Suzaku Observation of the Black Hole Candidate MAXI J1836-194 in a Hard/Intermediate Spectral State
Rubens C. Reis, Jon M. Miller, Mark T. Reynolds, Andrew C. Fabian, Dominic J. Walton
aa r X i v : . [ a s t r o - ph . H E ] M a r D RAFT VERSION N OVEMBER
8, 2018
Preprint typeset using L A TEX style emulateapj v. 5/2/11
Suzaku
OBSERVATION OF THE BLACK HOLE CANDIDATE MAXI J1836–194 IN A HARD/INTERMEDIATESPECTRAL STATE
R. C. R
EIS , J. M. M ILLER , M. T. R EYNOLDS , A. C. F ABIAN AND
D. J. W
ALTON Draft version November 8, 2018
ABSTRACTWe report on a
Suzaku observation of the newly discovered X-ray binary MAXI J1836–194. The source isfound to be in the hard/intermediate spectral state and displays a clear and strong relativistically broadenediron emission line. We fit the spectra with a variety of phenomenological, as well as physically motivated diskreflection models, and find that the breadth and strength of the iron line is always characteristic of emissionwithin a few gravitational radii around a black hole. This result is independent of the continuum used andstrongly points toward the central object in MAXI J1836–194 being a stellar mass black hole rotating with aspin of a = 0 . ± . (90% confidence). We discuss this result in the context of spectral state definitions,physical changes (or lack thereof) in the accretion disk and on the potential importance of the accretion diskcorona in state transitions. Subject headings:
X-rays: binaries – X-rays: individual: MAXI J1836-194 – Accretion, accretion disks – Blackhole physics – Line: profiles – Relativistic processes INTRODUCTIONThe X-ray spectra of X-ray binaries provide important clueson the nature of the compact objects and on the broad prop-erties of the accretion flow. In particular, the various re-flection features endemic to stellar mass black hole binariesin all active states have been successfully used to constrainthe dimensionless spin parameter of various black holes inbinary systems, as well as providing invaluable insight intothe manner in which accretion flow varies with mass accre-tion rate (e.g. Miller 2007). By far the most prominent –and probably the most important – of these reflection featuresis the relativistic iron line appearing at approximately 6.4 to6.97 keV depending to the ionisation state of the emitting ma-terial (e.g. Tanaka et al. 1995).
Suzaku combines exceptionalenergy resolution below ≈
10 keV with broadband observa-tion, and, as such, is unique amongst other current X-ray satel-lites in the study of black hole transients.The observed spectrum often exhibit the presence of ther-mal emission, originating in an optically-thick accretion disktogether with a hard component often referred to as thecorona, and reflection features. There is a complex link be-tween the disk, coronal hard X-rays and reflection emissions(e.g. Done et al. 2007), which manifests in various spectralstates (see Remillard & McClintock 2006 and Belloni 2010).Characterising the driving force between these state transi-tions is a fundamental challenge for both theoretical and ob-servational studies of accretion-flow properties.The prevailing paradigm requires that in quiescence (verylow ˙ m ), the inner accretion disk is fully replaced by an Ad-vection Dominated Accretion Flow (ADAF; e.g. Esin et al.1997). This has led to the idea that the transition betweenactive states is a manifestation of changes in the innermostextent of the accretion disk, so that a transition from the disk-dominated high/soft state to a powerlaw-dominated low/hard state marks the point of the disk recession. The constant pres- Dept. of Astronomy, University of Michigan, Ann Arbor, Michi-gan 48109, USA Institute of Astronomy, University of Cambridge, Madingley Rd.,Cambridge CB3 0HA, UK ence of radio jets in the low/hard state can also be associ-ated, albeit in a qualitative manner, with the truncation radius.Thus, it is clear that knowledge of the inner extent of the ac-cretion disk can have fundamental consequences to our under-standing of the nature of the accretion flow at low ˙ m as well asthe connection between accretion disk, corona and radio jets.This radius can be determined via the study of both the con-tinuum emission from the accretion disk or by the reflectediron emission line, again making Suzaku , with its broad bandcoverage and high spectral resolution, ideal for this science.In fact, the advent of
XMM-Newton and
Suzaku has stronglychallenged the paradigm that the accretion disk is truncated inthe bright phases of the low/hard state in black hole binaries(but see Done & Diaz Trigo 2010; also see Miller et al. 2010).This challenge is exemplified by the recent
Suzaku and
XMM-Newton observations of XTE J1752–223(Reis et al. 2011b) and the 42 ks XMM-Newton observationof XTE J1652–453 (Hiemstra et al. 2011). XTE J1752–223was caught during the decay of its 2009 outburst in boththe intermediate ( Suzaku ) and low/hard ( XMM-Newton )spectral states. Interestingly, in both observations wefound the presence of a strong, relativistic iron emissionline which independently yielded strong constraints onthe inner radius: R inner = 2 . +0 . − . r g and . ± . r g (90% confidence; Reis et al. 2011b) for the intermediate and low/hard state respectively, as well as thermal diskcomponents clearly following the L ∝ T relation expectedfor geometrically-thin accretion disk, thus strongly rulingout disk truncation in either state. Similar results have beenfound for XTE J1652–453, where Hiemstra et al. (2011) findthe disk to be at ≈ r g in the hard/intermediate state , andfor GX 339-4 where the disk does not appear to truncateuntil at least − L Edd (Miller et al. 2006; Reis et al. 2008;Tomsick et al. 2009; Wilkinson & Uttley 2009).In order to determine whether these sources are anomalous,or if state transitions are really not linked with a recession ofthe innermost extent of an optically-thick, geometrically-thinaccretion disk, we urgently need observations at low fractionsof the Eddington limit, down to − and below. However, itis clear that at in some phases of the low/hard state, at least asdefined by McClintock et al. (2006), the disk does not appearto truncate beyond the radius of the Innermost Stable Circu-lar Orbit (ISCO). Whether this is due to a different phase ofthe low/hard state – i.e. an “ADAF state” – or whether thedisk only truly begin to truncate at much lower ˙ m remainsto be seen. Either way, it is clear that the disk plays an im-portant role in state transitions and even jet creation, be it byphysically truncating and allowing for the existence of an in-ner ADAF zone or by dissipating less gravitational energy andallowing for a more powerful accretion disk corona.The shape of the iron line is determined by the relativedepth of the disk within the potential well of the black holeand, as such, conveys information on its spin. Understandingthe role of black hole spin ( a = cJ/GM , − ≤ a ≤ )in shaping accretion flows onto and jets from black holes isan important goal, having strong repercussion in all areas ofastronomy. Stellar-mass black holes, for example, are likelyto gain most of their angular momentum during birth, andtheir spin is a consequence of the supernovae that results inthe creation of the central black hole (see e.g. Miller et al.2011). Knowing the spin distribution for these objects thusprovides a window into the nature of one of the most power-ful explosions in the Universe. At present, we have approx-imately a dozen spin measurements made by the use of therelativistic iron lines (e.g. Miller et al. 2008, 2009; Reis et al.2008, 2009a, 2011b), and a handful obtained from the ther-mal disk continuum (McClintock et al. 2006; Gou et al. 2009;Steiner, Reis, McClintock et al. 2011). However, in order tomake any claim on the possible role of spin on, for example,radio jet power (Fender et al. 2010), we need to increase ourspin demographics.In this paper, we draw on the recent Suzaku
