1: \documentclass[12pt,preprint]{aastex}
2:
3: %\usepackage{emulateapj5}
4:
5: \begin{document}
6: \slugcomment{Submitted to ApJ Letter}
7: \title{The optical counterpart of a Ultraluminous X-Ray Object in M81}
8:
9: \author{Ji-Feng Liu, Joel N. Bregman and Patrick Seitzer}
10: \affil{Astronomy Department, University of Michigan, MI 48109}
11:
12: \begin{abstract}
13:
14: Ultraluminous X-Ray Objects are off-nucleus point sources with
15: $L_X=10^{39}$-$10^{41}$ erg/s and are among the most poorly understood
16: X-ray sources. To help understand their nature, we are trying to identify
17: their optical counterparts by combining images from the Hubble Space Telescope
18: and the Chandra Observatory. Here we report an optical counterpart for ULX NGC
19: 3031 X-11, which has average X-ray luminosity of $\sim2\times10^{39}$ erg/s and
20: has varied by a factor of 40\% over the last 20 years. We find a unique optical
21: counterpart with the magnitude and color of an O8V star and we identify this as
22: the secondary in a binary system. The primary is believed to be a black hole
23: of approximately 18 $M_\odot$, based on analyses of ACIS and ASCA spectra.
24: This binary system probably is powered by mass transfer from the O8V star onto
25: the black hole, with the Roche Lobe equal to the stellar radius. This model
26: predicts an orbital period of $\sim1.8$ days, which can be tested by future
27: observations.
28:
29: \end{abstract}
30:
31: \keywords{Galaxy: individual(M81) --- X-rays: binaries}
32:
33: \section{INTRODUCTION}
34:
35:
36: X-ray emission from the nuclei of AGN and quasars, powered by accretion onto
37: nuclear supermassive black holes, are usually $>10^{41}$ erg/s, while for the
38: stellar binary systems in the Milky Way, the luminosities are typically
39: $10^{33}$-$10^{38}$ erg/s. X-ray objects with luminosities of
40: $10^{39}$-$10^{41}$ erg/s have been observed in some external galaxies as
41: off-nucleus point sources (Fabbiano 1989; Marston et al. 1995; Zezas et al.
42: 2002). These objects have several names, the most popular designation being
43: ``Ultraluminous X-ray Objects'', or ULXs.
44:
45:
46: One suggestion for the nature of these ULXs is that they are binary systems
47: with $10^3$-$10^4$ $M_{\odot}$ black holes as primaries ( Colbert and Mushotzky
48: 1999). This suggestion is consistent with the X-ray spectra analyses of some
49: ULXs (Makishima et al. 2000). However, the formation of such massive black
50: holes is not predicted by stellar evolution theory and it may be impossible to
51: form such objects even in dense star clusters. Alternatively, these sources
52: may be stellar mass black holes or neutron stars whose emission is beamed,
53: thereby representing micro-quasars (King et al. 2001). If the emission is
54: beamed, the intrinsic luminosities become sub-Eddington, and there are known
55: examples of beamed Galactic X-ray sources. It is also possible that the
56: luminosities are truly super-Eddington, for example, obtainable from accretion
57: disks with radiation-driven inhomogeneities (Begelman, 2002).
58:
59:
60: An essential step in uncovering the nature of the ULXs is to identify them at
61: other wavelengths, such as optical. Associations of ULXs to emission-line
62: nebulae based on ROSAT positions have already been done for a number of ULXs
63: (e.g., Pakull \& Mirioni, 2002). However, identifications down to single point
64: sources were impossible given the large positional errors of EINSTEIN and ROSAT
65: observations, with the smallest error circles had radii of
66: $\sim5^{\prime\prime}$ (for ROSAT HRI). With the advent of Chandra X-ray
67: satellite, this task becomes possible owing to the sub-arcsecond spatial
68: resolution of Chandra observations. For example, by comparing Chandra ACIS-S
69: and HST/WFPC2 observations of the ''Antennae'' starburst galaxies, Zezas et al.
