1. Ultraluminous X-ray Sources
Tim Roberts
ULXs in the interacting galaxy pair NGC 4485/4490 (Gladstone & Roberts also poster B.11)

2. A definition
ULX: an X-ray source in an extra-nuclear region of a galaxy with an observed luminosity in excess of 1039 erg s-1
Heterogeneous population - includes some recent supernovae - but bulk of sources are black holes accreting from a secondary star
The Antennae - Chandra ACIS
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

3. A new class of black hole?
But Eddington limit for spherical accretion:
LEdd ~ 1.3 × 1038 (M/M) erg s-1
hence ULXs contain  10 M compact objects – larger still if accretion sub-Eddington – massive black holes.
Not super-massive BHs (MBH  106 M); fall to Galactic centre in a Hubble time due to effects of dynamical friction.
Too massive for stellar remnants (3M  MBH  18M).
Are we observing a new, 102 – 105 M “intermediate mass” class of accreting black hole (IMBHs; e.g. Colbert & Mushotzky 1999)?
Again pretty self-evident – but worth adding lots of emphasis to this slide as IMBHs are still very sexy.
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

4. X-ray evidence for IMBHs
X-ray spectroscopic evidence – cool accretion discs (Miller et al. 2003).
NGC 1313 X-1
T  M-0.25
Animation – first shows T (temp of accretion disc) prop to Mass of BH to power of – i.e. bigger black hole, cooler disc. Next two clicks bring up Miller’s measured temp for ULXs, then typical temp for stellar mass BH. Final click shows mass inferred from Miller’s paper, due to cooler disc.
kTin ~ 0.15 keV
→ ~ 1000 M BHs
c.f. kTin ~ 1 keV for stellar BHs
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

5. Tim Roberts - Ultraluminous X-ray Sources
LX – kTin relationship
IMBH candidates occupy separate part of parameter space to stellar-mass BHs.
Strong evidence for IMBHs as new class underlying luminous ULXs.
From Miller et al. (2004)
LX  T4
Be brief – self-evident!
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

6. Vanishing IMBHs problem
But some problems with IMBHs, most notably…
X-ray luminosity function (XLF), normalised to star formation rate, unbroken over 5 decades.
XLF break at ~ 0.1 LEdd for 1000-M IMBHs.
No other source population switches off at 0.1 LEdd like this.
From Grimm, Gilfanov & Sunyaev (2003)
Break at ~ 2 × 1040 erg s-1
Emphasize that IMBHs certainly not a perfect solution – some severe problems!
Big point here is that all other sources (NSs, BHs, AGNs) easily surpass a tenth Eddington accretion rate – rather strange that ULXs do not if they’re IMBHs…
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

7. The link with massive stars
High mass stars can feed stellar mass black holes at a sufficient rate to produce the extreme X-ray luminosity
From Gao et al. (2003)
Potential X-ray luminosities for accretion onto a 10 M BH from 2 – 17 M secondaries (Rappaport, Podsiadlowski & Pfahl 2005)
Populations of ULXs (10+) detected in bright starbursts - ULXs must be short-lived, so cannot all be IMBHs
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

8. Physical processes
Still need to break the Eddington limit; suggested methods include:
Relativistic beaming (e.g. Körding et al. 2002)
Radiative anisotropy (e.g. King et al. 2001)
Truly super-Eddington discs (e.g. Begelman 2002; Heinzeller & Duschl 2007)
Can combine at least two of the above, e.g. King (2008) - within Rsph local energy release is kept ~ Eddington by driving a bi-conical outflow; so apparent line-of-sight Bolometric luminosity is
This summarises a lot of stuff very briefly. Probably best not to go into detail of models. Note on Super-Edd mass transfer based on figure (from Rappaport et al. paper). Note blue stellar companions from ID’ed counterparts (usually HST data – references omitted).
For beaming factor b and super-Eddington rate M/MEdd
. .
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

