Transcript ULX accretion states

ULX accretion state(s)
Roberto Soria
University College London (MSSL)
Thanks also in random order to Doug Swartz, Manfred Pakull, Hua Feng,
Christian Motch, Luca Zampieri, Fabien Grise’, Jess Broderick, Tim Roberts
Outline
Canonical accretion states and state transitions
Accretion states as mass indicators for ULXs
Classifying X-ray properties of ULXs into “states”
Comptonization-dominated state
Slim disk state
(High) hard state
and where is the standard high-soft state?
Mechanical vs radiative states (jets or no jets)
1. Canonical BH states (short review)
Mostly defined from stellar-mass Galactic BHs
State transitions in Cyg X-1
harder – non-thermal – radio loud
October 1972
softer – thermal – radio quiet
“Canonical” BH accretion states
(From the 1980s… eg, Cyg X-1, GX339-4)
m
1
F(0.3-10 keV)
Very high state
Heavily Comptonized disk
Radio flaring
0.1
High/Soft state
Standard disk
Radio quiet
0.01
Low/Hard state
Jet? Corona?
ADAF? CENBOL?
Radio loud
1 keV
5 keV
E
Disk + pl
GRS1758
Power-law
disk
Radio lobes
(ATCA 5 GHz)
(Hao, Soria et al 2010, in preparation)
Canonical state evolution of Galactic BHs
LC  Ldisk
Very high
LEdd
High/soft
Thin flow
Quiet
No jet
0
Thick flow
Noisy
Jet
0.5
Low/hard
1
LC ( LC  Ldisk )
“Canonical” BH accretion states
Power-law
IC in inner disk
or base of outflow
(+BMC from outflow?)
M M
Edd
1
Thermal
Optically-thick
emission from disk
0.1
Power-law
0.01
Truncated disk + ADAF
Full disk + jet + corona
0.001
High/soft state = disk-blackbody spectrum
Ldisk  LX ~
2 4
Rin Tin
~
2
4
M BH Tin
Ldisk  LX ~ m
1/ 4
Tin ~ m
Ldisk  LX ~ LEdd  10 M BH erg/s
38
High/soft state can be used to estimate BH mass
2. Accretion states as indicators
of BH mass in ULXs
(where no direct BH mass measurements)
ULX luminosity function
Chandra survey of ~200
nearby star-forming galaxies
Number of sources N(>L)
(Swartz et al 2010, in prep)
100
Steepening
or cut-off?
D Swartz’s talk today
10
0.3-10 keV isotropic L
of the most luminous ULXs
1
1E39
1E40
Intrinsic 0.5—8 keV Luminosity (1039 erg/s)
Most or all of these sources consistent
with “heavy” stellar BHs up to ~ 70 Msun
Different class? IMBHs?
Cartwheel: ~ 1E41 erg/s
M82:
~ 1E41 erg/s
NGC2276: ~ 1E41 erg/s
NGC5775: ~ 8E40 erg/s
ARP240:
~ 7E40 erg/s
NGC7714: ~ 7E40 erg/s
(ESO243-49: ~ 5—8 E41 erg/s)
Let’s take a ULX at LX ~ 1E40 erg/s:
What accretion state do we expect?
If BH mass > 1,000 Msun
we expect to find it in the low/hard state (hot corona, jet)
If BH mass ~ 100 – 1,000 Msun
we expect to find it in the high/soft state (diskbb, no jet)
If BH mass ~ 30 -- 100 Msun
we expect to find it in some kind of very high state
(mildly super-Eddington, Comptonized disk)
If BH mass ~ 10 -- 30 Msun
we expect to find it in a new kind of strongly super-Edd state
(thick outflows, beamed?)
If BH mass > 1,000 Msun
If BH mass ~ 100 – 1,000 Msun
If BH mass ~ 30 -- 100 Msun
super-stellar
stellar
Direct collapse of a metal-poor star (Z ~ 0.1)
with initial mass ~ 120—150 Msun
If BH mass ~ 10 -- 30 Msun
Core mass up to 70 Msun
+ fallback + accretion
If ULXs are stellar (M < 100 Msun )
is there a big difference between:
-- “normal” stellar BHs (M ~ 5 – 20 solar)
-- “heavy” stellar BHs (M ~ 30 – 100 solar)
?
Let’s look at the apparent luminosity
1.3 10  M BH   3