70: (2002) found that 10 out of 14 ULXs are associated with stellar clusters, with
71: offsets between the X-ray sources and the clusters less than
72: $2^{\prime\prime}$. Wu et al. (2002) found that the ULX in edge-on spiral
73: galaxy NGC4565 is associated with a stellar cluster on the outskirt of the
74: bulge. Goad et al. (2002) found that within the $0\farcs6$ positional error
75: circle of the ULX in NGC5204, there are three blue objects which they argue are
76: very young stellar clusters.
77:
78: As part of our search for optical counterparts of ULXs in nearby galaxies (Liu
79: et al. 2001), we report the results for a ULX in NGC 3031, NGC 3031 X-11 as
80: designated in Roberts \& Warwick (2000). This ULX has been observed more than
81: 20 times by EINSTEIN, ROSAT and Chandra X-ray satellites. Over the 20 year
82: baseline, its X-ray luminosity varied by more than 40\%, with an average of
83: $\sim2\times10^{39}$ erg/s. Matonick \& Fesen (1997) found that this ULX is
84: coincident with their SNR No. 22 and a discrete radio source in a VLA 20cm map,
85: and suggested that it may be a SNR, despite its long term variability and high
86: luminosity. In $\S$2, we present the observations utilized, the data analysis
87: procedures, and the results. We discuss the implications of its optical
88: identification as an O8V star in $\S$3. For the distance to M81, we use 3.63
89: Mpc ($\mu=27.8\pm0.2$, Freedman et al. 1994). At such a distance, 1 pixel in
90: the Planetary Camera chip, i.e., $0\farcs05$, corresponds to a physical scale
91: of 0.9 parsecs.
92:
93:
94: \section{OBSERVATIONS AND DATA ANALYSIS}
95:
96: NGC 3031 X-11 was observed by HST/WFPC2 in June, 2001 for our program. Three
97: filters, F450W(B), F555W(V), and F814W(I) were used; for each filter, four
98: 500-second exposures were made. We used HSTphot (Dolphin 2000) to perform PSF
99: fitting photometry. To transform between WFPC2 image positions (X,Y) and
100: celestial coordinates (RA,DEC), we used IRAF tasks {\tt metric/invmetric},
101: which correct the geometric distortion introduced by the camera optics. The
102: final relative positions are accurate to better than $0\farcs005$ for targets
103: contained on one chip, and $0\farcs1$ for targets on different chips. The
104: absolute accuracy of the positions obtained from WFPC2 images is typically
105: $0\farcs5$ R.M.S. in each coordinate (HST data handbook).
106:
107: A 50 kilosecond Chandra ACIS observation of NGC 3031 (ObsID 735, observed date:
108: 2000-05-07) was retrieved from the Chandra archive. {\tt CIAO 2.2.1}, {\tt
109: CALDB 2.12} and {\tt xspec 11.2.0} were used to analyze the ACIS data. Pixel
110: randomization of event positions was removed to recover the original positions.
111: {\tt Wavdetect} was run to detect discrete sources on ACIS-S chips. The
112: accuracy of absolute positions for these sources has a typical R.M.S of
113: $0\farcs6$ (Aldcroft et al. 2000).
114:
115: Two bright X-ray sources fall into the WFPC2 field of view, i.e., the ULX on
116: PC1 and SN 1993J on WF4, with a seperation of $\sim2^\prime$. Using the
117: relative position of these two sources, we can register the ULX onto the
118: Planetary Chip very accurately. We list in Table 1 the ACIS and WFPC2
119: positions of SN 1993J, the ACIS position for the ULX, and its derived WFPC2
120: position. The centroiding error for ACIS-S3 positions computed by {\tt
121: Wavdetect} as quoted in Table 1 is negligible. The relative positional error
122: comes mainly from the ACIS plate scale variation\footnote{see
123: http://cxc.harvard.edu/cal/Hrma/hrma/optaxis/platescale/geom\_public.html},
124: which produces an error of $0\farcs1$ for a $2^\prime$ seperation, and the
125: WFPC2 interchip error of $0\farcs1$. The rotation R.M.S. for both WFPC2 and
126: ACIS-S chips are $\sim0\fdg01$, and the resulted positional error is about
127: $0\farcs02$. The final positional error is $\sim0\farcs15$, i.e., about 3
128: Planetary Chip pixels. Within this error circle, there is only one object,
129: which is indicated by a cross in Figure 1.
