1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass{emulateapj}
3: %\lefthead{Dewangan et al.}
4:
5: \slugcomment{Accepted for publication in ApJ {\it Letters}}
6:
7: \def\mes{M{\'e}sz{\'a}ros}
8: \newcommand\asca{{\it ASCA}}
9: \newcommand\sax{{\it BeppoSAX}}
10: \newcommand\chandra{{\it Chandra}}
11: \newcommand\rosat{{\it ROSAT}}
12: \newcommand\rxte{{\it RXTE}}
13: \newcommand\xmm{{\it XMM-Newton}}
14: \newcommand\ftools{{\it FTOOLS}}
15: \newcommand\s{{\rm~s}}
16: \newcommand\ks{{\rm~ks}}
17: \newcommand\mhz{{\rm~mHz}}
18: \newcommand\mpc{{\rm~Mpc}}
19: \newcommand\pc{{\rm~pc}}
20: \newcommand\hz{{\rm~Hz}}
21: \newcommand\kev{{\rm~keV}}
22: \newcommand\ev{{\rm~eV}}
23: \newcommand\kms{\ifmmode {\rm~km\ s}^{-1} \else ~km s$^{-1}$\fi}
24: \newcommand\Hunit{\ifmmode {\rm~km\ s}^{-1}\ {\rm Mpc}^{-1}
25: \else ~km s$^{-1}$ Mpc$^{-1}$\fi}
26: \newcommand\ctssec{\ifmmode {\rm~count\ s}^{-1} \else ~count s$^{-1}$\fi}
27: \newcommand\ergsec{\ifmmode {\rm~erg\ s}^{-1} \else
28: ~erg s$^{-1}$\fi}
29: \newcommand\funit{\ifmmode {\rm~erg\ s}^{-1}\;{\rm cm}^{-2} \else
30: ~ergs s$^{-1}$ cm$^{-2}$\fi}
31: \newcommand\phflux{\ifmmode {\rm~photon\ s}^{-1}\;{\rm cm}^{-2}
32: \else ~photon s$^{-1}$ cm$^{-2}$\fi}
33: \newcommand\efluxA{\ifmmode {\rm~erg\ s}^{-1}\;{\rm cm}^{-2}\;{\rm
34: \AA}^{-1} \else ~erg s$^{-1}$ cm$^{-2}$ \AA$^{-1}$\fi}
35: \newcommand\efluxHz{\ifmmode {\rm~erg\ s}^{-1}\;{\rm cm}^{-2}\;{\rm
36: Hz}^{-1} \else ~erg s$^{-1}$ cm$^{-2}$ Hz$^{-1}$\fi}
37: \newcommand\cc{\ifmmode {\rm~cm}^{-3} \else cm$^{-3}$\fi}
38: \newcommand\FWHM{\ifmmode {\rm~FWHM} \else ${\rm~FWHM}$\fi}
39: \newcommand\Msun{\ifmmode M_{\odot} \else $M_{\odot}$\fi}
40: \newcommand\Lsun{\ifmmode L_{\odot} \else $L_{\odot}$\fi}
41: \newcommand\ltsim{\raisebox{-.5ex}{$\;\stackrel{<}{\sim}\;$}}
42: \newcommand\gtsim{\raisebox{-.5ex}{$\;\stackrel{>}{\sim}\;$}}
43: \newcommand\hbeta{\ifmmode {\rm H}\beta \else H$\beta$\fi}
44: \newcommand\Kalpha{\ifmmode {\rm K}\alpha \else K$\alpha$\fi}
45: \newcommand\nh{\ifmmode N_{\rm H} \else N$_{\rm H}$\fi}
46: \usepackage{graphicx}
47: \usepackage{here}
48:
49:
50: \begin{document}
51:
52:
53: \title{Black Hole Mass of the Ultraluminous X-ray source M82 X--1}
54:
55: \author{Gulab C. Dewangan\altaffilmark{1}, Lev
56: Titarchuk\altaffilmark{2} \& Richard E. Griffiths\altaffilmark{1}}
57: \altaffiltext{1}{Department of Physics, Carnegie Mellon University,
58: 5000 Forbes Avenue, Pittsburgh, PA 15213 USA; {\tt email: gulabd@cmu.edu} }
59: \altaffiltext{2}{Center for Earth Observing and Space Research, George
60: Mason University, Fairfax, VA 22030; and US Naval Research
61: Laboratory, Code 7655, Washington, DC 20375-5352}
62:
63: \begin{abstract}
64: We report the first clear evidence for the simultaneous presence of a
65: low frequency break and a QPO in the fluctuation power spectrum of a well
66: known ultraluminous X-ray source (ULX) in M82 using long \xmm{}
67: observations. The break occurs at a frequency of $34.2_{-3}^{+6}\mhz$.