TOO ob-servations of the nearly discovered black hole candidateMAXI J1836–194 to learn about the nature of the innermostaccretion flow in this source, and to increase our black holespin demographics. The following section summarises all theobservations of the source and details the current
Suzaku ob-servation. Section 3 begins by exploring some of the morephenomenological models used to explain the spectra of X-ray binaries and confirms the black hole nature of the cen-tral source. We conclude this section by using a fully self-consistent and physically motivated model to estimate the spinparameter of the black hole in MAXI J1836–194. Section 4summaries our results and discusses the implications for cur-rent ideas of black hole state transitions and interpretations. OBSERVATION AND DATA REDUCTIONMAXI J1836–194 was discovered by the MAXI/GSCobservatory on 2011 August 30 (Negoro et al. 2011). Itscurrent evolution, at various wavelengths, have been reportedin various ATels (Kennea et al. 2011; Cenko et al. 2011;Strohmayer & Smith 2011; Rau et al. 2011; Nakahira et al.2011b; Del Santo et al. 2011; Trushkin et al. 2011;Russell et al. 2011) with the latest report by Russell et al.(2011) strongly suggesting that MAXI J1836–194 is indeeda black hole X-ray binary based on VLT mid-IR detections.Similar conclusions were made by Miller-Jones et al. (2011)based on EVLA radio detections. The 6 ks RXTE /PCA obser-vation of MAXI J1836–194 reported by Strohmayer & Smith(2011) had the spectrum described by an absorbed pow-erlaw with photon index of 1.84. The authors reportedthe presence of an iron emission line at . ± . together with a smeared edge at . ± . and alluded (cid:1) (cid:2)(cid:3)(cid:4) (cid:5) (cid:6) k (cid:8) (cid:9) . k (cid:6) (cid:11)(cid:8) u e )d - dV u (cid:18) (cid:11) (cid:19) V0VVVV0VVdV0V)VV0V)dV0V.V (cid:22)(cid:23)(cid:9)(cid:11)ue(cid:24)(cid:25)(cid:26)(cid:6)u(cid:27)(cid:28)(cid:2)(cid:9)u.V))-Vf-.ruVV8VV8VV9u!"
V )V .V TV iV dV F IG . 1.— Swift/BAT lightcurve for MAXI J1836–194 in the 15–50 keV energy range. The dashed horizontal lines shows theaverage rate and +/- 1 sigma standard deviation for this sourcesince its discovery. The arrow shows the time of the Suzaku observation reported in this paper. Lightcurve available from:http://heasarc.gsfc.nasa.gov/docs/swift/results/transients/weak/MAXIJ1836-194/ to a reflection interpretation of these features. The quoted3–20 keV flux of . × − erg cm − s − is remark-ably similar to that of XTE J1752–223 in a similar state( ≈ × − erg cm − s − ; Γ = 1 . ± . ; Reis et al.2011b).Figure 1 shows the hard X-ray evolution ofMAXI J1836–194 as observed by the Swift/BAT HardX-ray Transient Monitor provided by the Swift/BAT team. Suzaku observed the source on 2011 September 14 for a totalof 10.7 ks during a period where the 15–50 keV flux showeda slight decrease. The three operating detectors constitut-ing the X-ray Imaging Spectrometer (XIS; Koyama et al.2007) on-board of Suzaku were operated in the 1/4 win-dow “burst” mode with both front and back illuminateddetectors in the 3x3 and 5x5 editing modes. Using thelatest
HEASOFT v6.11.1 software package we processed theunfiltered event files for each CCD following the
Suzaku
Data Reduction Guide . Due to the observation having beenpreformed in “burst” mode, we started by producing detailedgood time intervals (GTIs) using the FTOOL XISTIME andsetting the option “bstgti=yes”. New attitude files were thencreated using the
AEATTCOR script (Uchiyama et al. 2008)in order to correct for shift in the mean position of the sourcecaused by the wobbling of the optical axis. The FTOOLXISCOORD was used to create new event files which werethen further corrected by re-running the
Suzaku pipeline withthe latest calibration, as well as the associated screeningcriteria files. The good time intervals provided by the XISteam were also employed in all cases to exclude any possibletelemetry saturations.
XSELECT was used to extract spectralproducts from these event files.In order to estimate the level of pile up suffered by the datawe used the script
PILE EST (Davis 2001) to create a pileupmap out of a Suzaku event data file. After experimentingwith various extraction regions we choose to employ a boxannulus region with a width of 240 pixels ( ∼ ”) and aheight of 290 pixels ( ∼ ”) and an inner radius of 70 pix-els ( ∼ ”). This resulted in a maximum pileup fraction of http://heasarc.gsfc.nasa.gov/docs/suzaku/analysis/ http://space.mit.edu/cxc/software/suzaku/aeatt.html http://space.mit.edu/cxc/software/suzaku/pest.htm ′′ elsewhere on the samechip. Individual ancillary response files (arfs) and redistri-bution matrix files (rmfs) were produced with the script XIS-RESP – which calls the tools XISRMFGEN and
XISARFGEN – with the “medium” input.Finally, we combined the spectra and response files fromthe two front-illuminated instruments (XIS0 and XIS3) usingthe
FTOOL ADDASCASPEC to increase signal-to-noise. The
FTOOL GRPPHA was used to give at least 100 counts perspectral bin in a total of 512 energy channels. The nominalenergy range covered by the XIS detectors is from ∼ . −
12 keV . However there are still calibration issues below ∼ , which are especially severe in the burst clocking mode.Therefore, we do not consider data below 1.2 keV or above10 keV , following the analyses of XTE J1752–223 presentedby Nakahira et al. (2011a). The energy band of 1.6–2.4 keV is also excluded to avoid large systematics uncertainties in theeffective area near the silicon K and gold M edge. The smallereffective area, together with the fact that the out-of-time eventrate is more significant in the BI instrument, as compared tothe FI, means that at the energies considered here (i.e. above1.2 keV ) the BI data contain even larger uncertainties, and forthis reason we did not use XIS1 data in this paper.We processed the The Hard X-ray Detector (HXD;Takahashi et al. 2007) with the standard criteria. The appro-priate response file (ae hxd pinxinome11 20110601.rsp) forXIS-nominal pointing was downloaded and the data werereprocessed in accordance with the Suzaku
Data ReductionGuide. As the non-X-ray background (NXB) file was yet to becreated by the HXD team at the time of writing, we estimatethe NXB by extracting the earth occulted data” (ELV < − ) .Dead time corrections were applied with HXDDTCOR . Thecontribution from the Cosmic X-ray Background (CXB) wassimulated using the form of Boldt (1987), with appropriatenormalisation for the XIS nominal pointing, resulting in aCXB rate of .