L

 1  ln m 
b

 M sun   5
38
Beaming ~ 0.2—0.5
BH mass >~ 10
Accretion rate > 1
For a fixed super-Eddington luminosity,
the required accretion rate decreases with BH mass
M ~ M BH m ~ M BH e
 L

1 

 LEdd 
~ M BH e
 aM BH
For super-Edd ULXs, the expected bright lifetime
increases almost exponentially with BH mass
M donor M donor aM BH
~
~
e
M
M BH
Assuming beaming ~ 2 (quasi-isotropic),
a ULX with Lx ~ 1 E 40 erg/s may have:
M
Rc
BH mass
(Msun)
m
(Msun / yr)
(km)
10
20
10,000
100
2E-3
5E-5
1E6
20,000
30
10
7E-6
2,500
100
1
2E-6
900
3. Observational classification
of ULX states
Main problem: spectral coverage only in 0.3-10 keV
Typical spectral “states” of ULXs
Lx
0.3
1
5
10
E (keV)
Typical spectral “states” of ULXs
Lx
…but very few (if any) diskbb ULXs
0.3
1
5
10
E (keV)
Holmberg II X-1 (Lx ~ 2E40 erg/s)
“soft excess”
kT ~ 0.15 keV
Holmberg II X-1 (Lx ~ 2E40 erg/s)
M99 X1
Power-law spectrum
Photon index G = 1.6
Lx ~ 2 E 40 erg/s
(Soria & Wong 2006)
NGC 5474 X1
(Swartz & Soria 2010, in prep)
Broken power-law:
G= 0.75 below 3 keV,
G = 1.4 above 3 keV
Lx ~ 2.5E40 erg/s
NGC 5575 X1
Hard power-law: G = 1.5
Lx ~ 7E40 erg/s
NGC 4631 X4
Power-law + soft excess
G ~ 1.8
Lx ~ 2E39 erg/s
Tin ~ 0.2 keV
Rin ~ 1500 km
(Soria & Ghosh 2009)
NGC 4631 X5
Simple power-law
G ~ 2.1
Lx ~ 5E39 erg/s
ULX
L0.3-10
M82 X1
2-10 E 40
2 E 40
2-3 E 40
2.7 E 40
2 E 40
4-6 E 39
1.7 E 40
3 E 40
2 E 40
1 E 40
2 E 40
3 E 40
1-3 E 40
4-6 E 39
2 E 40
7 E 39
1.5 E 40
1 E 40
1.5 E 40
1.3 E 40
1 E 40
0.7-1 E 40
M82 X2
NGC925
IC342 X1
IC342 X2
Ho IX
Ho II
NGC1313 X1
NGC1313 X2
NGC5055
NGC4559 X1
NGC4559 X2
NGC1068
NGC5474
NGC3628
NGC5408
G
+ soft x?
curved
(curved)
1.2 +/- 0.1
1.3-1.5
2.0 +/- 0.3
Y
comp / sd
1.6-1.8
comp / sd
1.9
Y
comp / sd
1.6-1.8
2.5 +/- 0.2
2.4 +/- 0.1
1.7-1.9
2.0-2.5
2.5 +/- 0.1
2.3 +/- 0.1
1.8-2.1
1.8-2.1
0.9 +/- 0.1
(~1)
1.8 +/- 0.1
2.6-2.7
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
sd?
broken po
(comp)
(comp)
HS state
diskbb?
ULX
L0.3-10
G
NGC5775 X1
7 E 40
1 E 40
1 E 40
3 E 40
1 E 40
5 E 39
4 E 40
1.5 E 39
2 E 40
1.5 E 40
0.7-2 E 40
0.7-2 E 40
1 E 40
3 E 40
1-1.5 E 40
1 E 40
0.7-0.9 E 40
7 E 40
4 E 40
4-12 E 40
7 E 40
1.7 +/- 0.2
1.9 +/- 0.2
1.5 +/- 0.1
1.8 +/- 0.1
1.8 +/- 0.1
1.8 +/- 0.2
1.2 +/- 0.1
1.2 +/- 0.2
1.6 +/- 0.1
1.9 +/- 0.1
1.3-1.8
1.1-1.4
1.7 +/- 0.1
2.5 +/- 0.5
1.6-2.0
1.2 +/- 0.1
2.1-2.4
2.1 +/- 0.2
(2.6 +/- 0.5)
1.9 +/- 0.2
1.5 +/- 0.5
NGC5775 X2
NGC1365 X1
NGC1365 X2
M99
NGC4579
Antennae X11
Antennae X16
Antennae X42
Antennae X35
Antennae X44
Antennae X?
NGC5204
NGC7714
Cartwheel N10
Arp240
+ soft x?
curved
Y
Y
(curved)
Y
comp
Y
comp
curved
HS state
Some have pure power-law spectra
(usually hard, photon index < 2)
Most ULXs classified as
Power-law + soft excess + downturn at E ~ 5 keV
Some have curved spectra:
thermal but not standard disk
Fitted by slim-disk model (p-free disks)
photon trapping & advection, outflows
(S Mineshige’s talk)
Power-law + soft excess + downturn at E ~ 5 keV
Likely physical interpretation:
Inner disk heavily Comptonized – covered or replaced
by scattering-dominated region with Te ~ a few keV
+
Standard disk at large radii
Expected from theory when mdot ~ 10
L ~ 2-4 LEdd
inner disk becomes effectively thin, hotter (a few keV),
scattering dominated, (scattering) ~ a few
M BH ~ 30  100 M sun
Inner disk heavily Comptonized – covered or replaced
by scattering-dominated region with Te ~ a few keV
+
Standard disk at large radii
Because it is the most common ULX state, sometimes called
“Ultraluminous state”
(T Roberts, J Gladstone)
Disk and “power-law” components
Standard disk
“reprocessing” region
Thermal spectrum
Power-law spectrum
Ldisk  30% LX
L po  70  100% LX
Large Rc
Low Tin
Low fqpo
Ldisk
Confusing definitions of ULX temperatures
(claims that “ULXs have hot disks” or “ULXs have cool disks”)
Or here?
ULXs are here?
Standard disk
(Soria 2007)
0.1
0.5
1
Tin
Ldisk
Confusing definitions of ULX temperatures
(claims that “ULXs have hot disks” or “ULXs have cool disks”)
Inner hot region
Outer standard disk
(soft excess)