130:
131: A cluster of typical size 2--3 parsecs, or a supernova remnant, can be readily
132: resolved in the PC image. Careful examination shows that this unique
133: counterpart of NGC 3031 X-11 is not extended, but a point source, with standard
134: magnitudes $B=23.87\pm0.04$, $V=23.89\pm0.03$, and $I=23.76\pm0.07$ as obtained
135: with HSTphot. The counterpart is located on a dusty spiral arm and in a region
136: of high stellar density, as can be seen in Figure 1. The Galactic extinction
137: toward NGC 3031 are $A_B=0.35$, $A_V=0.27$, and $A_I=0.16$, with a HI column
138: density\footnote{http://heasarc.gsfc.nasa.gov/cgi-bin/Tools/w3nh/w3nh.pl} of
139: $4.1\times10^{20}$ cm$^{-2}$ (Schlegel et al. 1998). When this extinction is
140: removed, the counterpart has $M_B=-4.28$, $M_V=-4.18$, $M_I=-4.20$,
141: $\mbox{(B-V)}_0=-0.10$, and $\mbox{(V-I)}_0=0.02$. While its absolute
142: magnitude is consistent with a O9V star (19 $M_\odot$, 7.8 $R_\odot$;
143: Schmidt-Kaler 1982), its color is much redder, which may indicate intrinsic
144: extinction by its dusty environments. If we assume an intrinsic
145: $\mbox{E(B-V)}=0.22$ and a Galactic-like $R_V=3.1$, the counterpart will have
146: $M_V=-4.9$ and $\mbox{B-V}=-0.32$, matching the color and absolute magnitude of
147: an O8V star (23 $M_\odot$, 8.5 $R_\odot$), though the I band is dimmer by 0.1
148: magnitude. We note that the total color excess $\mbox{E(B-V)}=0.3$ is
149: consistent with the absorbing column density $2.1\times10^{21}$ cm$^{-2}$ derived
150: from X-ray spectra according to a Galactic empirical relation
151: $N_H=5.8\times10^{21}*E(B-V)$ (Bohlin et al. 1979). This star is not a
152: supergiant, since a supergiant is much redder and much brighter, for example, a
153: B2I supergiant has $B-V=-0.17$ and $M_V=-6.4$.
154:
155: \section{DISCUSSION}
156:
157: The above determination of the mass and radius of the optical counterpart is
158: based upon comparison with Galactic stars whose properties are known.
159: Alternatively, one can try to obtain these quantities theoretically by
160: comparing its observational properties to isochrones (Z=0.020) from the Geneva
161: stellar models (Lejeune et al. 2001), such as in Figure 2, where we also show
162: other bright stars in the field. The ages of the bright field stars are from
163: about $10^6$ to $10^8$ years, with the anticipated implication that star
164: formation in the spiral arm is not co-eval. The location of the counterpart in
165: the color-magnitude diagram is consistent with it being a star younger than
166: $10^{6.5}$ years. If it has spent $10^{6.5}$ years on the main sequence, its
167: absolute magnitude corresponds to that of a star with a mass of 27 $M_\odot$
168: and a radius of 10.0 $R_\odot$. A limiting case is that it is a star that has
169: just reached the main sequence, which makes it hotter and more massive, with a
170: mass of 41 $M_\odot$ and a radius of 8.7 $R_\odot$. The stellar metalicity
171: may differ from solar, which will lead to small changes in the inferred mass
172: and radius, so we consider two extreme cases of low and high metallicity. For
173: a star of metalicity Z=0.004, the age would be less than $10^{6.7}$ years from
174: its location on the Main Sequence, and its mass ranges from 50 to 28 $M_\odot$
175: with a radius between 8.3 to 10.3 $R_\odot$ with increasing age. For a star of
176: metalicity Z=0.040, it would be younger than $10^{6.4}$ years, its mass lies
177: between 36 and 26 $M_\odot$ with a radius between 8.8 and 9.8 $R_\odot$.