9. Evidence from our own Galaxy
Super-Eddington luminosities are seen!
GRS has intermittently exceeded LEdd over its ~15 yr outburst (Done et al. 2004)
SS433 is super-critically accreting (perhaps exceeding M/MEdd by >103) - if seen face-on it would be an ULX (Fabrika & Mescheryakov 2001, Poutanen et al. 2007)
. .
SS433: cartoon showing jet precession & inclination
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

10. Tim Roberts - Ultraluminous X-ray Sources
A pause for reflection
Dichotomy
X-ray evidence such as extreme luminosities and cool accretion discs point to IMBHs, but…
Other evidence stacking up in favour of smaller black holes.
Which one is the correct interpretation?
This slide is simply a bridge into the second half of the talk (my recent stuff).
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

11. X-ray timing – PSDs & break frequency
Break frequencies in PSDs related to black hole mass and accretion rate (McHardy et al. 2006)
But most ULXs show little or no variability power (Feng & Kaaret 2005)
Break feature in NGC 5408 X-1 ~3 mHz (Soria et al. 2004; Strohmayer et al. 2007) implies mass of M
Frequency regime probed by XMM for bright ULXs
Adapted from Vaughan et al. (2005)
Scaling of break frequencies with mass, assuming accretion at mdotEdd
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

12. ULX QPOs
Two ULXs with known QPOs - both luminous with LX > 1040 erg s-1
Cannot be beamed
Scaling arguments from Galactic black holes - masses ~ M if in known state (talk by Zampieri; Casella et al. 2008)
Double QPO in NGC 5408 X-1 (from Strohmayer et al. 2007)
QPO in M82 X-1 (from Strohmayer et al. 2003)
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

13. EPIC-pn light-curve of Ho II X-1
Ho II X-1: timing
Goad et al. 2006
Ho II X-1 is a good example of a ULX with little variability power - can we explain this using known accretion states?
Not disc-dominated
Insufficient power for high or classic very high states
Energy spectrum not low/hard state
Similar to “χ”-class of GRS in VHS?
Band-limited PSD - but don’t see variability, so must be at high-f  MBH < 100 M.
EPIC-pn light-curve of Ho II X-1
(0.3 – 6 keV, 100 s binning)
Animation – show on second point – emphasizes effectively zero power measured in PSD of Holmberg II.
Logical argument is that zero power rules out high state (note that we expect to see red noise power, as we are dominated by the coronal component)
Alternatives are very high state or low state – energy spectrum looks like VHS.
But VHS is mainly red noise power too – except for some states seen in GRS 1915
In these states PSD noise limited to narrow band at high frequency. If BH is much bigger would expect to see this noise shifted to lower frequency – but we don’t. So BH must be small.
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

14. ULX spectra revisited Look at best archival XMM-Newton data
Stobbart, Roberts & Wilms 2006
Look at best archival XMM-Newton data
Demonstrate that keV spectrum fit by a broken power-law in all of the highest quality data
Invalidates IMBH model - hard component is not a simple power-law
Disc
Power-law
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

15. ULX spectra vs Galactic black holes
Physical accretion disc plus corona model: cool discs (kT ~ keV), optically-thick coronae ( ~ )
ULXs operate differently to common black hole states, but…
“Strong” VHS in XTE J (Done & Kubota 2006)
Disc appears cool as its inner regions are obscured by an optically-thick corona.
from Kubota & Done (2004)
“ultraluminous branch” (from Soria 2007)
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

16. A new, ultraluminous accretion state?
Spectrum defined by apparently cool disc, power-law turning over at > 2 keV. Little or no variability power present. Occurs at extreme accretion rates
Low hard state in GX339-4 vs a classic ULX, Ho IX X-1
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

17. The importance of winds
Hydrodynamical simulations of extreme accretion rates (M >> MEdd) onto stellar-mass black holes - Ohsuga (2006, 2007)
Extreme wind driven - column ~ 3  1024 cm2 at the poles, much higher elsewhere
Explains coronae, lack of variability power, giant nebulae…link to high-Z QSOs, Galactic-scale feedback • •
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