Ldisk  Tin
(Soria 2007)
0.1
0.5
Slim disk
Ldisk  Tin2
Standard disk
4
Ldisk  Tin
1
Tin
Slim-disk models suggest L ~ 1 -- a few LEdd
“Warm” scattering model suggests L ~ 1 -- a few LEdd
Either way, most ULXs should have M ~ 30—100 Msun
Hard power-law ULXs still not well understood
No clue on BH mass yet
ULXs never lose scattering corona
LC  Ldisk
ULXs?
LEdd
High/soft
Thin flow
Quiet
No jet
0
Thick flow
Noisy
Jet
0.5
Low/hard
1
LC ( LC  Ldisk )
X1 2006
X1 2007
NGC1365 X1, X2
X1: Lx = 3E40 (in 2006)
5E39 (in 2007)
G ~ 1.8
X2: Lx = 4E40 (in 2006)
1.5E39 (in 2007)
G ~ 1.2
X2 2006
X2 2007
(Soria et al 2007,2009)
ULXs may not follow canonical state transitions
ULXs do not settle into high/soft state
(never collapse accretion flow into a thin disk)
Direct transitions low/hard to ultraluminous state?
Saturated Comptonization with Te ~ 5 keV?
Decrease of scattering electron Temp
T ~ 100 keV
T ~ 10 keV
(Galactic BHs)
(ULXs)
Increase of scattering optical depth
 ~ 0.1
 ~ a few
(Galactic BHs)
(ULXs)
State transition cycle is driven by 2 parameters:
Accretion rate
“something else” (ang mom? magnetic energy of the inflow?)
Cygnus X-1 never properly switches
to a disk-dominated state
Cyg X-1
GX339-4
(Belloni 2009)
(Zhang et al 1997)
Seyfert 1 galaxy Ark 564 behaves like a ULX
Ark 564
GX339-4
(Belloni 2009)
Perhaps most AGN are always dominated
by scattering corona, not pure disk
4. Radiative and mechanical output
ULXs have strong winds (shock-ionized bubble nebulae)
Do they also have jets?
Do ULXs also have jets?
LC  Ldisk
ULXs?
LEdd
High/soft
Thin flow
Quiet
No jet
0
Thick flow
Noisy
Jet
0.5
Low/hard
1
LC ( LC  Ldisk )
NGC1313 X2
ULX bubbles
Shock-ionized nebulae
with E >~ 1E52 erg and d >~ 100 pc
See talks by M Pakull, D Russell
Holmberg IX X1
Grise’ et al 08
IC342 X1
Pakull & Mirioni 02, 03
Feng & Kaaret 08
Pakull & Mirioni 02, 03
Grise’ et al 08
Non-nuclear radio jet with long-term-avg power ~ 5 E 40 erg/s
in a microquasar of NGC7793
Accretion state with jet power ~ maximum ULX luminosities
M W Pakull’s talk
Pakull, Soria & Motch 2010, Nature, accepted
Summary
Accretion states are a BH mass indicator
L < LEdd, M < 100 Msun
L < LEdd, 100 < M < a few 1000 Msun
If VH or slim disk state
If high/soft state
(see Hua Feng’s talk)
Most ULXs dominated by p-l or Compt. component
Many have soft excess + p-l + high-energy break (“ULX state”)
inner disk modified by scattering-thick region at T ~ a few keV
L ~ 1 – a few LEdd , M ~ 30 – 100 Msun
Some ULXs have hard power law spectrum
Direct evolution between low/hard and “high/hard” state?
Very few ULXs are found in the high/soft state (never thin disk)
We expect ULX to have jets.
Observational challenge to find them. States with jet power ~ rad power
Director’s cut for this talk
