has weak nonthermal emission.


The Center of Centaurus A

NOAO

Hubble

Figure 14.12:OPtical images of the center of Centaurus A.

14.6.5 Feeding a Nearby Monster

Fig. 14.12, displays the AGN nearest to us: the giant elliptical galaxyNGC 5128, which harbors the strong radio source Centaurus A.

• Centaurus A is only about 107 light years away. Its radio lobesspan 10◦ in our sky (20 times the Moon’s diameter)

• It has a faint optical jet and strong radio jet, and violent activityat the center causing the jets that produce the huge radio lobes.

• IR observations suggest that behind the dark dust lanes lies a gasdisk 130 light years in diameter, surrounding a black hole ofabillion solar masses that is powering the activity in the core ofthe galaxy.


The Center of Centaurus A

NOAO

Hubble

The firestorm of starbirth seen in the above figure has been causedby a collision between a smaller spiral galaxy and a giant ellipticalgalaxy within the last billion years.

• The dark diagonal dust lanes are thought to be the remains ofthespiral galaxy, seen almost edge-on and encircling the core of thegiant elliptical.

• The logical conclusion is that the supermassive black holeat thecenter of Centaurus A has been triggered into pronounced activ-ity by the collision of the parent galaxy with another one.

• This collision has led to strong radio emission because of rela-tivistic jets powered by enhanced accretion onto the black hole.

• It has also produced enhanced star formation across broad re-gions because of the compression of gas and dust in the collisionof the two galaxies.


Figure 14.13:X-ray jet from Centaurus A. In this false-color Chandra X-ray Ob-servatory map the highest X-ray intensity is indicated by white and red colors.

This black hole central-engine interpretation is supportedfurther by Fig. 14.13.

1. This image shows an X-ray map obtained by theChandra X-ray Observatory superposed on an opti-cal image of Centaurus A.

2. The Chandra data reveal a bright central region(white ball near the center) that likely surrounds theblack hole.

3. It also indicates a strong jet oriented to the upper left25,000 light years in length that is probably beingejected on the polar axis of the central black hole.

4. There also is a fainter jet oriented in the opposite di-rection, making it harder to see.


Blazars sometimes emit extremely high-energy photons.

• For example, 1012 eV gamma-rays have been de-tected coming from Markarian 421, a BL Lac at aredshift ofz= 0.031.

• Such gamma-rays cannot be produced by AGN ther-mal processes, but the unified model of AGN andquasars provides a possible explanation.

• Box 14.4 (next page) illustrates a mechanism calledinverse Compton scatteringby which lower-energyphotons can be boosted to extremely high energy inan AGN.

As illustrated there, 1012 eV gamma-rays are possible ifX-rays produced by irradiation from the accretion disk en-ter a highly-relativistic jet and are inverse Compton scat-tered by the jet.


Box 14.4 Producing Very High Energy Photons

Some AGN emit extremely high-energy photons that cannot be produced by anythermal processes expected in an AGN. Instead, they are thought to involve a pro-cess calledinverse Compton scattering,illustrated below.

ScatteredElectron

Electron

RelativisticElectron

ScatteredElectron

Higher EnergyPhoton

Higher EnergyPhoton

Lower Energy Photon

Lower EnergyPhoton

Compton Scattering

Inverse Compton Scattering

In Compton scatteringa high-energy photon strikes an electron and accelerates it,giving up energy. Inverse Compton scattering is the reverseprocess: a high-energyelectron strikes a photon, imparting energy. A specific way in which an AGN couldproduce a high-energy photon is illustrated in the following figure.

Accretion Disk

Black Hole

RelativisticElectron

Jet

InverseComptonScattering

High-EnergyPhoton

OpticalPhotons

Ionizing Photon

BLR Cloud

UV from the disk can photoionize clouds in the broad line region (BLR) and theseclouds then emit light largely at optical wavelengths. If these photons enter thejet, they can be inverse Compton scattered to much higher energies. The frequencyboost factor is given byν/ν0 = 1/(1− v2/c2), wherev is the jet velocity, so in ahighly relativistic jet, boost factors of 106 can be obtained (Exercise 14.2). Thiscan convert optical photons into intermediate-energy gamma-rays and X-rays into1012 eV gamma-rays.


14.7 Gamma Ray Bursts

Discovered serendipitously by military satellites, gamma-ray bursts (GRB) were initially one of the great mysteriesin modern astronomy.

• On average, about one burst a day is observablesomewhere in the sky.

• However, until the 1990s we had little idea aboutwhere they originated or even how far away theywere.