68: The QPO has a centroid at $\nu_{QPO} = 114.3\pm1.5\mhz$, a coherence $Q
69: \equiv \nu_{QPO}/\Delta\nu_{FWHM} \simeq 3.5$ and an amplitude (rms) of
70: $19\%$ in the $2-10\kev$ band. The power spectrum is approximately
71: flat below the break frequency and then falls off above the break
72: frequency as a power law
73: with the QPO superimposed. This form of the power spectrum is
74: characteristic of the Galactic X-ray binaries (XRBs) in their high
75: or intermediate states. M82 X-1 was likely in an intermediate state
76: during the observation. The EPIC PN spectrum is well described by a
77: model comprising an absorbed power-law ($\Gamma \sim 2$) and an
78: iron line at $\sim 6.6\kev$ with a width $\sigma \sim 0.2\kev$ and
79: an equivalent width of $\sim 180\ev$. Using the well established
80: correlations between the power and energy spectral parameters for
81: XRBs, we estimate a black hole mass
82: for M~82 X-1 in the range of $\sim 25 -
83: 520M\odot$ including systematic errors that arise due to the
84: uncertainty in the calibration of the photon spectral index versus QPO
85: frequency relation.
86: \end{abstract}
87:
88: \keywords{accretion, accretion disks --- stars: individual (M82 X-1)
89: --- X-rays: stars}
90:
91: \section{Introduction}
92: Ultra-luminous X-ray sources (ULXs) are extra-nuclear point X-ray
93: sources with luminosities exceeding $\sim 10^{39}{\rm~erg~s^{-1}}$.
94: These objects were first discovered with {\it Einstein} observations
95: (reviewed by Fabbiano 1989).
96: %Later they were detected
97: %in large numbers by {\it ROSAT} (Colbert \& Mushotzky 1999; Roberts \&
98: %Warwick 2000; Colbert \& Ptak 2002) and recently with {\it Chandra}
99: %and {\it XMM-Newton} (e.g, Swartz et al. 2004)
100: %from many galaxies.
101: The nature of ULXs
102: continues to be an enigma, since their adopted isotropic high energy
103: output surpasses the Eddington limit of even the most massive stellar
104: mass black holes (BHs), sometimes by large factors.
105: %
106: Several models have been proposed to explain the high luminosities of
107: ULXs. The most popular is the ``intermediate mass
108: black hole (IMBH)'' with mass $M_{BH} \simeq 10^2 -
109: 10^4{\rm~M_{\odot}}$ (e.g., Colbert \& Mushotzky 1999, hereafter CM99) bridging
110: the gap between stellar mass BHs in XRBs and
111: super-massive BHs in active galactic nuclei \footnote{In CM99
112: the BH mass was evaluated using the best-fit parameters
113: of the {\tt XSPEC} BMC model (see Shrader \& Titarchuk 1999
114: for details)}.
115: Other popular models include XRBs with
116: anisotropic emission (King et al. 2001), beamed XRBs with relativistic
117: jets directly pointing towards us i. e., scaled down versions of
118: blazars (Mirabel \& Rodriguez 1999), and XRBs with super-Eddington
119: accretion rates (Begelman 2002). Several observations suggest that
120: ULXs may be similar or scaled up versions of XRBs. Discovery of orbital modulations from several ULXs
121: (Bauer et al. 2001; Sugiho et al. 2001) implies their binary nature. Recent
122: \chandra{} and \xmm{} observations of ULXs show soft X-ray excess
123: emission which has been interpreted as the emission from accretion
124: disks with temperatures in the range $\sim 100 -500\ev$ (see
125: Shrader \& Titarchuk 2003 and review by Miller \& Colbert 2003 and
126: references therein). Observations of spectral transitions between
127: low/hard and high/soft states in two ULXs in IC~342 (Kubota et al.
128: 2001) further demonstrate their similarity with the XRBs.
129: The recent observation of a break at a frequency of $\sim 2.5\mhz$ in
130: the power density spectrum (PDS) of the ULX in NGC~5408
131: suggests a BH mass of $\sim 100M\odot$ (Soria et al. 2004).