178: Taking into account the range of possibilities, the inferred mass is between 26
179: and 50 $M_\odot$, and the radius is between 8.3 and 10.3 $R_\odot$. This result
180: can be compared to the mass and radius inferred from its spectral type, i.e.,
181: 23 $M_\odot$ and 8.5 $R_\odot$ for an O8V star. We conclude that the radius is
182: about $9.3\pm1.0$ $R_\odot$, while the mass is somewhat uncertain, ranging from
183: 23 to 50 $M_\odot$. The lower end of this mass range is preferred because it
184: is unlikely that the star has just reached the main sequence since its binary
185: companion has already evolved into a compact object. This optical counterpart
186: is presumably a secondary in a binary system, and is contributing material onto
187: an accreting compact primary.
188:
189: Swartz et al. (2002) analyzed the radial profile of this ULX in their Chandra
190: ACIS-S3 observation, and found it consistent with a point source; it is
191: consistent with a point source in our WFPC2 observations. They also analyzed
192: the spectrum and found, after deconvolving the effects of pileup, the spectrum
193: is best described by a model for an absorbed blackbody disk. The best-fit
194: absorbing column density is $N_{20}=21.7\pm1.0$ in units of $10^{20}$ cm$^{-2}$;
195: the best-fit innermost disk temperature is $T_{in}=1.03\pm0.11$ Kev, the
196: corresponding radius is $R_{in}=161\pm16$ km, which correspond to an 18
197: $M_\odot$ non-rotating black hole. From an analysis of ASCA spectra, Mizuno
198: (2000) found the following parameters, $N_{20}=21\pm3$,
199: $T_{in}=1.48\pm0.08$ Kev, $R_{in}=83\pm8$ km, which he argued correspond to a
200: Kerr black hole with similar mass. In Figure 3, we compared the fluxes of the
201: counterpart in B, V, and I bands with the predictions of the absorbed disk
202: blackbody model derived from Chandra ACIS and ASCA spectra, and found that the
203: observed optical fluxes are at least 10-30 times brighter than the disk itself,
204: indicating that most of the optical flux is from the X-ray inactive secondary
205: O8V star. Furthermore, the optical disk has a color of $\mbox{B-V}\approx0.1$,
206: which is significantly flatter (redder) than the color of the counterpart.
207: These two pieces of evidence lead us to conclude that the optical light is from
208: a star rather than from an accretion disk.
209:
210: The nature of the X-ray variability, the X-ray spectral properties, and the
211: optical colors indicate that NGC 3031 X-11 is not a supernova remnant (Matonick
212: \& Fesen 1997) but is a high mass X-ray binary (HMXB) system, consisting of an
213: $M_1=18M_\odot$ black hole and an $M_2=23M_\odot$ O8V companion. HMXB systems
214: in our Galaxy with black hole masses above about 10 $M_\odot$ have
215: $L_X<10^{39}$ erg/s, with very few exceptions. An example is Cygnus X-1, which
216: has a BH of 15 $M_\odot$ and a companion of O9.7Iab supergiant, with
217: $L_X<10^{38}$ erg/s (Tanaka \& Lewin, 1995). These modest luminosities can be
218: understood given the low accretion efficiency of stellar wind from the
219: companion supergiants. For this ULX system, since the companion is a main
220: sequence star, its stellar wind is not as strong as supergiants, and it may be
221: accreting gas via Roche Lobe Overflow.
222:
223: This scenario can be tested by future observations. Black hole masses derived
224: from spectral analyses are best fits to simple models, so changes to the
225: models, especially for the inner region, may lead to different masses. In
226: addition, there is the concern that stellar evolutionary models do not predict
227: $18 M_\odot$ black holes. For example, models for 60 $M_{\odot}$ main sequence
228: stars at the point of core collapse (Vanbeveren et al. 1998, and references
229: therein) predict a mass of $\sim 11 M_\odot$. These uncertainties in the mass
230: determination can be avoided by obtaining a direct measure of black hole mass
231: (function) through measurements of an orbital period and the associated orbital
232: velocities.
233:
234: We can predict the period of the binary based on the mass of the secondary and
235: the requirement that it is filling its Roche lobe. If the counterpart has the
236: radius and mass of an O8V star, i.e., a radius of $\sim8.5R_\odot$ and a mass
237: of 23 $M_{\odot}$, we infer that the seperation between the primary and
238: secondary is $a=(0.6q^{2/3}+ln(1+q^{1/3}))q^{-2/3}/0.49R_{cr}\approx21R_\odot$
239: for their mass ratio $q=M_2/M_1\approx1.3$ (Eggleton 1983). This in turn
240: implies an orbital period of $P_{orb} \approx1.8$ days according to Kepler's
241: 3rd law. For the mass range of 23-50 $M_\odot$ and the radius range of
242: $9.3\pm1.0$ $R_\odot$, the expected orbital period ranges from 1.1 days to 2.1
243: days. Although challenging, the measurement of the period would greatly
244: illuminate the nature of this, and presumably other ULX systems.
245:
246: King et al. (2001) raised the possibility that ULXs could be HMXB systems during
247: their short thermal timescale mass transfer via Roche Lobe overflow, and our
248: identification of ULX NGC 3031 X-11 with an O8V star supports this idea. Such
249: high mass systems have very short ages, consistent with the general findings
250: that ULXs tend to occur in star forming regions (e.g., Zezas et al. 2002).
251: However, ULXs also occur in some elliptical galaxies, e.g., NGC 1399 (Angelini
252: et al. 2001), and those may be intrinsically different from what we
253: discuss here.
254:
255:
256: \acknowledgements
257:
258: We are grateful for the service of Chandra Data Archive. We would like to thank
259: Jimmy Irwin, Renato Dupke, Eric Miller, Rick Rothschild, and Steven Murray for
260: helpful discussions. We gratefully acknowledge support for this work from NASA
261: under grants HST-GO-09073.
262:
263:
264: \begin{thebibliography}{}
265: \bibitem{} Aldcroft, T.L., Karovska, M., Cresitello-Dittmar, M.L., Cameron,
266: R.A., and Markevitch, M.L., 2000, Proc. SPIE, 4012, 650
267:
268: \bibitem{} Angelini, L., Loewenstein, M., and Mushotzky, R.F., 2001, \apj, 557,
269: L35
270:
271: \bibitem{} Begelman, M.C., 2002, \apj, 568, L97
272:
273: \bibitem{} Bohlin, R., avage, B.D., and Drake, J.F., 1978, \apj, 224, 132
274:
275: \bibitem{} Colbert, E.J.M. and Mushotzky, R.F., 1999, \apj, 519, 89
276:
277: \bibitem{} Dolphin, A.E., 2000, PASP, 112, 1383
278:
279: \bibitem{} Eggleton, P.P., 1983, \apj, 268, 368
280:
281: \bibitem{} Fabbiano, G. 1989, ARA\&A, 27, 87
282:
283: \bibitem{} Freedman, W.L., Hughes, S. M., Madore, B. F. et al., 1994, \apj,
284: 427, 628
285:
286: \bibitem{} Goad, M.R., Roberts, T.P., Knigge, C., and Lira, P. 2002,
287: astro-ph/0207428
288:
289: \bibitem{} King, A.R., Davies, M.B., Ward, M.J., Fabbiano, G., and Elvis, M.
290: 2001, \apj, 552, L109
291:
292: \bibitem{} Lejeune, T. and Schaerer, D., 2001, A\&A 366, 538
293:
294: \bibitem{} Liu, J., Bregman, J.N., and Seitzer, P., 2001, AAS, 199, 0503
295:
296: \bibitem{} Marston, A.P., Elmegreen, D., and Elmegreen, B. 1995, \apj, 438, 663
297:
298: \bibitem{} Makishima, K., Kubota, A., Muzuno, T., Ohnishi, T., Tashiro, M., et
299: al. 2000, \apj, 535, 632
300:
301: \bibitem{} Mizuno, T., 2000, Ph.D. Thesis, Univ. Tokyo
302:
303: \bibitem{} Matonick, D.M. and Fesen, R.A. 1997, ApJSS, 112, 49
304:
305: \bibitem{} Pakull, M.W. and Mirioni, L., 2002, astro-ph/0202488
306:
307: \bibitem{} Roberts, T.P. and Warwick, R.S., 2000, MNRAS, 315, 98
308:
309: \bibitem{} Schmidt-Kaler, Th., 1982, {\it Landolt-Bornstein: Numerical Data and
310: Functional Relationships in Science and Technology}, eds. Schaifers, K. and
311: Voigt, H.H., VI/2b, P.15-31
312:
313: \bibitem{} Schlegel, D.J., Finkbeiner, D.P., and Davis, M., 1998, \apj, 500, 525
314:
315: \bibitem{} Swartz, D.A., Ghosh, K.K., McCollough, M.L., Pannuti, T.G., Tennant,
316: A.F., and Wu, K., 2002, astro-ph/0206160
317:
318: \bibitem{} Tanaka, Y. and Lewin, W.H.G. 1995, in {\it X-ray Binaries}, eds.
319: Lewin, W.H.G, van Paradijs, J., and van den Heuvel, E.P.J. P.265
320:
321: \bibitem{} Vanbeveren, D., de Loore, C., and van Rensbergen 1998, A\&ARev, 9, 63
322:
323: \bibitem{} Wu, H., Xue, S.J., Xia, X.Y., Deng, Z.G., and Mao, S.D. 2002,
324: astro-ph/0204359
325:
326: \bibitem{} Zezas, A., Fabbiano, G., Rots, A.H., and Murray, S.S. 2002,
327: astro-ph/0203175
328:
329: \end{thebibliography}
330:
331: \begin{deluxetable}{lcccccccccc}
332: \tablecaption{The optical counterpart for ULX NGC 3031 X-11}
333: \tabletypesize{\tiny}
334: \tablehead{
335: \colhead{} & \multicolumn{4}{c}{ACIS-S3} &\colhead{}& \multicolumn{5}{c}{WFPC2} \\
336: \cline{2-5} \cline{7-11} \\
337: \colhead{Object} & \colhead{RA} & \colhead{DEC} & \colhead{RA\_ERR} &
338: \colhead{DEC\_ERR} & \colhead{}&\colhead{Chip} & \colhead{X} & \colhead{Y} &
339: \colhead{RA} & \colhead{DEC} \\
340: \colhead{} & \colhead{} & \colhead{} & \colhead{(arcsec)} &
341: \colhead{(arcsec)} &\colhead{}& \colhead{} & \colhead{(pix)} & \colhead{(pix)} &
342: \colhead{} & \colhead{}
343: }
344:
345: \startdata
346: SN 1993J & 9:55:24.79 & 69:01:13.44 & 0.03 & 0.01 && 4 & 480.72 & 240.89 & 9:55:25.11 & 69:01:14.87 \\
347: NGC 3031 X-11 & 9:55:32.95 & 69:00:33.36 & 0.03 & 0.01 && 1 & 437.57 & 408.32 & 9:55:33.27 & 69:00:34.79 \\
348:
349: \enddata
350: \end{deluxetable}
351:
352: \begin{figure}
353: \plotone{f1.ps}
354: \caption{The optical counterpart of NGC 3031 X-11 and its environments. The
355: counterpart, indicated by a cross, is a point source with standard magnitudes
356: $B=23.87\pm0.04$, $V=23.89\pm0.03$, and $I=23.76\pm0.07$. When the Galactic and
357: intrinsic extinction is removed, its absolute magnitude and colors correspond
358: to those of an O8V star. The size of the image is about
359: $22^{\prime\prime}\times22^{\prime\prime}$ and the distance between the inner
360: edges of two diagonal line segments of the cross is twice the diameter of the
361: positional error circle of the counterpart. }
362:
363: \end{figure}
364:
365: \begin{figure}
366: \plotone{f2.ps}
367: \caption{Color magnitude diagram for the ULX and nearby bright stars with
368: signal-to-noise ratio $\sigma>12$. Galactic extinction is removed from all the
369: stars and the counterpart also is corrected for intrinsic extinction, as
370: inferred from its X-ray absorption. The stellar isochrones are taken from
371: Lejeune et al. (2002) for Z=0.020 and the dashed line indicates the
372: de-reddening vector for HI absorption assuming $R_V=3.1$.}
373:
374: \end{figure}
375:
376: \begin{figure}
377: \plotone{f3.ps}
378: \caption{Spectral energy distribution of NGC 3031 X-11. The X-ray spectra from
379: ACIS can be fitted by an absorbed disk blackbody model, which we extend down to
380: optical to compare with the B, V, and I photometry of the counterpart. Both
381: the model and the BVI magnitudes are dereddened. The BVI fluxes of the
382: counterpart are 30-10 times larger than the disk itself.}
383:
384: \end{figure}
385:
386: \end{document}
387: