Crossing the Eddington limit: examining disc spectra at high accretion rates
Introduction Super-Eddington accretion states in ultraluminous X-ray sources (ULXs) Broadened disc ULXs: ~Eddington rate accretion? A new study comparing bright sub-eddington accretion discs with broadened disc ULXs Motivation Results Implications 2
Definition ULXs are defined as: Extra-Galactic Non-nuclear Bright X-ray point sources This is an empirical definition It is important to study ULXs in the context of fainter Xray binaries 3
ULX spectra The highest quality XMMNewton ULX spectra differ from sub-eddington states Rules out sub-eddington massive black holes Most ULXs are in a new accretion state Characterised by a soft excess And a high energy break Gladstone et al. (2009) 4
The ultraluminous state Three spectral types of ULXs: Broadened disc Hard ultraluminous Soft ultraluminous Initially suggested that this could be a sequence with increasing accretion rate But we now think that inclination is important too Gladstone et al. (2009) 5
Hardness luminosity diagram Soft ultrauminous LX (erg s-1) Hard ultralumious Broadened discs below ~3 1039 erg s-1 f(1 10 kev) / f(0.3 1 kev) Sutton et al. (2013) 6
The most luminous ULXs Hard ultraluminous Soft ultraluminous Variable soft ultraluminous Key feature: funnelshaped supereddington outflow Inclination and mass accretion rate determine the observed spectral and timing properties Modified from Middleton et al. (2011) 7
Hardness luminosity diagram Soft ultrauminous LX (erg s-1) Hard ultralumious Broadened discs below ~3 1039 erg s-1 f(1 10 kev) / f(0.3 1 kev) Sutton et al. (2013) 8
Broadened disc spectra From a comparison with the hard/soft ultraluminous sources Broadened disc ULXs may be the subtle transition to two-component spectra e.g., M33 X-8, Middleton et al. (2011) 9
Sub-Eddington accretion discs Differentiated by mass accretion rate, to first-order 2-components: accretion disc and 'power-law' Thin accretion disc component dominates at high accretion rates in the thermal dominant state Ultra-Soft State Thermal Dominant State Very High State Low/Hard State Done et al. (2007) 10
Slim discs As L LEdd radiation pressure dominates and the disc becomes geometrically 'slim' As the scale height increases, advection becomes important Resulting in a broader, less peaked disc spectrum Dotan & Shaviv (2011) 11
Broadened disc spectra Broadened disc ULXs may be an extension of the thermal dominant state With geometrically slim, advection dominated discs 0.27 LEdd 0.86 LEdd CXOM31 J004253.1+411422 Straub et al. (2013) 12
A new study: motivation To test if sub-eddington thermal dominant spectra can really be approximated by thin disc models in soft X-rays (below 2 kev) Carry out a comparison of the most luminous subeddington thermal dominant spectra and broadened disc ULX spectra What is the accretion physics in broadened discs? What are their Eddington ratios? 13
BHB Sample selection Obtained RXTE light curves for 21 BHBs out to LMC (Zhang 2013) Searched for XMM-Newton and Swift observations during periods where RXTE data exceeded 1038 erg s-1 Eliminated obvious non-thermal dominant spectra and GRS 1915+104 8 observations of 3 sources: GX 339-4, LMC X-3 and LMC X-1 GX 339-4 14
Methods Extract BHB spectra Re-sample the BHB spectra such that they are of comparable quality to the ULX sample (~25000 counts) Extract broadened disc spectra from ULXs with LX<3 1039 erg s-1 Fit all 3 sets of spectra with various models appropriate for thin discs, slim discs and bright ULXs Extract covariance spectra to test for evidence of multiple spectral components 15
Results: disc plus power-law We apply this test to these BHBs They are all consistent with having broadened disc shaped spectra LMC X-3 M31 ULX1 vfv We use a disc plus power-law model to classify ULX spectra Ratio Energy (kev) 16
Results: thin disc model Kerrbb in XSPEC, with parameters: Spin Black hole mass Accretion rate Theoretically appropriate for L/LEdd 0.3 M31 ULX1 vfv LMC X-3 Ratio Energy (kev) 17
Results: thin disc model The kerrbb model is not broad enough to reproduce the BHB and most ULX spectra An additional hard powerlaw can account for some of the broadening in GX339-4 No strong evidence of this in other objects 18
Results: thin disc model The model parameters imply that broadened disc ULXs contain subeddington (L/LEdd~0.2 0.3) massive stellar black holes But there are problems with interpreting this physically As it requires near ubiquitous maximal spin in the ULXs Broadened disc ULXs Thermal dominant BHBs 19
Results: two-component ULX model Multi-colour disc + Comptonisation Fits most of the broadened disc ULX spectra And most of the (ULX quality) thermal dominant BHB spectra M31 ULX1 vfv LMC X-3 Ratio Energy (kev) 20
Results: two-component ULX model Similar model parameters in both the ULXs and BHBs Broadened disc ULXs Thermal dominant BHBs Except for normalisation But, the model is clearly only phenomenological in the BHBs 21
Results: slim disc model p is the exponent of the radial dependence of the disc temperature Fits most of the ULX spectra And most of the (ULX quality) BHB spectra LMC X-3 M31 ULX1 vfv p-free disc Ratio Energy (kev) 22
Results: slim disc model Similar parameter ranges in both ULXs and BHBs Except normalisation p = 0.75: thin disc p < 0.75: slim disc We find p < 0.75, even at ~0.1 LEdd Broadened disc ULXs Thermal dominant BHBs 23
Timing: covariance spectra Covariance is the correlated variability relative to some reference band: N 1 σ = ( X i X )(Y i Y ) N 1 i=1 2 cov Xi and Yi = count rate in the ith time bin, in the energy band of interest (X) and reference band (Y) N = time bins per light curve segment With some normalisation Here we show fractional covariance spectra 24
Timing: BHB covariance spectra GX 339-4 GX 339-4 LMC X-3 Constant fractional covariance is ruled out in the top plot of GX 339-4 Implying multiple spectral components Consistent with an additional power-law No strong evidence of a hard power-law in other observations 25
Timing: covariance spectra Covariance spectra could also be extracted in two of the ULXs Both are consistent with single-component energy spectra NGC 253 XMM2 NGC 4736 ULX1 26
Implications Either broadened disc ULXs are sub-eddington massive stellar black holes (~40-90 M ) But poor fit statistics and ubiquitous near maximal spin in the broadened disc ULXs makes this unlikely Or, broadened disc spectra persist even below ~0.3 LEdd in the black hole binaries Radiation pressure is insufficient to support a slim disc at such low Eddington ratios Magnetic pressure has been suggested Or quenching by magnetically driven winds mimics a slim disc 27
Summary ULXs below LX ~ 3 1039 erg s-1 have broadened disc spectra and may represent ~Eddington accretion Emerging 2-component spectra have been suggested, but we find no strong evidence of this Instead, they appear to be dominated by a single broad, disc-like component Some bright thermal dominant BHBs are similar in shape to broadened disc ULXs Broadened disc ULXs may have thin disc spectra at ~0.1-0.3 LEdd Or, accretion discs appear to remain 'slim' down to ~0.1 LEdd But this implies ~40-90 M black holes with maximal spin Some other effect smoothly takes over from radiation pressure 28