132:
133: The enigmatic nature of ULXs can be understood by determining their BH
134: masses. The most reliable method is to measure the mass function
135: through the secondary mass and orbital parameters such as
136: velocities, period, orbit size etc., which can be measured only if the
137: secondary is optically identified. There are two cases of optical
138: identification (Kuntz et al. 2005; Liu et al. 2004); however,
139: the orbital parameters of
140: any ULX binary system are yet to be measured. Good progress can still be made
141: by establishing a direct physical connection between ULXs and XRBs.
142: This can be done by comparing the characteristic time scales such as
143: those associated with the low frequency quasi-periodic oscillations
144: (QPOs). However, this comparison is ambiguous without the information
145: about a low frequency break, the shape of the PDS and
146: energy spectrum.
147:
148: %
149: M82 X-1 is one of the brightest ULX (CXO M82 J095550.2+694047, source 7
150: in Matsumoto et al.
151: 2001) that is well suited for a determination of the shape of the PDS.
152: \chandra{} High Resolution Camera observations showed that the source
153: is unresolved and off-nuclear (Matsumoto et al. 2001). The ULX has no
154: optically bright counterpart (Kaaret et al. 2001): there is a likely
155: IR counterpart which is a super star cluster (Kaaret et al. 2004). Recently,
156: Strohmayer \& Mushotzky (2003), hereafter SM03, discovered a QPO in
157: M82 X-1 at a frequency of $54{\rm~mHz}$ based on a $27{\rm~ks}$ \xmm{}
158: observation. They also found QPOs in the $50-100{\rm~mHz}$ frequency
159: range in the \rxte{} data.
160: Fiorito \& Titarchuk (2004), hereafter FT04 apply a new method to
161: determine the BH mass of M82 X-1. The method uses the spectral index-QPO
162: low-frequency correlation that has been recently established in
163: BH XRBs GRS 1915+105, XTE J1550-564, 4U 1630-47, and
164: others (Titarchuk \& Fiorito 2004, hereafter TF04) . Using scaling
165: arguments and the correlation derived from
166: Galactic BHs, they conclude that M82 X-1 is an intermediate mass BH
167: of order 1000 M$_{\odot}$.
168: We revisit the BH mass estimate in M82~X-1 using new data
169: on the power spectrum and energy spectra.
170:
171: \section{Observation, Data Reduction \& Analysis}
172: M82 was observed using \xmm{} on 2001 May 5 for $30\ks$ and again on 2004 April
173: 21 for $103\ks$. The first observation led to the discovery of a
174: $54\mhz$ QPO from the ULX (SM03). Here we consider the EPIC data from
175: the second observation only. The EPIC PN and MOS cameras were operated
176: in the full frame mode using the medium filter. We used the SAS
177: version 6.1 and the most recent calibration data base to
178: process and filter the event data. Examination of the background rate
179: above $10\kev$ showed that the observation is completely swamped by
180: the particle background after an elapsed time of $71.5\ks$ and this
181: latter period was therefore
182: excluded from the rest of the analysis.
183:
184: In the central regions of M82, \chandra{} images revealed diffuse emission
185: and several point sources, M82 X-1 being the brightest among them
186: (Matsumoto et al. 2001). The diffuse emission is evident in the EPIC
187: images but the point sources are not resolved. Following SM03, we
188: extracted PN and MOS events using a $18\arcsec$ circular region around
189: the bright point source whose position is consistent with the
190: \chandra{} position of M82 X-1. Above $2\kev$, the diffuse emission
191: contributes only $< 10\%$ to the point source inside the $18\arcsec$
192: circular region. At the faintest flux level of M82 X-1, nearby point
193: sources contribute $\la 30\%$ to the $0.5-10\kev$ X-ray emission in
194: the $18\arcsec$ region (Matsumoto et al. 2001; SM03).
195: %\section{Analysis \& Results}
196: \subsection{The power spectrum}
197: For temporal analysis, we combined the PN and MOS data and used the
198: continuous exposure of $70.2\ks$ during which both the PN and MOS
199: cameras operated simultaneously. We calculated a power density
200: spectrum (PDS) using the background corrected PN+MOS light curves
201: sampled at $0.5\s$. Figure~\ref{f1} ({\it left}) shows the $2-10\kev$
202: PDS of M82 X-1 rebinned by a factor of $1024$ yielding a frequency
203: resolution of $7.8\mhz$. The PDS continuum is approximately flat at
204: low frequencies below $\sim 30\mhz$ and then falls off approximately
205: following a power law up to $\sim 200\mhz$ where the white noise
206: arising from the poisson errors starts to dominate the PDS. There is
207: a prominent QPO with its peak frequency near $114\mhz$.