019 count s − . The earth occulted NXB andCXB spectra were then combined using MATHPHA to givea total background spectrum, to which a 2 per cent system-atic uncertainty was added. The source spectrum was finallygrouped to at least 100 counts per spectral bin. The PIN spec-trum is restricted to the 15.0–42.0 keV energy range and fitsimultaneously with the XIS data by adding a normalisationfactor which is set to 1.16 in all fits with respect to that ofthe FI spectrum as recommenced by the
Suzaku data analysisguide. To test the robustness of our result, in the followingsections we also investigate the effect of allowing this cross-normalization to vary. All errors reported in this work are90 per cent confidence errors obtained by allowing all param-eters to vary, unless otherwise noted. DATA ANALYSES AND RESULTS
EXPLORING PHENOMENOLOGICAL MODELS
In order to compare the spectral properties ofMAXI J1836–194 with past work on other X-ray bina-ries, we begin by fitting the data with a simple combination of http://suzaku.gsfc.nasa.gov/docs/suzaku/analysis/xisresp This has now become available and we have checked the consistency inthe results. The ”tuned” background has a 15-45 keV flux that is approxi-mately 3% less than the earth-occulted background, and is fully consistencywithin errors. The spectral shape is also fully consistent with one another anddoes not alter the results presented here. (cid:1) (cid:2) (cid:3)(cid:4) (cid:5) gh gs((g((gk(gV (cid:13)(cid:14)(cid:15)(cid:16)(cid:17)(cid:18)t2(cid:21)(cid:15)(cid:22)3 k 5 ( k ) (cid:21) (cid:15) (cid:22) k t (cid:26) (cid:27)(cid:5) (cid:3) (cid:5)(cid:14) (cid:28) t (cid:29) (cid:30) r k (cid:28) r ( (cid:21) (cid:15) (cid:22) r ( ( gk g5 F IG . 2.— Response unfolded νF ν spectrum of MAXI J1836–194. Thetotal, disk and powerlaw components are shown as cyan, blue and green solidlines respectively. Bottom:
Data model ratio to a absorbed
DISKBB plus
POWERLAW fit ignoring the 4–7 keV energy range. an absorbed powerlaw together with a
DISKBB (Mitsuda et al.1984) model. Figure 2 shows this fit with the 4–7 keV rangeignored in order to best model the continuum. The total0.01-100 keV unabsorbed flux is ∼ . × − erg cm − s − of which approximately 42 percent is associated with theaccretion disk. For comparison with the standard work ofMcClintock et al. (2006) and Belloni (2010), we quote the to-tal 2–20 keV unabsorbed flux as ∼ . × − erg cm − s − and a disk fraction of ≈ . . Combined with a spectral indexof ∼ . , this observation of MAXI J1836–194 is consistentwith having caught the source in the hard/intermediate spectral state. A further hint of this is also seen in Fig. 1,where it is clear that the source was observed during a timeof slight decrease in the hard X-ray flux and suggests a shortexcursion away from the low/hard state.In Figure 2, the 4–7 keV energy range is shown above thecontinuum in order to highlight the presence of various fea-tures in this range. A possible explanation used by a numberof authors to account for the residuals seen in Fig. 2 assumesthat these features are a combination of a narrow iron Kα emission line and its associated absorption edge, which, whenarising from the region around a black hole, suffers from ahigh degree of smearing and thus is better described by the SMEDGE model (e.g. Ebisawa et al. 1994). A fit with sucha smeared edge having an energy of 7.11 keV (as expectedfrom neutral iron) and a width of 10 keV was not able to ac-count for the residuals ( χ /ν = 630 . / Model 1a; Table1 and Figure 3), with broad residuals remaining both aboveand below the edge energy. Adding a narrow ( σ = 1 eV ) GAUSSIAN line at 6.4 keV did not improve the residuals inany way. A much better fit is indeed achieved when the ener-gies of both the
GAUSSIAN line and that of the smeared edgeare allowed to be free and the emission line is allowed to bebroad ( χ /ν = 380 . / ; Model 1b; Table 1). The neutralhydrogen density found in these models are also mildly con-sistent with the value of . ± . × cm − presented byKennea et al. (2011) based on a Swift/XRT observation. Wewill discuss possible reasons for the variation in N H , as ob- TABLE 1P
HENOMENOLOGICAL FITS USED TO DETERMINE THE ROBUSTNESS OF THE EMISSION LINE .Model 1a Model 1b Model 2 Model 3a Model 3b Model 3c Model 3d N H ( × cm − ) . ± .
02 0 . ± .
02 0 . +0 . − . . ± .
03 0 . ± .
03 0 . +0 . − . < . E smedge ( keV ) 7.11(f) . +0 . − . — — — — — τ smedge . ± . . ± . — — — — — Γ 2 . ± .
01 2 . ± .
01 1 . +0 . − . . ± .
06 2 . ± .
03 2 . ± .
02 2 . ± . kT disk ( keV ) . ± .
005 0 . ± .
005 0 . ± .
003 0 . +0 . − . . ± .
01 0 . ± .
01 0 . ± . kT electron ( keV ) — — — +3 − +2 +2 +6 τ CompPs — — — . +0 . − . . ± .
02 0 . +0 . − . . +0 . − . N hard . ± .
01 0 . ± .
01 0 . ± .
001 0 . +0 . − . . +0 . − . . ± .
03 0 . ± . N diskbb ( × ) . ± . . ± . . +0 . − . — — — — N CompPs ( × ) — — — +4 − +16 − +13 − +10 − E gauss ( keV ) — . − . — — . ± . — — σ gauss ( eV ) — +80 − — — +160 − — — E laor ( keV ) — — . − . — — . +0 . − . . +0 . N line ( × − ) — . ± . . +0 . − . — . +0 . − . . +0 . − . . +0 . − . W line ( eV ) — ±
90 360 ± — +80 − +20 − +60 − q in — — . ± . — — . +0 . − . . ± . θ (degrees) — — ± — — 25(f) 25(f) r in ( r g ) — — < . — — . +0 . − . . +0 . − . ξ — — — < <
100 70 +80 − +2600 − χ /ν Notes: Results of phenomenological fits with a variety of continuum models. The continuum in Model 1 is assumed to consist of a simplepowerlaw having a normalisation N hard and a DISKBB component. The feature around the 6–7 keV range is modelled with a smeared edge(Model 1a) together with a further Gaussian line (Model 1b). Model 2 replaces the smeared edge and the Gaussian line with a single relativisticline component. Model 3 replaces the DISKBBand powerlaw models with the Comptonization code of Poutanen & Svensson (1996) and themodel PEXRIV of Magdziarz & Zdziarski (1995) respectively. For Model 3, a “slab” geometry was assumed and the reflection option incompPSwas deactivated. A lower limit for the electron temperature of 10 keV was imposed. The feature is modelled with a broad Gaussian(Model 3b) and a relativistic line (Model 3c). Model 3d self consistently convolves the PEXRIV model with the same parameters as therelativistic line since they both originate from the same region. All errors are 90 per cent confidence. served between the different models, in §
4. The scenario sofar presented here however, has an edge energy of ∼ . which is much less than the value for neutral iron absorption;itself a lower limit since iron is likely to be highly ionised inthe inner parts of the accretion disk. In fact, it is more likelythat the edge is trying to compensate for a broad emission fea-ture (see §
4) and, for this reason, although the model is de-tailed in Table 1, we do not believe it to convey any physicalinformation.Given the obvious presence of a broad line feature in Fig-ure 2, we replaced the
SMEDGE model with a
LAOR lineprofile (Laor 1991) as expected if emission is coming fromthe inner disk around a black hole. The line energy is con-strained to lie between 6.4 and 6.97 keV , thus encompass-ing the full range of possible ionisation states of iron. Westart in Model 2 with a simple powerlaw emissivity profilesuch that ǫ ( r ) ∝ r − q in . The outer disk radius was frozenat the maximum value in the model of 400 r g . This modelis detailed in Table 1 and resulted in a satisfactory fit with( χ /ν = 432 . / ). Allowing for a broken powerlaw emis-sivity profile with indices q in within a radius r break and q out beyond, further improved the fit ( ∆ χ = 13 . for 2 degrees offreedom). This fit has an emissivity index of q in ∼ withina radius of ∼ r g and then breaks to q out ∼ . . In boththese instances, the inner radius obtained is very low imply-ing that not only is the central object in MAXI J1836–194 astellar mass black hole, it is also likely to be rapidly spinning.In all models considered so far, the DISKBB component con-sistently required a disk with a temperature of approximately0.45 keV .It has been suggested by a number of authors that this sim- (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)dc(cid:3)V((cid:10)))() (cid:12)(cid:13)(cid:4)(cid:14)(cid:15)(cid:16)d1(cid:18)(cid:4)(cid:19)E (cid:24) (cid:25) (cid:26)(cid:27) (cid:2) (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)dc(cid:25)V((cid:10)))() (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)d2V((cid:10)))() (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)d)(cid:25)V((cid:10)))() F IG . 3.— Data/Model ratio for the various phenomenological models de-scribed in Table 1. From top to bottom: phabs ∗ smedge ∗ ( diskbb + powerlaw ) with the smeared edge energy frozen at 7.11 keV ; phabs ∗ ( laor + diskbb + powerlaw ) ; phabs ∗ ( compps + pexriv ) ; phabs ∗ ( compps + pexriv + gaussian ) ; phabs ∗ ( compps + pexriv + laor ) ; phabs ∗ ( compps + kdblur ⊗ pexriv + laor ) . ple view of a multicolour disk is not appropriate and shouldbe replaced with a broader disk model. Kolehmainen et al.(2011) further suggested that the broad iron line constantlyseen in a number of stellar mass black hole binaries might bean artificial effect caused by the usage of narrower disk com- I n c li n a ti on ( d e g r ee s ) Inner radius (r g )+ 2 4 6 8 I n c li n a ti on ( d e g r ee s ) Emissivity index (q in )+ E m i ss i v it y i nd e x ( q i n ) Inner radius (r g ) + F IG . 4.— Contour plots investigating the effect of the inclination (left) and emissivity index (right) on the inner radius for Model 3d. The centre panel shows thecontour plot for the inclination versus emissivity index. The 68, 90 and 95 per cent confidence range for two parameters of interest are shown in black, red andgreen respectively. The cross marks the global minima. It is clear that even for the most curved continuum where the emission feature would appear the narrowest(i.e. Model 3), the inner extent of the accretion disk, as obtained solely from the phenomenological fit to the iron line profile, is still consistent with not beingtruncated far beyond the innermost stable circular orbit for a Schwarzschild black hole (6 r g ) in the hard/intermediate spectral state observed here. However, dueto the local minima in these phenomenological models we cannot rule out a disk that is mildly truncated at this stage. ponents similar to DISKBB , used here. Despite the noticeableproblems with this interpretation — iron lines are mostly seenin black hole binaries in the low/hard state where the disk tem-perature is relatively cold and its contribution to the spectrumat ≈ is unimportant — we nonetheless test the effectthis may have on the residuals seen between 4–7 keV (Fig-ure 2), by replacing both the DISKBB , LAOR , and
POWER-LAW components with compPS (Poutanen & Svensson 1996)and
PEXRIV (Magdziarz & Zdziarski 1995). The former fullycharacterises the process of Comptonization for a varietyof coronal geometries, electron distributions and seed pho-tons injection geometries. We used the code assuming apurely thermal electron distribution in the corona (Gmin=-1). The
PEXRIV model represents a exponentially cut-offpower law spectrum reflected from ionized material. Whenthe reflection fraction, R > , this models give the sumof the illuminating powerlaw hitting the ionised disk to-gether with the corresponding reflection component (Fe-edgeand Compton hump) but it does not include the Fe Kα emission line. We initially have the ionisation parame-ter ( ξ = 4 πF/n erg cm s − , where F is the incident fluxand n is the number density of hydrogen nuclei), reflectionfraction and inclination of the PEXRIV component frozenat − , 1 and 25 degrees respectively, in linewith the values used for XTE J1752–223 by Nakahira et al.(2011a). The geometry of compPS is chosen to be a “slab”(geom=1) and the covering fraction frozen at 1. Reflectionfrom the compPS component is turned off (rel refl=0).This combination resulted in a unsatisfactory fit with χ /ν = 629 . / . Allowing the ionization of PEXRIV tovary did not resolve the problem ( χ /ν = 572 . / ), with abroad feature clearly present in the residuals between 4–7 keV (Model 3a in Table 1; Figure 2). Further allowing the reflec-tion fraction to change again did not provide a satisfactoryfit ( χ /ν = 546 . / ). The reflection fraction artificiallyshoots up to ∼ . as the PEXRIV model tries to compensatefor the lack of an Fe-emission line by increasing the depth ofthe iron-edge (see Figure 9). Indeed, adding a broad
GAUS-SIAN line (Model 3b) resulted in a dramatic improvement,with ∆ χ = 200 . for 3 degrees of freedom compared toModel 3a. However, a line energy of E Gauss = 5 . ± . is not consistent with emission of iron. Furthermore, a width The cut-off energy is frozen at 300 keV , similar to the value used in thereflection model
REFLIONX described in detail in the next section. of ∼
800 eV is highly suggestive of emission from close tothe black hole where gravitational broadening effects are im-portant, again taking us back to the need of a relativistic ironline.We therefore proceed by replacing the
GAUSSIAN line inModel 3b with the relativistic line expected around a spin-ning black hole. The inclination in the
LAOR model isfrozen at 25 degrees, as with
PEXRIV . This model immedi-ately improved the quality of the fit ( ∆ χ = 15 . for 1 de-gree of freedom) and, more importantly, brought the emis-sion line energy to a range consistent with emission from iron( E Laor ≈ . ). The inner radius as obtained from theiron line profile in Model 3c is again consistent with the cen-tral object in MAXI J1836–194 being a rotating black hole.Allowing the reflection fraction in the PEXRIV model to befree yields R = 0 . ± . , and does not change any of theother fit parameters nor does it affect the quality of the fit( χ /ν = 356 . / ). Importantly, Model 3c as it stand is notphysically consistent. If the Fe line is being emitted in the in-ner parts of the accretion disk, as appears to be the case, thenthe other reflection features should also experience the effectof strong gravity. For this reason we convolve the PEXRIV model — which models the illuminating continuum as well asthe absorption edge of iron and the Compton reflection hump— with the relativistic kernel
KDBLUR (Laor 1991). We forcethe parameters of
KDBLUR to be the same as that of the rela-tivistic iron line. This model (Model 3d) result in a slightlyworst quality of fit ∆ χ = 3 . for the same number of de-grees of freedom as the previous model, however it is nowphysically consistent. Again, allowing the reflection fractionin the PEXRIV model to be free resulted in an improvementof ∆ χ = 4 . for 1 degree of freedom, giving R < . but did not change the line profile in any way, with the innerradius and emissivity index remaining at r in = 3 . +0 . − . r g and q in = 3 . ± . , respectively. We also investigated the ro-bustness of the fit with respect with the cross-normalizationconstant between the XIS and PIN data which we have so farfixed at 1.16. Allowing this to go free barely improved thequality of the fit ( ∆ χ = 1 . for 1 degree of freedom) andrecovered a value of . +0 . − . , fully consistent with the ex-pected value.To expedite the computational time, in all incarnations ofmodel 3 we have assumed that the inner disk inclination has avalue of 25 degrees. As a last step in our exploration of thesephenomenological models, we allow the inclination to be free TABLE 2F
ITS WITH PHYSICALLY MOTIVATED REFLECTION MODELS
Model 4a Model 4b Model 5a Model 5b N H ( × cm − ) . ± . . +0 . − . . ± .
