What Variability of (nonblazar) AGNs is Telling Us
Martin Gaskell
Dept. Physics & Astronomy
Univ. Nebraska
[email protected]
Huatulco, April 18, 2007
Another area Deborah has worked in – variability of nonblazar AGNs!
1.
2.
3.
4.
5.
OUTLINE
Review of simple accretion disk theory and what it
means for variability.
How similar is the variability of different types of
AGNs?
The nature of variability
Explaining wavelength-dependent variability
The X-ray / optical connection
1. DISKS AND VARIABILITY
Basic theory of quasar energy
generation
Schwarzschild radius
Rg = 2GM/c2.
⇒ τvar ∝ Mbh
~ 3 km per solar mass.
For NGC 5548 MBH ~ 108 solar masses
(Koratkar & Gaskell 1991 etc.)
⇒Rg ~ 3×108 km
~ 2 a.u. radius
~ ½ light hour diameter
Black-Hole Accretion
• Accretion onto a supermassive black hole
(Zel’dovich & Novikov 1964, Salpeter 1964).
• Gravitational PE = -GM/R
• For bound system in equilibrium, virial theorem
⇒
KE = -½ PE
∴½ PE lost
• Energy lost
– mechanically (mass outflow)
– radiatively
Matter settles into an accretion disk
(Lynden-Bell 1969)
Structure of an accretion disk
Pringle & Rees (1972), Shakura & Sunyaev (1973)
Luminosity structure
L (ergs s-1) per cm3 at any point depends on
• PE (= GM/R) released per gram, and
• the number of grams entering per sec per cm3.
which in turn depends on
• total accretion rate (dM/dt)
• local density (ρ).
Conservation of mass flux
Area of a shell is πR2, so volume of a shell ∝ R2
⇒ ρ ∝ 1/R2. [note: no assumptions about thickness]
Putting it all together
L ∝ (GM/R) (dM/dt)(1/R2) or
L ∝ R-3.
Temperature Structure
Assume black-body spectrum
L ∝ T4.
so T ∝ (L)1/4 ∝ (R-3)1/4 or
T ∝ R-3/4
Integrated spectrum
Integrate flux from rings (add up black-body
curves)
If T ∝ R-p
Fν ∝ ν(3-2/p)
If p = 3/4,
Fν ∝ ν+1/3
Shakura & Sunyaev (1973)
Pringle & Rees (1972)
Don’t get obsessed with Fν ∝ ν+1/3 !
• Standard thin disk:
T ∝ R-0.75 ⇒ Fν ∝ ν+0.33
• Non-local processes – additional energy
transport (e.g., irradiation from inner regions)
flattens T(R)
• Observed spectrum is very sensitive to T(R).
• E.g., “slim disk” (Abramowicz et al. 1988):
T ∝ R-0.5 ⇒ Fν ∝ ν-1
Average spectral energy distributions of real AGNs:
Elvis et al. (1994)
Observed distribution of UV-optical spectral indices:
Gaskell, Goosmann, Antonucci & Whysong (2004)
Reddening
(See also talks by Luc Binette and
Sinhue Haro-Corzo)
Bechtold et al. (1997)
Filled - optically selected
R
ng
i
n
e
d
ed
Open – radio-loud from
Netzer et al. (1995)
Deduced temperature structure
• α = - 0.5 ⇒ p = 0.57
• Reasonable (cf. 0.75 for standard thin disk
and 0.50 for “slim” disk)
Relative Sizes of Regions
Real AGN Variability
• All AGNs vary (gives a technique for finding them – see
Vicki Sarejedini’s talk)
• So AGN variability is normal.
Lyuti (2006)
First two things we get from variability:
1. Amplitude
2. Timescale
Amplitude
Tells if variability is important or unimportant.
E.g.:
• Variable stars – amplitude small (bolometrically) ⇒
variability unimportant (main energy mechanism not
varying)
• Supernovae – amplitude huge ⇒ variability fundamental
(main energy generation mechanism is varying)
• Quasars?
– amplitude large (see later)
assert: main energy generation mechanism is varying ⇒
variability fundamental.
2. HOW SIMILAR IS THE
VARIABILITY OF DIFFERENT
CLASSES OF AGNs?
Questions:
1. Do bright radio-quiet AGNs vary as much
in the optical as radio-loud AGNs of
comparable brightness?
2. Do high-accretion-rate AGNs vary as
much in the optical as low-accretion-rate
AGNs of comparable luminosity?
Conventional wisdoms:
• Radio-loud quasars show higher amplitude
optical variability than radio-quiet ones because
there is contribution of a jet-related, non-thermal
component in the optical (a “blazar” component).
• Some indications that NLS1s (high-accretionrate) AGNs vary less in the optical.
(but NLS1s known to vary more in soft X-rays).
Compare the three brightest
nearby quasars
• 3C 273 (z = 0.158, MV= -26.6) brightest and nearest
high-luminosity radio-loud quasar – well-known
variable; light-curve well studied for decades.
Two recently-discovered, comparable-luminosity,
nearby, radio-quiet quasars:
• PDS 456 (z = 0.184, MV = -26.9) Broad lines like 3C
273 (Torres et al. 1997; Simpson et al. 1999)
• PHL 1811 (z = 0.192, MV = -25.9) Narrow-Line
Seyfert 1 quasar (Leighly et al. 2001)
PHL 1811 V-band light curve
(2003)
14.80
14.90
V Magnitude
15.00
15.10
15.20
15.30
PHL 1811
15.40
15.50
0
20
40
60
80
100
2003 (relative day number)
120
PDS 456 V-band light curve
(2000)
14.00
V Magnitude
14.05
PDS 456
14.10
14.15
14.20
14.25
14.30
-10
0
10
20
30
40
Day Number
50
60
70
80
Comparison of Variability –
Seasonal RMS Variability
• PDS 456 (quiet; broad lines)
±0.042m (2000)
±0.036m (2003)
• PHL 1811 (quiet; narrow lines)
±0.104m (2003)
• 3C 273 (radio-loud)
±0.042m mean for 11 seasons with comparable
coverage.
rms seasonal variability has exceeded ±0.10m
only once over the past 30 years.
Comparing
high- and lowaccretion rate
quasars.
Klimek, Gaskell, &
Hedrick (2004), ApJ,
609
No clear differences
once you allow for
selection effects (we
observe NGC 5548
because it varies!)
SHORT TIMESCALES
1. Radio-louds and radio-quiets show
similar occurrences of optical sub-diurnal
variability (de Diego, Dultzin-Hacyan, Ramirez, & Benitez
1998; Wiita, Stalin, Gopal-Krishna, & Sagar 2004)
2. High-accretion-rate AGNs show a similar
occurrence of optical sub-diurnal
variability (Klimek, Gaskell, & Hedrick, 2004)
“Conventional wisdoms” seem to be wrong
on both long and short timescales.
⇒ OPTICAL VARIABILITY MECHANISMS
FUNDAMENTALLY THE SAME FOR
DIFFERENT CLASSES OF AGNs
3. The Nature of the Variability
ARE
THERE
FLARES?
Gaskell (2004)
(Soft X-rays)
Variability depends on flux level.
(Lyutyi & Oknyanskij,1987; Gaskell 2004)
Gaskell (2004)
Variability has a log-normal distribution
Gaskell (2004)
(a) Explains
light curves
without
flares
(b) No “high”
and “low”
states
Gaskell (2004)
(One of
these is
IRAS 132243809!)
4. Wavelength-Dependent
Variability
(Gaskell, 2007, astro-ph/0612474)
Wavelength-dependent delays
• Expected delays on sound-crossing or
dynamical (orbital) timescales (long – see
table above)
FIRST SURPRISE:
• Not seen at first (e.g., NGC 5548, Korista
et al. 1995; NGC 4151, Edelson et al.
1996). Upper limits ruled out long
(dynamic) timescales ⇒ light-crossing
timescales.
• NGC 7469 – Wanders et al. (1997), Collier et al.
(1998), Kriss et al. (2000)
UV
Delays found on light-crossing timescales
Important discovery (Sergeev et al.
2005): Delay ∝ Luminosity
Current model: - “Lamp post” model
(E.g., Goosmann et al. 2006)
Expect τ = R/c ∝ T-4/3 ∝ λ4/3
Collier et al. (1998)
Wavelength
Mrk 279
(Gaskell et al., in prep – see
Arav et al. 2007)
PROBLEM 1: Lopt can vary by an order of
magnitude.
⇒ Irradiance would dominate over viscous
energy production in the disk!!
⇒ Main energy source would not be the disk!!
(i.e., our old model is totally inconsistent!)
2. What is this
amazing light
bulb?!
3. Even if it does exist,
WHY DON’T WE
SEE IT?!
Have to have “fullcutoff” fixtures!
(International Dark
Sky Association
approved!)
AVERAGE NORMALIZED DELAYS FOR 14 AGNS
τ ∝ λ4/3 looks good,
but …
. . . IT PREDICTS WRONG UV-OPTICAL DELAY BY
ALMOST AN ORDER OF MAGNITUDE.
Upper limits for
NGC 4151 and
NGC 5548
A NEW MODEL
(astro-ph/0612464)
1. Intrinsic continuum variability has
essentially no wavelengthdependent lag.
2. OBSERVED LAGS PRODUCED
BY CONTAMINATION BY A
SMALL AMOUNT OF LIGHT
WITH A LARGE DELAY FROM
THE DUSTY TORUS.
IR emission comes from hottest dust
= dust at sublimation temperature
Example: NGC 4151 – 2.2 μm lags 0.55 μm by ~ 50 days
Minezaki et al. (2006)
NGC 4151
WHAT DOES HOT DUST LOOK LIKE?
A well-known device
operating at the dust
sublimation temperature:
A candle shines in the
optical!
So hot AGN dust shines in
the optical too!
3. Delay depends linearly on the relative strengths of
the simultaneous component and the delayed one.
I, R
Wien tail of torus
emission
Hot dust
(Hα )
Model also automatically explains why
Delay ∝ Luminosity …
(Sergeev et al. 2005)
… because 2.2 μm delay ( = inner radius of
torus = dust sublimation radius) ∝ L1/2.
Suganuma
et al. (2006)
Hence I and R band lags ∝ L1/2.
Model also predicts: hysteresis
in colour-magnitude (or colourcolour) diagrams.
Similar V fluxes; different K fluxes because of history.
NGC 4151
Based on Minezaki et al.
(2006)
Can quantitatively predict (V-I) vs. V just from observed V
light curve:
• Predict IR delay from IR luminosity – radius relationship.
• Create a “psuedo I” from (simultaneous) V and delayed K.
• Add non-varying host galaxy starlight contributions.
I, R
Torus
RESULTS:
Observed
Model
Bachev & Strigachev (2003)
(Not “high” and “low”
states.)
5. The X-ray / Optical Connection
(Gaskell, 2006 - astro-ph/0701008)
The short-term relationship
between X-ray and optical flux
Gaskell & Klimek (2003)
~ 1 day lag (V – X)
Gaskell & Klimek (2003)
On short timescales the optical can
ignore the X-rays.
Gaskell (2006)
NGC 3516 - Maoz et al. (2002)
KEY FACTS:
• X-ray timescale is short
• Optical timescale can be short too (but
only sometimes).
⇒
ORIGIN OF VARIABILITY IS ELECTROMAGNETIC
Why?
• Needs to move near the speed of light.
• Gas dynamical timescales much too long.
• If we believe in energy equipartition:
Emagnetic ~ Ekinetic.
• The large amplitudes in short times require
relativistic beaming (e.g., Boller et al.
1998, Reeves et al. 2002).
The relationship between optical
and X-ray flux is not simple.
• Sometimes correlated with small lags
• Sometimes X-rays vary without optical
• Sometimes optical varies without the Xrays
A QUALITATIVE MODEL
X-rays, no optical
Both
Optical, no X-rays
Gaskell (2006)
Long-term X-ray/optical relationship
Uttley et al. (2003)
Gaskell (2006)
• Long-term X-ray and optical have close to
zero phase lag (within a day or two)!
• ⇒ OPTICAL AND X-RAY HAVE SAME
UNDERLYING LONG-TERM DRIVING
MECHANISM.
UV
X-rays
Optical
Gaskell (2006)
UV
X-rays
Optical
Gaskell (2006)
Few of the results of this talk
make sense in the standard
accretion disk picture!!
CONCLUSIONS
• Variability is very similar for all AGNs.
• Most variability timescales are too short to be explained
by standard (or modified) accretion disks.
• Variability amplitudes are much too large to be explained
by standard accretion disks.
• Variability timescales correlate with BH mass.
• Simple reprocessing models don’t work.
• Optical/X-ray – long-term correlations with zero lag, but
short-term correlations erratic.
• Long-term variability drives short-term variability
• Electromagnetic and relativistic effects very important
• PLENTY STILL TO EXPLAIN!