208:
209: We fitted the PDS with two models: ($i$) a broad Lorentzian for the
210: continuum and a narrow Lorentzian for the QPO, and ($ii$) a broken
211: power law (BPL) for the continuum and a Lorentzian (L) for the QPO.
212: We also used a constant to account for the poisson noise. Both models
213: provided statistically acceptable fits, giving a minimum $\chi^2$ of
214: $128.6$ for 121 degrees of freedom (dof) and $127.9$ for 120 dof for the
215: double Lorentzian and BPL$+$L models, respectively. The errors on the
216: best-fit PDS model parameters, quoted below, are at a $1\sigma$ level.
217: The double Lorentzian model resulted in a QPO centroid frequency
218: $\nu_{QPO} = 114.6\pm1.5\mhz$, a width $\nu_{FWHM} =
219: 31.3_{-2.5}^{+2.9}\mhz$, an amplitude $A_{QPO}=0.037.4\pm0.0024$ and
220: the broad Lorentzian centroid $\nu_{0} = 11.4_{-4.4}^{+4.2}\mhz$, a
221: width $\nu_{FWHM} = 60.2_{-6.3}^{+7.5}\mhz$ and an amplitude $A =
222: 0.038\pm0.0034$. The BPL$+$L model resulted in a QPO centroid frequency
223: $\nu_{QPO} = 114.3\pm1.5\mhz$, a width $\nu_{FWHM} =
224: 32.7_{-2.6}^{+2.9}\mhz$, an amplitude $A_{QPO}=0.038\pm0.0024$, and
225: broken power-law indices of $\Gamma_1 = 0.11_{-0.09}^{+0.10}$
226: below the break frequency $\nu_b = 34.2_{-2.9}^{+5.7}\mhz$ and
227: $\Gamma_2 = 2.32_{-0.39}^{+0.58}$ above the break frequency, and an
228: amplitude $A_{BPL} = 0.36\pm0.03$ at a reference frequency of
229: $10\mhz$. The total integrated power ($0.001-1\hz$) and the QPO power
230: expressed as $rms/mean$ are $23\%$ and $19\%$, respectively. The use
231: of a broken power law is an improvement ($\Delta \chi^2 = -24.8$ for
232: two additional parameters) over a simple power law at a statistical
233: significance level of $>99.99\%$ based on the maximum likelihood ratio
234: test. Thus the break in the continuum of the power spectrum is real.
235: The best-fit BPL$+$L model is shown in Fig.~\ref{f1}({\it left}) as a
236: thick line.
237:
238:
239: For comparison we have also calculated the PDS of an XRB
240: XTE~J1550-564 using the \rxte{} observation of 10 September 1998.
241: XTE~J1550-564 shows a large range in its low frequency QPO centroid
242: ($\nu_{QPO} \sim 0.08 - 18\hz$; Sobczak et al. 2000a). Our choice of
243: the particular \rxte{} observation relies on the fact that
244: XTE~J1550-564 showed a power-law (PL) photon index of $\Gamma \sim 2.0$ on
245: 10 September 1999 (Sobczak et al. 2000b), which is very similar to the
246: $3-10\kev$ photon index of M82 X-1 (see below). Fig.~\ref{f1}({\it
247: right}) shows the PDS of XTE~J1550-564 calculated from the PCA light
248: curve sampled at $0.125\s$ which was obtained from the \rxte{} public
249: data archive. We used a model comprising a broken PL and
250: two Lorentzians to fit the PDS. This model resulted in a statistically
251: acceptable fit (minimum $\chi^2 = 94.2$ for 118 dof). The broken PL
252: has best-fit parameters: $\Gamma_1 = 0.03\pm0.006$ below the
253: break frequency $\nu_b = 0.27\pm0.007\hz$, $\Gamma_2 = 0.89\pm0.01$
254: above the break and an amplitude $A_{BPL} = 0.026\pm0.0005$ at a
255: reference frequency of $10\mhz$. The two QPOs have centroid
256: frequencies $\nu_{QPO} = 1.034\pm0.004$ and $2.055\pm0.016\hz$, widths
257: $\nu_{FWHM} = 0.125\pm0.010$ and $0.39\pm0.05\hz$ and amplitudes
258: $A_{QPO} = 0.026\pm0.0005$ and $0.0046\pm0.0003$ for the fundamental and first
259: harmonic, respectively. The total integrated power ($0.01-4\hz$) and the QPO (fundamental) power expressed as $rms/mean$ are $23\%$ and $16\%$, respectively.