Two or 3 ULXs are in a weird “supersoft state”, T <~ 0.1 keV
Like Galactic SS sources (= nuclear burning WDs)
But can a WD reach L ~ 1E39 erg/s
Photosphere of massive outflows around a BH?
HLX1 in ESO243-49 showed a (brief) state transition
from power-law dominated to pure thermal
True high/soft state?
True IMBH?
S Farrell’s talk
Why do some BHs lack a thermal dominant state?
Different BH mass range? (ULXs 5 times bigger? 100 times?)
That should not matter
Different BH spin? (why?)
That seems very contrived
Different mode of mass transfer?
No. ULXs are Roche Lobe fed, like LMXBs
Different magnetic field?
Most Galactic BH transients have low-mass donor stars
strongly magnetized accretion flow?
Most ULXs have OB-type donor stars
weakly magnetized accretion flow?
Why do some BHs lack a thermal dominant state?
Possible effect of the magnetic field
Corona may be produced via irradiated disk evaporation
(balance between disk evaporation and condensation…Liu & Taam 2007,2009)
Mass evaporation rate scales with thermal conductivity
(Meyer-Hofmeister & Meyer 2006)
Heat conduction strongly reduced in magnetized plasmas
(Chandran & Cowley 1998)
Most Galactic BHs have low-mass (magnetic) donor stars
(strongly magnetized accretion flow….less evaporation into corona?)
Most ULXs and AGN have non-magnetic accretion flows
(weakly magnetized accretion flow….more evaporation into corona…
…denser, thicker corona… more difficult to collapse it into pure disk state?)
New discovery:
ULX & bubble in
NGC 5585
(d ~ 7 Mpc)
SDSS image
Check with Matonick & Fesen’s Ha survey
300 pc
Chandra image
ULX with Lx = 5 E 39 erg/s
Magellan image (BVR)
Liu & Soria (August 09)
Galex
New discovery:
NGC 7793 S26
(d ~ 3.9 Mpc)
S26 nebula discovered by Blair & Long 1997
Radio nebula by Pannuti et al 2002
X-ray counterpart identified by Pakull et al 2008
X-ray “triple source” in S26
X-ray core + hot spots
Proof of collimated jet
Core (active BH): power-law spectrum
G  1.4 L0.38  6 1036 erg/s
Hot spots: thermal spectrum
kT  0.3  0.9 keV
L0.38  2 1037 erg/s
Chandra spectra of core and hot spots in S26
(Pakull, Soria & Motch 2010; Soria et al 2010)
5.5 GHz (ATCA)
9.0 GHz (ATCA)
Radio spectral index
Ha map
5.5 GHz contours
X-ray core/hot spots
LHa  3 1038 erg/s
FWHM = 250 km/s
(~ expansion velocity)
Size: 290 x 130 pc
Core not detected
(2001 CTIO image)
HeII 4686
VLT image 2002
Nebula emission
Core emission:
EW ~ 30 A
(consistent with
Wolf-Rayet star)
He II 4686 map
with Ha contours
Shock ionization models
suggest v(shock) ~ 275 km/s
Density (ISM) ~ 1 cm-3
Zoomed-in view
of the S lobe
(Magellan image 2009)
Core (BH)
Optical counterpart:
B ~ 23 mag
Radio hot spot
X-ray hot spot
Energy in the bubble
Standard bubble expansion model
(self-similar solution, Weaver et al 1977)
 Pj t
r  0.76 
 