New observations have begun to clear away the mysteryand we now have a much better understanding of theseremarkably energetic events.

14.7. GAMMA RAY BURSTS 391

+90o

-90o

+180o-180o

Figure 14.14:The sky at gamma-ray wavelengths in galactic coordinates. Whiteand yellow denote the most intense and blue the least intensesources. The diffusehorizontal feature is from gamma-ray sources in the plane ofthe galaxy. Brightspots to the right of center in the galactic plane are galactic pulsars. Brighter spotsabove and below the plane of the galaxy are distant quasars. Some of the faintersources in this map are of unknown origin.

14.7.1 The Gamma-Ray Sky

Our sky glows in gamma-rays, in addition to the other more familiarwavelengths.

• Since gamma-rays are high-energy photons, they are not easy toproduce and tend to indicate unusual phenomena; thus they areof considerable astrophysical interest.

• Because gamma-rays are absorbed strongly by the atmosphere,a systematic study of gamma-ray sources in the sky requires anorbiting observatory.

In Fig. 14.14 we show the continuous glow of the gamma-ray sky, asmeasured by the Compton Gamma-Ray Observatory from orbit.


+90

-90

-180+180

Figure 14.15:Location on the sky of more than 2000 gamma-ray bursts plottedin galactic coordinates. The data were recorded by the Burstand Transient SourceExperiment (BATSE) that was carried aboard the orbiting Compton Gamma-RayObservatory. The highly isotropic distribution on the sky of GRB events arguesstrongly that they occur at cosmological distances.

Superposed on the steady flux are sudden bursts of gamma-rays.

• These events can be as short as tens of milliseconds and as longas several minutes.

• Figure 14.15 shows the position of more than 2000 gamma-raybursts observed by the Burst and Transient Source Experiment(BATSE) of the Compton Gamma-Ray Observatory.

• This plot indicates that the distribution of bursts isisotropic onthe sky.


+90

-90

-180+180

• If bursts had a more local origin, such as in the disk of the galaxy,they should be concentrated along the galactic equator in thisfigure, not randomly scattered over the whole diagram.

• This tells us that either the gamma-ray bursts come from eventsat great distances (cosmological distances), or perhaps they comefrom events in the more spherical halo of our galaxy.

• Below we shall see that Doppler shifts for spectral lines confirmthatgamma-ray bursts are cosmological in origin.

• That they are at cosmological distances makes GRB quite re-markable, since to be seen at such large distances they must cor-respond to events in which energy comparable to that of a super-nova is liberated in a short period in the form of gamma-rays.

• Furthermore, the mechanism producing the gamma-rays mustallow them to escape with little interaction with surroundingmatter. Even a small interaction with baryons would downscat-ter gamma-rays to light of longer wavelength.


14.7.2 Characteristics of Gamma-Ray Bursts

We now know that gamma-ray bursts have the following characteris-tics:

1. The isotropic distribution on the sky found by BATSE suggests acosmological origin. This has been confirmed by direct redshiftmeasurements on emission lines in GRB afterglows.

2. Known redshifts for gamma-ray bursts range fromz= 0.0085toz= 4.5, with an averagez∼ 1 (as of 2007).

3. The spectrum is non-thermal, typically peaking around 200 keVand extending perhaps as high as GeV.

4. The duration of individual bursts spans 5 orders of magnitude:from about 0.01 seconds up to several hundred seconds.

5. There is a variety of time structure, from rather smooth tomil-lisecond fluctuations (implying a very compact source, by causal-ity).

6. There appear to be two classes of bursts: long-period and short-period bursts, having many common features but with sufficientdifferences to suggest that the bursts in the two classes aretrig-gered by different events.


To determine the source of events with these character-istics, it was necessary first to understand where theyoriginated. For example, were they associated with anypreviously-known galaxies?

• Initial progress was slow because BATSE observa-tions could localize the position of a burst only withinseveral degrees on the sky, and enormous numbersof stars and galaxies and other potential sites for agamma-ray burst lie within a patch of sky that large.

• Therefore, it was very difficult to know exactly whereto point telescopes to find evidence associated withthe gamma-ray burst at other wavelengths before thatevidence faded from sight.


Figure 14.16:Localization of the X-ray afterglow for a long-period GRB bythesatellite BeppoSAX.

14.7.3 Localization of Gamma-Ray Bursts

A major breakthrough came in 1998 when it became possible to cor-relate some GRB with other sources in the visible, RF, IR, UV,andX-ray portions of the spectrum.

• First enabled by a small Dutch–Italian satellite called BeppoSAXthat could pinpoint the position of X-rays following GRB with 2arc-minute resolution in a matter of hours.

• This permitted other satellite and ground-based instruments tolook quickly at the burst site.

Figure 14.16 shows an X-ray transient observed by BeppoSAX at thelocation of a previously-seen gamma-ray burst.


BAT

XRT

Figure 14.17:Optical association of short-period GRB 050509B with a large el-liptical galaxy at a redshift ofz= 0.225. The larger (red) circle is the error circle forthe Burst Alert Telescope (BAT) on SWIFT. The smaller (blue)circle is the errorcircle for the X-Ray Telescope (XRT), which was slewed to point at the event whenalerted by BAT. The XRT error circle is shown enlarged in inset, suggesting thatthe GRB occurred on the outskirts of the large elliptical galaxy partially overlappedby the XRT error circle on the right.

• These transients are thought to be associated with a rapidly fad-ing fireball that is produced by the primary gamma-ray burst.

• One of the first localizations for a short-period burst by the SWIFTsatellite is illustrated in Fig. 14.17.

Correlation of GRB with sources at other wavelengthshave allowed distances to be estimated to GRB becausespectral lines and their associated redshifts have been ob-served in the transients after the burst.These observationsshow conclusively that GRB are occurring at cosmologi-cal distancesand not in the halo of the local galaxy.


Duration

Hard

ness

Short-Period

Long-Period

Figure 14.18:Hardness (propensity to contain higher-energy photons) ofthe spec-trum versus duration of the burst, illustrating the separation of the GRB populationinto long, soft bursts and short, hard bursts [?]. The parameterHR measures thehardness of the spectrum andT90 is defined to be the time from burst trigger for90% of the energy to be collected. Bursts withT90 shorter than 2 seconds are clas-sified as short-period and those withT90 greater than 2 seconds are classified aslong-period bursts. Although GRB form a continuum of events, we see from thisplot that shorter bursts generally have a harder spectrum.

14.7.4 Two Classes of Gamma-Ray Bursts

Figure 14.18 suggests two classes of gamma-ray bursts:

1. Short-period bursts,which last less than two seconds and in gen-eral exhibit harder (higher energy) spectra.

2. Long-period bursts,which have softer (lower energy) spectraand can last from several seconds up to several hundred seconds.

These and other distinctions in the data set suggest that GRBrepresenta family of events having more than one source.


One unifying idea is that gamma-ray bursts are poweredby a collapse of large amounts of spinning mass to a blackhole, but that there are several mechanisms to cause this.The preferred mechanisms based on current data are

1. Particular classes of core-collapse supernovae in-volving massive stars with high angular momentumfor the long-period bursts.

2. The merger of two neutron stars or a neutron star anda black hole for the short-period bursts.

In both cases the outcome is a Kerr black hole havinglarge angular momentum and strong magnetic fields, sur-rounded by an accretion disk of matter that has not yetfallen into the black hole.

• This scenario likely leads to highly-focused relativis-tic jet outflow on the polar axes of the Kerr blackhole.

• These jets are powered by

– rapid accretion from the disk,

– neutrino–antineutrino annihilation,

– strong coupling to the magnetic field.

Thus, the GRB black hole engine may have many similar-ities with the engine powering AGN and quasars, but on astellar rather than galactic-core scale.


14.7.5 Association of GRB with Galaxies

The localization provided by afterglows has permitted anumber of long-period and (more recently) short-periodGRB to be associated with distant galaxies.

1. Long-period (soft) bursts appear to be strongly cor-related withstar-forming regions(strong correlationwith blue light in host galaxies).

2. Short-period (hard) bursts are generally fainter andsampled at smaller redshift than long-period bursts.They do not appear to be correlated with star-formingregions.

3. There is some evidence that long-period bursts arepreferentially found in star-forming regions havinglow metallicity.

These observations provide further evidence that long-period and short-period bursts are initiated by differentmechanisms.


HD56925 (Wolf-Rayet)

Figure 14.19:A Wolf–Rayet star surrounded by shells of gas that it has emitted.These massive, rapidly spinning main-sequence stars are thought to be the progen-itors of Type Ib and Type Ic core-collapse supernovae, and oflong-period GRB.

14.7.6 Association of Long-Period GRB with Supernovae

There is an intimate relationship between long-period GRB and par-ticular types of core-collapse supernovae called Types Ib and Ic.

• The supernova mechanism in both cases is thought to involvecore collapse in a rapidly-rotating, massive(15–30M⊙) main-sequence star called aWolf–Rayet star.

• These stars exhibit large mass loss and can shed their hydrogenand even helium envelopes before their cores collapse.

• They are so massive that they can collapse directly to a rotating(Kerr) black hole, instead of a neutron star.

Figure 14.19 shows a Wolf–Rayet star surrounded by large shells ofgas that it has ejected.


It is thought that in a Type Ib supernova the H shell hasbeen removed before collapse of the core, and it is thoughtthat in a Type Ic supernova both the H and He shells havebeen removed before the stellar core collapses.

On the other hand, there is little observational evidencethat short-period bursts are associated with star-formingregions, or supernovae. The favored mechanism for short-period bursts involves the formation of an accreting Kerrblack hole by the merger of two neutron stars (or a neutronstar and a black hole).


GRB Central Engine

Interstellar Medium (ISM)

Relativistic Jet

Internal shocks: collisions of shells of ejecta moving at different speeds emit the gamma-rays of the GRB.

External shocks: collisions of a shock with the ISM causes emission in gamma-rays, X-rays, optical, and radio, producing the GRB afterglow.

Fe-lines may appear from illumination of a pre-ejected shell or the envelope of a massive star.

ExternalShock

InternalShocks

Figure 14.20:Fireball model for transient afterglows following a GRB. Internalshocks in the relativistic jet produce the gamma-rays and the external shocks result-ing from the jet impacting the interstellar medium (ISM) produce the afterglows.

14.7.7 The Fireball Model and Afterglows

There is broad agreement that the transients observed at various wave-lengths following gamma-ray bursts can be accounted for by arela-tivistic fireball model, as illustrated in Fig. 14.20.

• In this model some central engine deposits a very large amountof energy in a small volume of space.

• That produces a fireball expanding at relativistic velocities andthis fireball is responsible for the observed transients.


14.7.8 Mechanisms for the Central Engine

The central engines responsible for gamma-ray bursts andthe associated afterglows are not well understood, but anacceptable model for them must embody at least the fol-lowing features:

1. All models require highly-relativistic jets to accountfor observed properties of gamma-ray bursts.

• Lorentzγ factors of at least 200, perhaps as largeas 1000, appear to be required by observations.

• Jets focused with opening angles∼ 0.1 rad andpower as large as∼ 1052 erg

• Long-period bursts must (at least sometimes) de-liver ∼ 1052 erg to a much larger angular range(∼ 1 rad) to produce the accompanying super-nova, and the central engine must be capable ofoperating for 10 seconds or longer in these long-period bursts to account for their duration.

2. The large and potentially long power timescale, par-ticularly for long-period bursts, implies accretiononto a compact object. Thus, acceptable modelsmust produce substantial accretion disks.

Almost the only way that we know to explain such a rapidrelease of that amount of energy is from a collapse involv-ing a compact gravitational source.


As we have indicated, two general classes of models arenow thought to account for GRB.

1. Short-Period Bursts: The merger of two neutronstars, or a neutron star and a black hole, with jet out-flow perpendicular to the merger plane producing aburst of gamma-rays as the two objects collapse to aKerr black hole.

2. Long-Period Bursts:A hypernova,where a spinningmassive star collapses to a Kerr black hole and jetoutflow from the region surrounding this collapsedobject produces a burst of gamma-rays.

The unifying theme is the collapse of stellar-size amountsof spinning mass to a black hole central engine that powersthe burst.


14.7.9 The Collapsar Model and Long-Period Bursts

An overview of the collapsar model is shown in Fig. 14.21 (nextpage).


Gamma-ray burst

produced by jets

outside the star

Exploding

supernova

γ

γ

Jet breakout typically

at of order 10 seconds

after launch

Inner coreOuter core

Rest of star

Massive Wolf-Rayet star with large angular

momentum that has lost its hydrogen and

possibly helium envelopes.

The inner core collapses directly to a (Kerr) black

hole and the outer core forms an accretion disk

because of the angular momentum. Highly-

collimated, relativistic jets form on the polar axis,

powered by neutrino-antineutrino annihilation,

magnetic energy from the accretion disk, and

rotational energy extracted magnetically from the

black hole.

Kerr black hole

Accretion disk

The jets produce the gamma-ray burst

outside the star, while shock waves from the

core collapse and the jets blow the star

apart, leading to a Type Ib or Type Ic

supernova.

Jet

Jet

Shock

waves

Expanding

afterglow

At much larger distances, the interaction of the

jets with the surrounding medium begins to

produce the afterglow that will be detected at

longer wavelengths.