03Γ 1 . ± . . +0 . − . . +0 . − . kT disk ( keV ) . +0 . − . . +0 . − . . +0 . − . N hard < . < .
01 0 . +0 . − . . +0 . − . N diskbb ( × ) . +0 . − . F Illum /F BB — — . ± . . +0 . − . n H ( × ) — — . +0 . − . . +1 . − . N refbhb — — . +0 . − . . +0 . − . N reflionx ( × − ) . ± . . +0 . − . — — q in > . > . > . > . q out . ± . . ± . . +0 . − . . +0 . − . r break ( r g ) . +2 . − . . +2 . − . . +0 . − . . +0 . − . θ (degrees) < < < < ξ +170 − +500 − — — r in ( r g ) . +0 . − . . +0 . − . . +0 . − . (2 . +0 . − . ) a Spin ( a ) — — — . ± . χ /ν Notes: Model 4 is described in XSPEC as phabs ∗ ( diskbb + powerlaw + kdblur ∗ reflionx ) . Model 4b is identical to 4a but we only fitthe 3–10 keV energy range. Model 5 replaces REFLIONX and DISKBB with the fully self-consistent reflection model REFBHB. In all modelsdescribed so far the kernel from the LAOR line profile was used to account for the gravitational effects close to the black hole. In Model 5b,we finally replaces the KDBLUR kernel with the relativistic code RELCONV were the spin is a parameter of the model. In all cases the hardemission illuminating the disk is assumed to be a powerlaw with index Γ . All errors are 90 per cent confidence for one parameter. a Innerradius is not a model parameter and was derived by using the relationship between the spin and ISCO (Bardeen et al. 1972). It is shown heremerely to allow for easy comparison. and investigate any possible degeneracy this might have on theinner accretion radius and emissivity index. Figure 4 showsthe 68, 90 and 95 percent confidence range for both param-eters as a function of inclination as well as the inclination asa function of emissivity index. From Fig. 4, it can be seenthat despite the fact that the inclination is not very well con-strained, ranging from anywhere between 5 and 40 degrees atthe 90 per cent confidence level, the inner radius of the ac-cretion disk, as obtained solely from the breadth of the lineprofile using purely phenomenological models, is still consis-tent with being at or close to the radius of the Innermost StableCircular Orbit around a rotating black hole. The global min-imum which is marked with a black cross in all panels, stillrequires r in < . r g at the 90 percent level however we dosee the presence of a further solution having a radius which ismarginally consistent with the ISCO of a non-rotating blackhole together with a very high emissivity index. In Fabian etal. (2012), we showed that a single power-law emissivity pro-file has only limited validity and that a slope of ∼ is a fairapproximation for an inner disk starting at r in ∼ r g . How-ever, it severely underestimates the profile within r in ∼ r g and is therefore a poor probe of the innermost region around arapidly spinning black hole. In that paper, we argued that if q= 3 is used, then it will likely yield an upper limit to the innerradius and thus a lower limit on the spin if the source is in-deed spinning rapidly. The evidence so far points toward theblack hole in MAXI J1836–194 not being rapidly spinning,and therefore, from theoretical arguments we should expectan emissivity index close to 3, similar to the value found in thethe global minima having an inner radius within r in < . r g .In the following section, we will investigate these possible de-generacies fully using a number of self-consistent reflectionmodels.In this section, we have established beyond any reason-able doubt the presence of a strong and mildly broad emis- sion line associated with iron in the Suzaku spectra ofMAXI J1836–194. We have shown that different continuummodels, as well as phenomenological features (such as theblurred edge component) does not eliminate the need for abroad emission line. Our efforts to “remove” the need for abroad emission line are somewhat artificial since reflection isa natural consequence of a system where hard X-rays are im-pinging on a cold accretion disk. However, it is only by doingso that we can safely rule out any continuum effect on thebreadth of the iron emission line. In the following section,we endeavour to interpret the spectra in a fully consistent andphysical manner.
PHYSICALLY SELF-CONSISTENT MODELLING
In all our previous fits, a broad feature has been shownto be robustly present above a thermal-disk and powerlaw-like continuum, peaking around the energies expected forneutral
Fe I K α emission ( ∼ . ) and highly ionisedH-like Fe XXVI L y α ( ∼ .
97 keV ; See Fig. 2). The nat-ural explanation for this feature is that it is indeed as-sociated with the reprocessing of hard X-ray emission byan accretion disk and, as such, the broad feature observedis the signature of iron fluorescence that has been rel-ativistic broadened due to the strong gravity around thecentral black hole. Similar reflection features are ob-served in a wide range of objects ranging from neu-tron stars (Bhattacharyya & Strohmayer 2007; Cackett et al.2008, 2009a,b; di Salvo et al. 2009; Reis et al. 2009b), stellarmass black holes (Miller 2007; Blum et al. 2009; Reis et al.2009a; Hiemstra et al. 2011; Walton et al. 2012) and AGNs(Tanaka et al. 1995; Fabian et al. 2009; Miniutti et al. 2009;Schmoll et al. 2009; Walton et al. 2010; Nardini et al. 2011;Brenneman et al. 2011; Reis et al. 2011a).Due to the immense diagnostic potential of reflection fea-tures, a large theoretical effort has been devoted to fully char- (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)d
V(cid:8) (cid:9) (cid:10) (cid:11)(cid:12) (cid:2) rEPggEg (cid:17)(cid:18)(cid:4)(cid:19)(cid:20)(cid:21)dt(cid:23)(cid:4)(cid:24)i y V gr yr kr(cid:1)(cid:2)(cid:3)(cid:4)(cid:5)dV(cid:10) (cid:9) (cid:10) (cid:11)(cid:12) (cid:2) rEPggEg (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)dk(cid:10) (cid:9) (cid:10) (cid:11)(cid:12) (cid:2) rEPggEg (cid:23) (cid:4) (cid:24) y d t (cid:28) (cid:29)(cid:2) (cid:11) (cid:2)(cid:18) (cid:30) d (cid:31) y (cid:30) g (cid:23) (cid:4) (cid:24) g i grEyrEV F IG . 5.— From top to bottom:
Response unfolded νF ν spectrum to thebest-fit Model 5b. The total and blurred
REFBHB components are shown ascyan and blue solid lines respectively. The powerlaw continuum falls belowthe y-scale; Data/model ratio for: (4a) phabs ∗ ( diskbb + powerlaw + kdblur ⊗ reflionx ) ; (5a) phabs ∗ ( powerlaw + kdblur ⊗ refbhb ) and(5b) phabs ∗ ( powerlaw + relconv ⊗ refbhb ) . acterising the reflection spectrum expected to arise from suchsystems (Lightman & White 1988; George & Fabian 1991;Matt et al. 1991; Ross & Fabian 1993; Zycki et al. 1994;Nayakshin et al. 2000; Ballantyne et al. 2001; Ross & Fabian2005, 2007; Garc´ıa & Kallman 2010; Garc´ıa et al. 2011).Amongst these, the most widely used reflection model is the REFLIONX code of Ross & Fabian (2005). This model self-consistently calculates the reflection arising from all ener-getically important ionization states and transitions expectedin disks around black holes. At low ionization parame-ters it reproduces the reflection continuum first described byLightman & White (1988), as well as self-consistently calcu-lating the fluorescent lines; at higher ξ , lower Z element be-comes ionized which results in a softening of the reflectionspectrum. We start by using a combination of REFLIONX ,relativistically convolved with
KDBLUR , together with diskemission (
DISKBB ) and a powerlaw illuminating continuum.The photon indices of the
REFLIONX and powerlaw compo-nents are assumed to be the same. The outer disk radius isassumed to be at 400 r g (the maximum allowed by the model)and the iron abundance of MAXI J1836–194 is fixed at solar.This model give a poor fit with χ /ν = 400 . / . Allowinginstead for a broken powerlaw emissivity profile results in asignificant improvement with χ /ν = 379 . / . This mod-els is described in detail in Table 2 (Model 4a) and shown inFig. 5.It is worth noting some similarities and differences betweenthe values obtained from the current reflection model andthose from the phenomenological fits described in the previ-ous section. To start, the high ionization parameter found here( ξ ≈ − ) is similar to that found in the mostphysically motivated version of Model 3 (i.e. Model 3d) andsuggest an intermediate to high ionization state. The moderateto low disk inclination shown in Fig. 4 is also confirmed here,where the current reflection fit suggests θ . ◦ . The disk (cid:1)(cid:2)(cid:3)(cid:4)(cid:5)d )(cid:8) (cid:9) (cid:10) (cid:11)(cid:12) (cid:2) ((r( (cid:15)(cid:16)(cid:4)(cid:17)(cid:18)(cid:19)dt(cid:21)(cid:4)(cid:22)i V ) 3 5 6 7 8 (y (cid:21) (cid:4) (cid:22) k d t !(cid:2) (cid:11) (cid:2)(cid:16) " d $ n k " n ( (cid:21) (cid:4) (cid:22) n ( i yr(yrkyr3 F IG . 6.— Response unfolded νF ν spectrum from Model 4b fit in the 3–10 keV energy range. The total, blurred reflection and disk components areshown as cyan, blue and red solid lines respectively. The powerlaw contin-uum falls below the y-scale. Bottom:
Data model ratio to Model 4b. Thereflection parameters obtained in this range are identical to that obtained bymodelling the full spectra (see Table 2). parameters found here are also in close agreement with thosefound in Model 2, however we cannot compare the parame-ters with Model 3 as the disk was modelled assuming a Comp-tonization model. A major difference between the fit with thereflection over that of the various phenomenological modelsis in the necessity for a broken emissivity profile over that ofa single powerlaw. In model 4a, the disk extends to within ∼ . r g . The steep emissivity occurs in a very narrow annulibetween this radius and ∼ r g , at which point it goes backto the value expected from a purely Newtonian geometry. Itis interesting to note that in the phenomenological model 3d,the inner radius obtained with an emissivity index of ∼ . issimilar to the break radius found here. In the case of the phe-nomenological models, the inner radius was obtained from apure emission line. The approach of using a model such asthe LAOR line profile to obtain spin has many limitations asthe information imprinted by the effects of strong gravity isnot limited to an emission line. In fact, such a limitation isclear in the inconsistency between the high ionization foundhere ( ξ ∼ − ) as well as that found in Model 3d( ξ ∼ − ) with the line centroid of ∼ .
48 keV found in that same model which indicates lowly ionised iron.In Model 3d, a further constraint on the inner radius was inplace by the act of convolving the
PEXRIV model with thesame kernel as the
LAOR line profile. Here, there was an in-terplay between broadening of the emission line componentas well as the iron absorption edge (modelled separately andtherefore not forced to be physical). The final product wasthe apparent presence of an emission line at ∼ .
48 keV , asexpected from cold matter with ξ .
100 erg cm s − , togetherwith the absorption edge from hot hydrogenic iron Ka FeXXVand/or FeXXVI with edge energies & .
85 keV as expectedfrom . ξ . − . This combination acted tomodel a broad feature with only a mild degree of relativisticbroadening. REFLIONX on the other hand, by virtue of theimposed self-consistency in its atomic physics, modelled thesame broad feature by allowing a greater degree of broaden-ing to be attributed to gravity, and hence resulted in a smallerinner radius.A question that generally arises when one is dealing withreflection models is whether the constraints on the various pa-rameters of interest (i.e. inner radius/spin, disk inclination and I n c li n a ti on ( d e g r ee s ) Inner radius (r g ) 1.5 2 2.5 3 I nn e r e m i ss i v it y i nd e x ( q i n ) Inner radius (r g ) 1.5 2 2.5 3 . . O u t e r e m i ss i v it y i nd e x ( q ou t ) Inner radius (r g )+1.5 2 2.5 3 . . . P I N / X I S R a ti o Inner radius (r g )+ 1.5 2 2.5 3 . F ill u m / F BB Inner radius (r g )+ × × × n H Inner radius (r g )+ F IG . 7.— Contour plots investigating the effect of the inclination (top-left), inner (top-centre) and outer (top-right) emissivity indices, PIN/XIS cross-normalisation constant (bottom-left), ratio of illuminating to black body flux (bottom-centre) and disk-surface hydrogen density (bottom-right) on the innerradius for Model 5a. The latter two being a proxy to the common disk ionisation parameter. The 68, 90 and 95 per cent confidence range for two parameters ofinterest are shown in black, red and green respectively. The cross marks the global minima. It is clear that the inner extent of the accretion disk is robust to theseparameters and is well constrained to ∼ . − . r g at the 90 per cent level of confidence. emissivity profile, etc) are driven by anything other than thereflection features. For example; could it be that extreme blur-ring, which would suggest a maximally rotating black hole,is artificially caused by the model trying to “smooth” the softpart of the reflection spectrum to mimic a disk component? Totest this we ignored the XIS data below 3 keV where the diskemission dominates (see Fig. 2) and removed the PIN dataaltogether. What remained is essentially the broad featurewhich we are associating with iron fluorescence emission. Werefit Model 4a after freezing the disk temperature and its nor-malization as well the neutral hydrogen column density andpowerlaw index, as these cannot be constrained from the lineprofile alone. We refer to this as Model 4b in Table 2, andshow this fit in Fig. 6. It is clear that all parameters obtainedfrom this narrow energy range are consistent, within errors, tothat obtained using the full spectra. Allowing Γ to vary doesnot change this conclusion.Until this point, we have made use of REFLIONX whichis designed to reproduce reflection spectra from the accre-tion disks around AGN. However, the disks around stellarmass black holes are significantly hotter, resulting in subtledifferences in the radiation processes, including reprocessing.The higher disk temperatures around stellar mass black holesmeans that Compton broadening is of greater importance andshould therefore be included at the correct level. With this inmind, Ross & Fabian (2007) developed a modified version ofthe previous grid,
REFBHB , in which the atmosphere of the ac-cretion disk is is illuminated not only by the hard, powerlaw-like corona, but also by a further blackbody radiation intrin-sic to the disk. This model self-consistently accounts for thedisk, relativistic line and reflection continuum present in thephenomenological models described in the previous section.The parameters of the model are the number density of hy-drogen in the illuminated atmosphere, n H , the temperature of the accretion disk, kT disk the index of the (assumed pow-erlaw) continuum and the ratio of the total flux illuminat-ing the disc to the total blackbody flux emitted by the disc, F illum /F BB . The disc reflection spectra is again convolvedwith KDBLUR to account for relativistic effects. This model(Model 5a; Table 2 and shown in Figure 5) provides the bestfit yet to the data with χ /ν = 344 . / despite being moreconstrained — compared to purely phenomenological mod-els — by virtue of being physically self-consistent. The inneraccretion disk radius of r in = 2 . +0 . − . r g is consistent withthe values found from the REFLIONX model together witha separate disc component. Assuming that this radius is thesame as the radius of the innermost stable circular orbit, wecan constrain the spin parameter (Bardeen et al. 1972) to be a = 0 . +0 . − . .We again investigate the dependence of the inner radius r in on a number of key parameters including the inclination an-gle, the inner and outer emissivity indices, the cross normal-isation between the XIS and PIN data, the ratio of the illu-minating powerlaw to the blackbody flux and the hydrogennumber density at the disk surface. Figure 7 shows that theinner radius found here of of r in = 2 . +0 . − . r g is extremelyrobust to changes in all the aforementioned parameters. How-ever, in order to make a formal constraint on the spin, we re-place KDBLUR with the sophisticated variable-spin relativis-tic smearing model
RELCONV (Dauser et al. 2010). Figure 8shows that the spin of the black hole in MAXI J1836–194 iswell constrained to be a = 0 . ± . at the 90 per centconfidence range. In the following, section we will discuss,amongst other things, the current strength and limitation ofthe various models used throughout this work and highlightsome of the factors contributing to the tight constraint on thespin parameter of MAXI J1836–194. (cid:1)(cid:1) a(cid:3) (cid:1)(cid:5) (cid:1)3 (cid:7) (cid:9) (cid:12)(cid:13)(cid:14)(cid:15)71(cid:17)S F IG . 8.— Goodness-of-fit versus spin for Model 5b. It is clear that the blackhole in MAXI J1836–194 is rapidly rotating. A maximally rotating Kerr or astatic Schwarzschild black hole are both rejected at greater than the 3 σ levelof confidence. DISCUSSIONThe fractional contribution of the disk component to the to-tal 2–20 keV flux ( ∼
25 % ) places the current observation inthe hard/intermediate state as defined by Belloni (2010) andin an intermediate state between the low/hard and high/soft state definition of McClintock et al. (2006) and. However, wecan see from figures 5 and 6 that the observed continuum isnot dominated by the powerlaw-like component expected tooriginate from the corona, but rather it is mostly reflectiondominated. Similar “reflection-dominated” spectra are seenin a number of narrow line Seyfert 1s and quasars during their low state (e.g. 1H0419-577, Fabian et al. (2005); NGC4051,Ponti et al. (2006); PG1543+489, Vignali et al. (2008); Mrk335, Grupe et al. (2008); PG1535+547, Ballo et al. (2008);PG2112+059, Schartel et al. (2010)). In this scenario, themajority of the X-ray reprocessing (reflection) occurs in theinner region of the accretion disk where strong gravitationallight bending is expected to occur (Martocchia & Matt 1996;Miniutti & Fabian 2004). Such behaviour is expected as a re-sult of strong light bending, where the reflected flux is en-hanced over the inner regions as a result of gravitational fo-cusing of the X-ray continuum down towards the black holeand onto the disk. The decrease in the number of X-rays thatcan escape as part of the continuum thus causes the source toappear reflection dominated.Wilkins & Fabian (2011) showed that a possible conse-quence of strong gravitation effects is an increase in the emis-sivity profile of the disk. Classically, the emissivity profile isexpected to be flat in the region directly below the source,while tending to r − when r >> h , where the flux re-ceived by the disc from the source falls off as the inversesquare of the distance with a further factor of /r arising fromthe cosine of the angle projecting the ray normal to the discplane. However, as briefly mentioned above and detailed inWilkins & Fabian (2011), strong gravity can potentially act tofocus more of the direct continuum into the inner parts of thedisk as well as increase the disk area being radiated – the latteras a consequence of gravitational warping. In this particulartreatment, these factors causes for a substantial steepening ofthe emissivity profile in the inner regions.This is indeed a possible explanation for what is observed inMAXI J1836–194, where within a radius of . r g the emis-sivity index is consistently > (see Table 2) and beyond it ! !" !"" " $" %" &" ’( ) * $ + , - ./ /1 + (cid:1) $ + (cid:1) ! + ( ) * (cid:1) ! :);<=/1>?+@ PEXRIV components in Models 3d (blue) and3c (red). The unblurred PEXRIV (red) is as expected from a “cold”( ξ = 70 erg cm s − ) accretion disk being illuminated by a Γ = 2 . pow-erlaw without being relativistic convolved. Compare this to the componentexpected from a moderately ionised ( ξ = 5200 erg cm s − ; blue) disk at3.3 r g illuminated by a Γ = 2 . powerlaw and the REFLIONX modelhaving both the absorption edge as well as the iron emission line (black). goes closer to the classical value of 3. Both the reflection-dominated spectrum and the high emissivity profile seen heresuggest that the primary X-ray continuum is located withina few gravitational radii of the black hole. Looking at Fig-ure 1, it is indeed possible that the corona briefly “collapsed”down close to the black hole causing the decrease in the hardX-ray flux seen during this Suzaku observation. If this is thecase, state transitions (at least that between the high/soft and low/hard ) should be seen to be much more tightly associatedwith changes in the corona as opposed to physical changes inthe accretion disk.In this work we have investigated a variety of possible mod-els striving towards a physically motivated interpretation forthe observed spectrum. It is worth stressing that equally goodfits – and at times (albeit not in this work) statistically betterfits – can be obtained with a purely phenomenological com-bination of components severely lacking in physical consis-tency. A case in point is Model 3c where the presence ofthe – clearly broad – iron line requires the emission to oc-cur from deep within 6 r g , yet the direct X-ray and reflectioncontinuum appear somehow exempt from the effect of stronggravity. When we account for relativistic effects, the ioniza-tion parameter of the disk increases by nearly two orders ofmagnitude from ξ ≈ to ≈ − . This changeis accompanied by a hardening of the photon index Γ possi-bly due to the powerlaw trying to compensate for the strongersoft emission and the weaker Compton hump of the blurredionised reflector in comparison to the unblurred, cold reflec-tor (see Fig. 9).A further point to note is that the apparent decrease in theequivalent width of the relativistic line profile (from 270 eV to180 eV from model 3c to model 3d) is clearly a consequenceof the, physically inconsistent, lack of gravitational blurring.From Fig. 9 we can see that the absorption edge in PEXRIV becomes much more smooth and symmetric after the compo-nent is convolved with KDBLUR and the ionization increases.Due to the decoupling between the LAOR line profile and PEXRIV , this smooth edge can act somewhat like an emission0line conspiring against the LAOR line component and thus de-creasing the equivalent width of latter. In order to couple theemission line strength with the absorption edge and all otherreflection features, we replaced the phenomenological com-bination of LAOR + PEXRIV with the REFLIONX reflectiongrid. The first thing to notice from Table 2 (model 4) is thatthis recovered the high ionization value from Model 3d butnow in a self-consistent manner. Nonetheless, the strong re-quirement for an emission line with an equivalent width ofat least ≈ 180 eV (Model 3d) is highly indicative of a broadline, adhering to the strong criterion of Reis et al. (2010) ar-guing against an accretion disk truncated far from the ISCO.The precise value of the neutral hydrogen column densitytowards MAXI J1836–194 is not known, however, based onSwift/XRT data which extends to much lower energies ascompared to the current Suzaku observation, Kennea et al.(2011) showed that a likely value is (2 ± . × cm − .A similar value is found here using our best model. Figure 9highlights how differing models impact the energies below ∼ where we would begin to see the curvatures expectedfrom such a low column density. The fact that our data cutsoff at 1.2 keV means that, in order to break the degeneraciesbetween the – essentially disk – continuum (see Fig. 2) andneutral column density, one must use a self consistent modelfor the full spectra. Such is the case for REFBHB .It is well known that in the hot inner regions of an ac-cretion disk the observed disk spectrum suffers from theeffects of electron scattering which results in an observed(colour) temperature, T col , which is higher than the effec-tive blackbody temperature, T bb , by approximately a factorof f col = T col /T bb (Ross et al. 1992). This “colour correc-tion factor” has been shown to have a value of . ± . (Shimura & Takahara 1995) for a wide range in luminosity, aslong as the disk effective temperature remains below ∼ (Davis et al. 2005), as is the case here. Above this tempera-ture, disk self ionisation can lead to an increase in f col how-ever it is found to be consistently below ∼ (Merloni et al.2000). The effective temperature in the REFBHB (Model 5)of kT = 0 . +0 . − . keV is precisely as expected from thevalue of the colour temperature from the DISKBB componentin Model 4 ( kT /kT = 1 . ± . ) .Lastly, although we cannot directly compare ξ between REFLIONX and REFBHB , we can estimate this value basedon the various parameters output in Table 2. Since the fluxof the blackbody in the model is related to its temperatureby the F bb ∝ T relation, we find F bb = (5 . +1 . − . × erg cm − s − . Combining this F Illum /F BB = 1 . ± . gives an illuminating flux of (6 . +2 . − . ) × erg cm − s − .The ionization parameter, which is defined as ξ =4 πF/n erg cm s − , is then found by using n = n H =2 . +0 . − . × H cm − . In this manner, we find ξ = 3300 ± − , in perfect agreement with the values foundfor both REFLIONX in both Models 4a and 4b and the blurred PEXRIV in Model 3d.All current methods of measuring black hole spin rely onboth the assumption that the accretion disk extends to the in- nermost stable circular orbit, as well as that emission withinthis radius is negligible. The latter is indeed valid for thestandard model of black hole accretion (Shakura & Sunyaev1973), where within the ISCO – in the region often referredto as the plunging region – there is no angular momentumtransport, and the region cannot support an X-ray corona toirradiate the material that is ballistically plunging onto theblack hole. However, it was shown by Krolik (1999) and in-dependently by Gammie (1999) that when magnetic fields areconsidered, the B-fields within the ISCO may be amplified toa point where the magnetic energy density can be comparableto the rest mass energy of the accreting material and may leadto the creation of an active inner X-ray corona. Shafee et al.(2008) investigated the effect of magnetic torque within theplunging region around a non-rotating black hole and con-cluded that “...magnetic coupling across the ISCO is relativelyunimportant for geometrically-thin discs”.A further study specifically aimed at addressing the robust-ness of the iron line/reflection fitting technique in diagnosingblack hole spin was presented by Reynolds & Fabian (2008).The authors used a high resolution three-dimensional magne-tohydrodynamic simulation of a geometrically-thin accretiondisc to show that the density of the plunging material dropsprecipitously over a very small radius within the ISCO. Thissudden drop in density results in the material being highlyphotoionised and suppresses any significant iron line emissionas well as all other reflection features from within the ISCO.The study by Reynolds & Fabian (2008) concluded that fora non-rotating black hole where the ISCO is at 6 r g , the re-flection edge – defined by the authors as the innermost ra-dius from which significant reflection emission is seen – isat approximately 5.8 r g . Furthermore, the discrepancy be-tween the true ISCO and the inferred radius diminishes asone considers more rapidly rotating black holes as is the casein MAXI J1836–194 (see Fig. 8). From an observational per-spective, it is also worth noting that strong support for thepresence of an “inner edge” in the accretion disk surroundingblack holes is provided by decades of empirical evidence asshown in Steiner et al. (2010).The culmination of this work is that the recently discoveredsystem, MAXI J1836–194, is indeed a stellar mass black holebinary, having a central black hole rotating with a spin pa-rameter of a = 0 . ± . (90% confidence). This strongconstraint is a result of being able to successfully, and mostimportantly self-consistently, model the reflection featuresclearly present in the Suzaku spectra. ACKNOWLEDGEMENTSRCR thanks the Michigan Society of Fellows and NASA. Afurther thank you goes to the Suzaku team, which scheduledthis TOO observation, and to the MAXI team for providingprompt notice of this new source to the astronomical commu-nity at large. RCR is supported by NASA through the Ein-stein Fellowship Program, grant number PF1-120087 and is amember of the Michigan Society of Fellows. ACF thanks theRoyal Society. This work was greatly expedited thanks to thehelp of Jeremy Sanders in optimising the various convolutionmodels. REFERENCESBallantyne D. R., Ross R. R., Fabian A. C., 2001, MNRAS, 327, 10Ballo L., Giustini M., Schartel N., et al., 2008, A&A, 483, 137Bardeen J. M., Press W. H., Teukolsky S. A., 1972, ApJ, 178, 347 Belloni T. M., 2010, in Lecture Notes in Physics, Berlin Springer Verlag,edited by T. Belloni, vol. 794 of Lecture Notes in Physics, BerlinSpringer Verlag, 53Bhattacharyya S., Strohmayer T. E., 2007, ApJ, 664, L1031