3 1/ 5




v  (3 / 5)(r / t )
Mechanical power Pj ~ 3 x 1040 erg/s
Characteristic age ~ 2 x 105 yrs
Total energy E ~ 1053 erg
Most of it is thermal energy of protons and ions
+ work to inflate the bubble against ISM pressure
(expanding at v ~ 250 km/s)
Main properties of S26
S26 has the same power as a ULX but in the jet
current X-ray luminosity << long-term average jet power
S26 nebula is 2 x larger and a few times more powerful than SS433/W50
Collimated jet
First evidence of steady collimated jet at accretion rates > Eddington?
P >~ 1040 erg/s
Ultraluminous X-ray sources
Ultrapowerful jet sources
Radio hot spots & lobes = synchrotron emission
X-ray hot spots = thermal plasma emission
Bright optical core with HeII 4686 emission (Wolf-Rayet? Accretion disk?)
Comparison between S26 and quasars
BH jets/winds ionize a gas bubble of radius
S26
 Pj t
r  0.76 
 

3 1/ 5




Typical quasar
Pj ~ 3 x 1040 erg/s
Pj ~ a few x 1046 erg/s
Active for ~ 2 x 105 yrs
Active for ~ 5 x 108 yrs
ISM densities ~ 1 cm-3
IGM/ISM densities ~ 0.01-1 cm-3
Can shock-heat a bubble
of size R ~ 100 pc
Can shock-heat a bubble
of R ~ a few hundred kpc
Conclusions
BHs with super-Eddington accretion can be detected as ULXs
(X-ray selected = radiation-dominated by definition!)
Most ULXs are likely to be due to super-Edd accretion
rather than intermediate-mass BHs.
(M82 X1 is perhaps unique exception so far)
Many ULXs also have powerful winds
(Mechanical power in addition to the X-ray emission)
Some super-Eddington BHs may be jet dominated
but radiatively faint (S26 in NGC7793)
Relative power in the jet and radiation
during super-Edd accretion is a fundamental issue
to understand quasar feedback
Low-mass and High-mass X-ray binaries
HMXBs found in starburst or actively starforming galaxies
Number of HMXBs proportional to star formation rate
LMXBs found in elliptical galaxies and old bulges
Number of LMXBs proportional to stellar mass of a galaxy
(Swartz et al 2003)
Luminosity function
is steeper for LMXBs
ULXs
LX ~ 3 1039 erg/s
Sources in Ellipticals
(LMXBs)
Sources in Spirals/Irr
(HMXBs)
We have discovered the optical counterpart of HLX1
by subtracting the diffuse stellar component of ESO243-49
(Soria et al 2010)
Properties of the optical counterpart:
R ~ 23.8 +/- 0.3 mag
V ~ 24.5 +/- 0.3 mag
Two possibilities:
Old, massive globular cluster in ESO243 (at 100 Mpc)
(like Omega Cen, mass ~ 1E6 solar masses)
IMBH in the core of a globular cluster?
Foreground M-star in the Galactic Halo (at 1—2 kpc)
Neutron star LMXB in the Galactic Halo?
Young star formation in ESO243-49 (UV emission at 2000 Ang, R contours)
Swift/UVOT (Soria et al 2010)
HLX1
But is it related to the HLX1 or just a chance association?
X-ray spectra are power-law + soft thermal component
Thermal component has kT ~ 0.15 keV
Accretion disk around an IMBH (at d ~ 100 Mpc)
It could be
Surface emission from a faint neutron star LMXB
in the Galactic Halo (d ~ 1—2 kpc)
XMM spectra
Swift spectrum
X-ray lightcurve shows rapid state transitions
Aug 2008
Apr 2010
State transitions of an IMBH accretion disk
or intermittent accretion onto a neutron star surface?
Conclusions
HLX1 was called a “proven” IMBH
We found an optical counterpart to HLX1
From optical and X-ray properties,
we argue that there are still 2 possibilities:
IMBH in the core of a globular cluster in that distant galaxy
in that case, L(0.3—10 keV) ~ 1E41 – 1E42 erg/s
Foreground neutron star LMXB in the Galactic Halo
in that case, L(0.3—10 keV) ~ 1E32 – 1E33 erg/s
(Personally, I would bet my money on the neutron star scenario)
Radiative MHD simulations by Ohsuga et al 2009
Slim disk
(ULX?)
Standard disk
(high/soft)
ADAF
(low/hard)