18. Other explanations for spectral break
Kerr disc models (Makishima et al. 2000)
“Slim” accretion discs (e.g. Watarai et al. 2000)
Accretion disc structure changes at highest accretion rates (close to the Eddington limit).
Model disc profile T(r)  r -p; standard disc has p = 0.75, slim disc p = 0.5.
Recent work finds p ~ 0.6 for ULXs (e.g. Tsuneda et al. 2006, Vierdayanti et al. 2006, Mizuno et al. 2007).
Fully comptonised VHS with spectrum modified by ionised fast outflow (Goncalves & Soria 2006).
Common thread: high accretion rate, small black holes (MBH < 100 M).
Animation – use at Goncalves & Soria paper point. This is an illustration of their model. It illustrates the “bite” taken out of the spectra by absorption from ionised material in a fast-moving wind (Roberto can certainly explain this better than I can!).
Make the points pretty much as listed!
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

19. A multi-wavelength perspective
Optical - counterparts and “beambags” (cf. Pakull & Grisé 2008)
Bubbles also seen in radio (e.g. Lang et al. 2007)
Spitzer observations of NGC AGN-like emission lines from ULXs (Vazquez et al. 2007)
Identified ULX counterparts are blue - OB stars (e.g. Liu et al. 2004, Kuntz et al. 2005)
Nebula around Ho IX X-1 (Grise & Pakull 2006)
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

20. New HST imaging of ULXs ACS WFC F606W F330W F435W F606W
Roberts, Levan & Goad (2008) - arXiv: v1
ACS WFC F606W
Early F supergiant? NB. high extinction
No counterpart, tho’ very high extinction
Consistent with late O or early B star
F330W F435W F606W
mF606W = 23.9
mF606W > 26
mF606W = 24.9
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

21. New HST imaging of ULXs (2)
ACS WFC F606W
Young stellar cluster? (MV ~ 8 - 9)
OB Star (odd colours?): U-B ~ -1.4, B-V ~ 0.1
Real ULX, or related to background galaxy?
F330W F435W F606W
mF606W = 22.0
mF606W = 24.9
mF606W = 25.6
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

22. Are these really secondary stars?
High LX will affect optical emission; reprocessing in accretion disc becomes more important to optical light as black hole mass increases
Stellar heating: stars may be later types than initial colour IDs suggest (late B, not late O/early B) (Patruno & Zampieri 2008; Copperwheat et al. 2007) - small black holes
Alternatively, IMBHs may be favoured (Madhusudhan et al. 2008)
Stellar-mass BHs
IMBHs
From Madhusudhan et al. (2008)
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

23. The goal: mass functions
Urgency to finding counterparts: race to get first ULX mass function
Best way to resolve mass controversy!
He II 4686Å line from accretion disc of NGC 1313 X-2; 300 km s-1 shift (Pakull et al. 2006). Could be used to constrain RV curve, hence constrain ULX black hole mass
Radial velocity curve from extragalactic Wolf-Rayet black hole binary IC 10 X-1. Uses He II 4686Å line to constrain mass function, find a black hole mass of ~ M (Silverman & Filippenko 2008)
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources

24. Tim Roberts - Ultraluminous X-ray Sources
So, what are ULXs?
Bulk of evidence - few keV X-ray spectral breaks, star formation link etc - argues most ULXs are extreme accretion rate, small (< 100 M) black holes
ULX is an accretion state, not a source class
Cannot rule out some larger IMBHs - NGC 5408 X-1, M82 X-1 and HLXs (with LX > 1041 erg s-1) are the best candidates?
Mass functions are within reach - will resolve the controversy for at least some ULXs
Wednesday 28th May 2008
Tim Roberts - Ultraluminous X-ray Sources