260:
261: \subsection{The energy spectrum}
262: We extracted PN and MOS spectra using a $18\arcsec$ circular region
263: centered at the position of M82 X-1. We also extracted PN and MOS
264: background spectra using nearby circular regions free of sources. We
265: created appropriate response files using the SAS tasks {\tt rmfgen}
266: and {\tt arfgen}. The spectra were grouped to a minimum of $20$
267: counts per spectral channel and analyzed with {\tt XSPEC 11.3}. The
268: errors on the best-fit spectral parameters are quoted at a $90\%$
269: confidence level. An absorbed power law (PL) model fitted to the PN
270: data showed strong soft excess emission below $\sim 3\kev$ due to the
271: presence of the diffuse soft X-ray emission in the source extraction
272: region. We ignored the data below $3\kev$ and performed the spectral
273: fitting of the PN and MOS data separately. A simple PL model
274: poorly describes both the PN and MOS data. We show the ratio of the PN
275: data and the best-fitting absorbed PL model in
276: Figure~\ref{f2}(a). A prominent iron K$\alpha$ line at $\sim 6.6\kev$
277: is evident. Addition of a Gaussian line (GL) to the PL model
278: improved the fit significantly ($\Delta \chi^2 = -71.7$ for three
279: additional parameters) and resulted in a good fit (minimum $\chi^2 =
280: 898.9$ for 944 dof. The Gaussian line has a centroid energy $E_{line} =
281: 6.61_{-0.08}^{+0.07}\kev$, a width $\sigma = 278_{-115}^{+161}\ev$, a
282: line flux $f_{line} = 1.9_{-0.6}^{+0.7} \times
283: 10^{-5}{\rm~photons~cm^{-2}~s^{-1}}$ and an equivalent width of $\sim
284: 180\ev$. The best-fit photon index is $2.00_{-0.07}^{+0.07}$ and the
285: the absorption column is $\nh = 2.5_{-0.6}^{+0.6}\times
286: 10^{22}{\rm~cm^{-2}}$. The observed $2-10\kev$ flux is
287: $9.6\times10^{-12}{\rm~erg~cm^{-2}~s^{-1}}$ and the corresponding
288: luminosity is $1.7\times 10^{40}{\rm~erg~s^{-1}}$ assuming a distance
289: of $3.9\mpc$ (Sakai \& Madore 1999). In Figure~\ref{f2}b, we show the
290: best-fit PL$+$GL model to the EPIC PN data. The PL$+$GL model fitted
291: to the MOS data resulted in slightly flatter PL ($\Gamma_X =
292: 1.80_{-0.09}^{+0.11}$), an absorption column
293: $\nh=1.8_{-0.3}^{+0.7}\times 10^{22}{\rm~cm^{-2}}$ and a weaker iron
294: line, $EW \sim 94\ev$.
295:
296:
297: \section{Discussion}
298: Based on long \xmm{} observations, we show clear evidence for the
299: simultaneous presence of a low frequency break at $\nu_b
300: =33.5_{-3.0}^{+6.7}\mhz$ and a QPO with centroid frequency $\nu_{QPO}
301: = 113.2\pm1.5\mhz$ in the power spectrum of M82 X-1. The QPO
302: frequency is similar to that found from the \rxte{} observation of
303: 1997 July 21, while it is about a factor of two larger than that found
304: in the first \xmm{} observation of 2001 May and an \rxte{} observation
305: of 1997 February 24 (SM03). Thus the QPO frequency of M82 X-1 is
306: variable.
307:
308: The M82 X-1 PDS is flat below the break frequency and then falls off above the break
309: as a power law with a prominent QPO on top of it. This form of the PDS
310: is characteristic of XRBs in their high/soft state
311: or intermediate state (e.g., McClintock \& Remillard 2003;
312: see also Fig.~\ref{f1}
313: for a comparison). PDSs of
314: many BH XRBs display low frequency QPOs in the frequency range
315: of $\sim 0.05 - 30\hz$ (Remillard et al. 2002). The QPO frequency is
316: strongly correlated with the break frequency (Wijnands \& van der Klis
317: 1999). The break frequency is always lower than the QPO frequency by
318: at least a factor of three (see, Wijnands \& van der Klis 1999). M82
319: X-1 follows this correlation as its QPO frequency is about a
320: factor three larger than the break frequency. {\it Thus the PDS of the ULX is
321: very similar to the XRB PDSs in their high/soft or transition
322: state.}
323:
324: Since the characteristic time scales of accretion powered sources scale
325: with BH mass, the similarity of the PDSs of M82 X-1 and XRBs provides
326: the opportunity to determine the ratio of their BH masses. However, as
327: stated above, the XRBs show a large range in their low
328: frequency QPO and break frequencies. Therefore, the frequency scaling
329: factor between any two power spectra of two XRBs will lead to
330: an incorrect determination of the ratio of their BH masses. Fortunately,
331: the PDS features (break and QPO frequencies) are well correlated with
332: the energy spectral parameters.