Figure 14.21:The collapsar model for long-period gamma-ray bursts and accom-panying Type Ib or Ic supernova.


Figure 14.22:Simulations of relativistic jets breaking out of the surfaces of Wolf–Rayet stars. Breakout of the jet is 8 seconds after launch from the center of a 15M⊙ Wolf–Rayet star. The Lorentzγ for the jet is about 200.

A simulation of relativistic jets breaking out of a Wolf–Rayet star is shown in Fig. 14.22.


Black hole

Rotation

axis

Accretion

disk

Figure 14.23:A rapidly-rotating WR star of 14 solar masses, about 20 secondsafter core-collapse (the polar axis is vertical). The density scale is logarithmic andthe 4.4M⊙ Kerr black hole has been accreting at about 0.1M⊙ per second for 15seconds at this point in the simulation.

In Fig. 14.23 a simulation of a Wolf–Rayet star 20 secondsafter core collapse is shown.


Figure 14.24:Nucleon wind off the accretion disk in a collapsar model.

In Fig. 14.24 a simulation showing a strong nucleon windblowing off the collapsar accretion disk is shown. Thiswind produces the supernova and the synthesis of56Nithat powers the lightcurve of the supernova by radioactivedecay.

The GRB and the supernova are powered in different waysin the collapsar model:

1. The GRB is powered by a relativistic jet deriving itsenergy from neutrino–antineutrino annihilation or ro-tating magnetic fields.

2. The accompanying supernova is powered by the diskwind illustrated in this figure.


100 3000 10000 30000 60000 100000

Temperature (millions of K)

0.875 ms 1.375 ms 2.125 ms

2.875 ms

3.625 ms

4.025 ms

Approximate

Schwarzschild

Radius

Simulation of the merger of two neutron stars. The elapsed time is about 3 ms and theapproximate Schwarzschild radius for the combined system is indicated. The rapid motionof several solar masses of material with large quadrupole distortion and sufficient density tobe compressed near the Schwarzschild radius indicates thatthis merger should be a strongsource of gravitational waves (Source: S. Rosswog simulation).

14.7.10 Merging Neutron Stars and Short-Period Bursts

The most widely accepted model for production of short-period gamma-ray bursts involves binary neutron starsmerging to form a Kerr black hole. A simulation of a neu-tron star merger is shown in the above figure.

http://eagle.phys.utk.edu/guidry/astro490/book/utilities/ rosswogMerger/rosswog1.mov


It is thought that it is not possible for the relativistic jets topunch out of the star quickly enough to account for GRBof a second or less duration in collapsar models.

• In the merger of two neutron stars a “centrifugal fun-nel” tends to form above and below the disk formedby the initial merger.

• These funnels have relatively low baryon density andthus relativistic jets emitted on the polar axes canpunch quickly into the interstellar medium, possiblyaccounting for very fast bursts.


Figure 14.25:Amplification of magnetic fields in merging neutron stars. Shearproduced at the merger boundary can substantially amplifying the already signifi-cant magnetic fields that are present.

An illustration of how the original magnetic fields of theneutron stars can be magnified in the merger is illustratedin Fig. 14.25.


140

120

100

80

60

40

20

0

Ð150 Ð100 Ð50 0x (km)

50 100 150

-6

-4

-2

0

2

4

Merger disk

Relativistic

Jet

Rotation axis

Neutrino-antineutrino

annihilation

Figure 14.26:Illustration of relativistic jet powered by neutrino–antineutrino an-nihilation in a neutron-star merger.

• One possible mechanism for powering the jets in GRB producedby neutron star mergers is neutrino–antineutrino annihilation aboveand below the plane of the merger disk, as illustrated in Fig.14.26.

• Another is accretion onto the rotating black hole from the sur-rounding disk in the merger.

• Another is to tap the power of the very strong magnetic fieldsthat are expected, for example as illustrated in Fig. 14.27 on thefollowing page.


ErgosphereHorizon

Jet

Jet

Figure 14.27:Illustration of relativistic jet powered by frame draggingof mag-netic fields by a Kerr black hole.

In the model illustrated in Fig. 14.27, the frame-draggingeffects associated with the Kerr black hole wind the fluxlines associated with the magnetic field around the blackhole and spiral them off the poles of the black hole rotationaxis, thus powering relativistic jets emitted on the polaraxes.

Seehttp://eagle.phys.utk.edu/guidry/astro490/book/utilities/semenovBH/for simulations.

chapter 14 black holes as central...

Documents