333: The low frequency QPOs are
334: related to the flux of the PL energy spectral component. The
335: variable frequency QPOs in XTE J1550-564 and GRO J1655-40 appear only
336: when the PL contributes more than $20\%$ of the $2-10\kev$ flux
337: (Sobczak et al. 2000a and references therein). GRS~1915+105 also shows
338: similar behavior (Muno et al. 1999). The PL index is well
339: correlated with the centroid of the low frequency QPOs in GRS~1915+105
340: (Vignarca et al. 2003) and XTE~J1550+564 (Sobczak et al. 2000a,
341: 2000b). The photon index increases with the QPO frequency until it
342: saturates. The frequency scaling factor between the low frequency QPOs
343: of two XRBs with similar PL photon indices can be used to
344: determine the ratio of their BH masses (TF04; FT04). Thus it is
345: crucial to determine the shape of X-ray spectrum and detect both the
346: QPO and the low frequency break in order to make sure that the
347: detected QPO lies at a low frequency, but above the break frequency,
348: and on top of a PDS power-law continuum.
349:
350:
351: The long \xmm{} observation of M82 X-1 has yielded all 3 pieces of
352: information required to determine its BH mass. The photon index of the
353: $3-12\kev$ PL is $2.0\pm0.1$ as determined from the high
354: signal-to-noise PN data. XRBs show a range of PL
355: photon indices from $\Gamma \sim 1.6$ in their low/hard state to
356: $\Gamma \sim 2.5$ in their high/soft state. Thus M82 X-1 is likely in
357: a state intermediate between the high and low states. Sobczak et al.
358: (2000a) studied the spectral behavior of XTE~J1550-564 during its
359: 1998--1999 outburst using 209 pointed observations with \rxte{}. They
360: used a model consisting of a multicolor disk blackbody and a PL
361: to fit the spectra and found that the spectra gradually steepen from
362: $\Gamma \sim 1.5$ to $\sim 2.9$ for XTE~J1550-564. The low frequency
363: QPO is also found to vary from $0.08 - 13\hz$. At a PL photon
364: index of $1.91$ ($2.12$), XTE~J1550-564 shows a low frequency QPO at
365: $0.81\hz$ ($1.6\hz$) (Sobczak et al. 2000a; 2000b). Noting that the
366: BH mass scales inversely in proportion to the QPO frequency and the BH
367: mass of XTE~J1550-564 is about $10M_\odot$ (Titarchuk \& Shrader 2002,
368: Shrader \& Titarchuk 2003), we can determine the BH mass of the M82 X-1
369: as
370: $M_{BH}$ (ULX) $\sim \nu_{QPO}$(XTE~J1550-564)/$\nu_{QPO}$(ULX)
371: $\times M_{BH}$ (XTE~J1550-564) $\sim 70 - 150 M\odot$. Another
372: XRB GRS~1915+105 shows a low frequency QPO with its centroid
373: at $0.642\pm0.004\hz$ ($1.609\pm0.007\hz$) when its PL photon
374: index is $1.88\pm0.04$ ($2.16\pm0.04$) (Vignarca et al. 2003). Noting
375: that that the BH mass of GRS~1915+105 is $14\pm4 M\odot$ estimated
376: dynamically from the IR spectroscopic observation of its companion
377: star (Greiner et al. 2001), the BH mass for M82 X-1 is $M_{BH}$ (ULX)
378: $\sim 50 - 260M\odot$. Thus, scaling the mass of M~82 X-1 from the
379: $\Gamma - \nu_{QPO}$ relation calibrated by the dynamical masses of
380: several galactic BHs, we find $M_{BH}$ (ULX) $\sim 50 - 260M\odot$.
381:
382: The QPO frequencies and photon indices for M82
383: X-1, XTE~J1550-564 and GRS~1915+105 correspond to a region of steep
384: dependence in the $\Gamma - \nu_{QPO}$ relation (see FT04). It is
385: therefore likely that the uncertainty in the calibration of the $\Gamma -
386: \nu_{QPO}$ relation may introduce an additional factor of two
387: uncertainty in the determination of the BH mass of M~82~X-1. Thus the
388: true BH mass of M~82 X-1 is most likely in the range of
389: $25-520M\odot$. This is at least a factor of $2$ lower than the
390: previous estimate by FT04, which can be explained by
391: the lower values of the QPO frequencies for M82 X-1 in our paper compared with
392: those in FT04. The above estimate of the BH mass is based on the
393: assumption that M~82 X-1 also follows the $\Gamma - \nu_{QPO}$
394: relation found for XRBs. \xmm{} and \rxte{}
395: observations of M~82 X-1 have revealed that the QPO frequency is
396: variable (SM03). The photon index is also variable (FT04) but appears
397: to be inconsistent with the $\Gamma - \nu_{QPO}$ relation for XRBs.
398: However, there may be systematic errors in the photon indices measured
399: with \xmm{} and \rxte{} due to contamination from nearby sources,
400: as indicated by large apparent changes in the effective absorption
401: column. \chandra{} ACIS observations of M~82 X-1 were used to suggest
402: an effective
403: column in the range $(5-9)\times10^{20}{\rm~cm^{-2}}$ (SM03), while
404: \xmm{} and \rxte{} observations have been used to suggest large columns
405: in the range
406: $2\times10^{21} - 8\times10^{22}{\rm~cm^{-2}}$ (see Section 3.2,
407: FT04). Both \xmm{} and \rxte{} observations suffer from contamination
408: by nearby sources. A small contribution from nearby sources may
409: significantly alter the spectral shape of the ULX and a varying
410: contribution may lead to the observed changes in the absorption
411: column. Thus, based on the currently available X-ray data on M~82 X-1,
412: it is neither possible to prove nor rule out the $\Gamma - \nu_{QPO}$
413: relation similar to that found for XRBs.
414:
415: M82 X-1 is one of the brightest ULXs. During the second \xmm{}
416: observation, its $2-10\kev$ luminosity was $1.7\times
417: 10^{40}{\rm~ergs~s^{-1}}$ (SM03), while it was a factor of $\sim 2$
418: brighter in the first \xmm{} observation. Based on the first \xmm{}
419: observation, SM03 quoted a bolometric luminosity of $(4-5)\times
420: 10^{40}{\rm~ergs~s^{-1}}$, equivalent to the Eddington luminosity for
421: a BH mass in the range of $300-400M\odot$. Given the uncertainty in
422: the contamination by nearby X-ray sources, our estimated BH mass of M82~X-1
423: is marginally consistent with the observed luminosity and the Eddington ratio.
424:
425:
426: \acknowledgements We thank an anonymous referee for useful comments, Aspen center for Physics
427: for organizing the workshop
428: `Revealing black holes' and A. R. Rao for his comments.
429: GCD acknowledges the support of NASA grants through the
430: awards NNG04GN69G and NNG05GN35G.
431: This work is based on
432: observations obtained with \xmm{}, an ESA science mission with
433: instruments and contributions directly funded by ESA Member States and
434: the USA (NASA).
435: %This research has made use of data obtained through
436: %the High Energy Astrophysics Science Archive Research Center Online
437: %Service, provided by the NASA/Goddard Space Flight Center.
438:
439: \begin{thebibliography}{}
440: \bibitem[Colbert \& Mushotzky(1999)]{1999ApJ...519...89C} Colbert,
441: E.~J.~M.~\& Mushotzky, R.~F.\ 1999, \apj, 519, 89 (CM99)
442: \bibitem[Fabbiano(1989)]{1989ARA&A..27...87F} Fabbiano, G.\ 1989,
443: \araa, 27, 87
444: \bibitem[Fiorito \& Titarchuk(2004)]{2004ApJ...614L.113F} Fiorito, R.,
445: \& Titarchuk, L.\ 2004, \apjl, 614, L113 (FT04)
446: \bibitem[Greiner et al.(2001)]{2001Natur.414..522G} Greiner, J., Cuby,
447: J.~G., \& McCaughrean, M.~J.\ 2001, \nat, 414, 522
448: \bibitem[Liu et al.(2004)]{2004ApJ...602..249L} Liu, J.-F., Bregman, J.~N.,
449: \& Seitzer, P.\ 2004, \apj, 602, 249
450: \bibitem[Kaaret et al.(2001)]{2001MNRAS.321L..29K} Kaaret, P., et al., \ 2001, \mnras, 321, L29
451: \bibitem[Kaaret et al.(2004)]{2004MNRAS.348L..28K} Kaaret, P.,
452: Alonso-Herrero, A., Gallagher, J.~S., Fabbiano, G., Zezas, A., \& Rieke,
453: M.~J.\ 2004, \mnras, 348, L28
454: \bibitem[Kuntz et al.(2005)]{2005ApJ...620L..31K} Kuntz, K.~D., et al.\ 2005, \apjl, 620, L31
455: \bibitem[Markwardt et al.(1999)]{1999ApJ...513L..37M} Markwardt,
456: C.~B., Swank, J.~H., \& Taam, R.~E.\ 1999, \apjl, 513, L37
457: \bibitem[Matsumoto et al.(2001)]{2001ApJ...547L..25M} Matsumoto, H.,
458: et al., 2001, \apjl, 547, L25
459: \bibitem[McClintock \& Remillard(2003)]{2003astro.ph..6213M} McClintock,
460: J.~E., \& Remillard, R.~A.\ 2003, ArXiv Astrophysics e-prints,
461: arXiv:astro-ph/0306213
462: \bibitem[Miller \& Colbert(2003)]{2003astro.ph..8402M} Miller,
463: M.~C.~\& Colbert, E.~J.~M 2003, ArXiv Astrophysics e-prints, 8402
464: \bibitem[Remillard et al.(2002)]{2002miqu.work...49R} Remillard, R.,
465: Muno, M., McClintock, J.~E., \& Orosz, J.\ 2002, New Views on
466: Microquasars, 49
467: \bibitem[Sakai \& Madore(1999)]{1999ApJ...526..599S} Sakai, S., \&
468: Madore, B.~F.\ 1999, \apj, 526, 599
469: \bibitem[Shrader \& Titarchuk (2003)]{2003ApJ...598..168T} Shrader, C.R. \& Titarchuk, L. 2003, \apj, 598, 168
470: \bibitem[Shrader \& Titarchuk (1999)]{1999ApJ...521..L121} Shrader, C.R. \& Titarchuk, L. 1999, \apj, 521, L121
471: \bibitem[Sobczak et al.(2000a)]{2000ApJ...531..537S} Sobczak, G.~J.,
472: et al., \ 2000a, \apj, 531, 537 %correlation
473: \bibitem[Sobczak et al.(2000b)]{2000ApJ...544..993S} Sobczak, G.~J., et al., \ 2000b, \apj, 544, 993
474: \bibitem[Soria et al.(2004)]{2004A&A...423..955S} Soria, R., Motch, C.,
475: Read, A.~M., \& Stevens, I.~R.\ 2004, \aap, 423, 955
476: \bibitem[Strohmayer \& Mushotzky(2003)]{2003ApJ...586L..61S}
477: Strohmayer, T.~E., \& Mushotzky, R.~F.\ 2003, \apjl, 586, L61 (SM03)
478: \bibitem[Titarchuk \& Fiorito(2004)]{2004ApJ...612..988T} Titarchuk,
479: L., \& Fiorito, R.\ 2004, \apj, 612, 988 (TF04)
480: \bibitem[Vignarca et al.(2003)]{2003A&A...397..729V} Vignarca, F.,
481: Migliari, S., Belloni, T., Psaltis, D., \& van der Klis, M.\ 2003,
482: \aap, 397, 729
483:
484: \end{thebibliography}
485:
486: % \clearpage
487: \begin{figure}
488: \centering \includegraphics[width=8cm]{f1a.ps}
489: \includegraphics[width=8cm]{f1b.ps}
490: \caption{PDS of the ULX M82 X-1 and an XRB
491: XTE~J1550-564 having very similar PL photon indices
492: ($\Gamma \sim 2.0$). {\it Left:} PDS of M82 X-1 derived
493: from the EPIC PN and MOS data above $2\kev$. The white noise
494: level expected ($\sim 1.4$) from the poisson errors has not been subtracted.
495: The frequency resolution is $7.8{\rm~mHz}$. The best-fitting model
496: comprising a broken power-law and a Lorentzian is shown as a
497: thick line. {\it Right:} PDS of the XRB
498: XTE~J1550-564 derived from the \rxte{} observations of 1998
499: September 10. The best-fit model consisting of a broken power law
500: and two Lorentzians is shown as a thick line. }
501: \label{f1}
502: \end{figure}
503:
504: \begin{figure}
505: \centering \includegraphics[angle=-90]{f2a.ps}
506: \includegraphics[angle=-90]{f2b.ps}
507: \caption{(a) Ratio of EPIC PN data and the best-fit absorbed
508: PL model showing the iron K$\alpha$ line. (b) The PN
509: spectrum and the best-fit absorbed PL$+$Gaussian line
510: model ({\it upper panel}) and their ratio ({\it lower panel}).}
511: \label{f2}
512: \end{figure}
513:
514:
515: \end{document}
516:
517:
518: