1: %\documentclass[12pt,manuscript]{aastex}
2: \documentclass[12pt,preprint]{aastex}
3:
4: %\usepackage{emulateapj5}
5:
6:
7: %\received{2002 October 22}
8: \begin{document}
9: %\slugcomment{Submitted to ApJ Letter}
10: \title{An Ultra-luminous X-Ray Object with a 2 hour period in M51}
11:
12: \author{Ji-Feng Liu, Joel N. Bregman, Jimmy Irwin, and Patrick Seitzer}
13: \affil{Astronomy Department, University of Michigan, MI 48109}
14:
15: \begin{abstract}
16: Ultra-luminous X-Ray Objects are off-nucleus point sources with
17: $L_X=10^{39}$-$10^{41}$ erg/s but the nature of such systems are largely
18: unidentified. Here we report a 2.1 hour period observed in a Chandra ACIS
19: observation for ULX M51 X-7, which is located on the edge of a young star
20: cluster in the star forming region in a spiral arm. In two ACIS observations
21: separated by one year, the ULX changed from a high-hard to a low-soft spectral
22: state, in contrast to most Galactic low mass X-ray binaries. Based on its
23: period and spectral behaviors, we suggest that this ULX is a low mass X-ray
24: binary system, with a dwarf companion of 0.2-0.3 $M_\odot$ and a compact
25: accretor, either a neutron star or a black hole, whose mass is not well
26: constrained. Relativistic beaming effects are likely involved to produce the
27: observed high X-ray luminosities given its low accretion rate as inferred from
28: a sustainable accretion scenario via Roche lobe overflow.
29:
30: \end{abstract}
31:
32: \keywords{Galaxy: individual(M51) --- X-rays: binaries}
33:
34:
35: \section{Introduction}
36:
37:
38: X-ray emission from active galactic nuclei and quasars, powered by accretion
39: onto nuclear super-massive black holes, are usually $>10^{41}$ erg/s, while for
40: the stellar binary systems in the Milky Way, the luminosities are typically
41: $10^{33}$-$10^{38}$ erg/s. X-ray objects with luminosities of
42: $10^{39}$-$10^{41}$ erg/s have been observed in some external galaxies as
43: off-nucleus point sources by X-ray satellites such as EINSTEIN, ROSAT, ASCA,
44: and recently Chandra X-ray Observatory (e.g., Fabbiano 1989; Marston et al.
45: 1995; Zezas et al. 2002). These objects have several names, a popular
46: designation being ``Ultra-luminous X-ray Objects'', or ULXs.
47:
48: One suggestion for the nature of these ULXs is that they are binary systems
49: with $10^3$-$10^4$ $M_{\odot}$ black holes as primaries (Colbert and Mushotzky
50: 1999). Such black holes, if they exist, are the missing links between stellar
51: mass black holes and super-massive black holes in the nuclei of galaxies. This
52: suggestion is consistent with the X-ray spectral of some ULXs (Makishima et
53: al. 2000). However, the formation of such massive black holes is not
54: predicted by stellar evolution theory and it may be impossible to form such
55: objects even in dense star clusters. Alternatively, these sources may be
56: stellar mass black holes or neutron stars whose emission is beamed, thereby
57: representing micro-quasars (King et al. 2001; King 2002). If the emission is
58: beamed, the intrinsic luminosities become sub-Eddington, and there are known
59: examples of beamed Galactic X-ray sources (c.f., Mirabel \& Rodriguez 1999). It
60: is also possible that the luminosities are truly super-Eddington, for example,
61: obtainable from accretion disks with radiation-driven inhomogeneities
62: (Begelman, 2002).
63:
64:
65: To identify which of the above suggestions is correct, one natural and fruitful
66: way is to associate them with known classes of objects by comparing spectral
67: behaviors, short-term and long-term variabilities, etc. Spectral analyses of
68: ULXs, not without problems, generally support the mass-accreting black hole
69: scenarios (e.g., Makishima et al. 2000). Spectral transitions were observed
70: for two ULXs in IC 342 between high-soft and low-hard states, reminiscent of
71: some black hole X-ray binaries in our Galaxy (Kubota et al. 2001). While many
72: ULXs are variable during individual observations, few periods have been
73: reported. So far, only a possible periodicity at several tens of hours of a ULX
74: in IC 342 has been reported (Sugiho et al. 2001). In this paper, we report a
75: 2.1 hour period for a ULX in M51 (NGC 5194), M51 X-7 as designated in Roberts
76: \& Warwick 2000. This ULX is located on the outskirts of a young open cluster
77: on a spiral arm of M51. In $\S$2, we present the two Chandra ACIS observations
78: utilized and the results. A period of 7620 seconds for the ULX is found from
79: the X-ray light, and a spectral transition is found between two Chandra
80: observations. In $\S$3, we discuss the implications of its period and spectral
81: evolution, and suggest that this ULX is a low mass X-ray binary system, with a
82: dwarf companion of 0.2-0.3 $M_\odot$ and a compact accretor. Possible
83: mechanisms to produce the high X-ray luminosity is also discussed. For the
84: distance to M51, we use 7.7 Mpc (Roberts \& Warwick 2000).
85:
86: \section{Data analysis and results}
87:
88: Two Chandra ACIS observations of M51, ObsID 354 (15 kilo-seconds,
89: 2000-06-20) and ObsID 1622 (27 kilo-seconds, 2001-06-23), were retrieved from
90: the archive. The data were analyzed with CIAO 2.2.1, CALDB 2.12, XSPEC
91: 11.2.0 and XRONOS 5.19. Pixel randomization of event positions was removed to
92: recover the original positions before CIAO task {\tt wavdetect} was used to detect
93: discrete sources on ACIS-S chips. The accuracy of absolute positions for these
94: sources is expected to be $0\farcs6$ (Aldcroft et al. 2000).
95:
96: The ULX M51 X-7 was observed in both data sets, with ACIS positions
97: differing by ($0\farcs5$, $0\farcs1$). For the true
98: position, we take the average, i.e., RA=13:30:0.99, Dec=47:13:43.9, with a
99: positional uncertainty of $0\farcs4$. Registering this position onto
100: an HST WFPC2 image, we find that it falls onto the edge of a star cluster in a
101: dusty star forming knot on a spiral arm of M51. Within its positional error
102: circle are a few stars fainter than 23 mag. The details of its potential
103: optical counterpart will be reported in a future paper.
104:
105: For timing and spectral analyses, we selected circular apertures with such
106: sizes that the number of source photons outside of the aperture is
107: approximately the number of background photons inside the aperture. The size
108: chosen is $4\farcs0$ for Observation 354, and $2\farcs0$ for Observation 1622.
109: The light curve of X-7 in the first ACIS observation (ObsID 354), defined by
110: about 550 photons, appears to have the structure of a single period, as shown
111: in Figure 1 in contrast with the background light curve. When fitting a
112: periodic signal, we derive a period of 7620 seconds (using XRONOS/efsearch with
113: 10 second resolution). The uncertainty in the period is approximately 500
114: seconds (a more accurate value is not possible due to the limited number of
115: periods involved). To determine whether this could be an instrumental effect,
116: we carried out a number of tests. We formed the light curves of all other
117: bright sources in the same ACIS field, but found no evidence for periods. The
118: period cannot be caused by background variation, since the count rate of the
119: background varies by less than 10\% (and is not periodic), while the count rate
120: of X-7 varies by an amplitude of more than 50\%, for both soft photons ($<1.5$
121: keV) and hard photons ($>1.5$ keV).
122:
123: Dithering may cause spurious periods when a source is dithering across a CCD
124: node boundary or gap, falls off of the detector, or crosses bad pixels
125: periodically. In this observation, the dither period is 1000 seconds in the
126: Chip Y direction, and 700 seconds in the Chip X direction. The dither pattern is
127: a Lissajous figure, spanning 16 arc-seconds from peak to peak. The ULX falls on
128: node 2 of chip S3 at (556,459). It is 22 arc-seconds away from the node 1-2
129: boundary, and no bad pixels fall within 30 arc-seconds of the source. Thus it is
130: unlikely that the period is due to the dithering period.
131:
132: In the second ACIS observation (ObsID 1622) that was taken 1 year after the
133: first, only 53 photons were collected for the ULX, too few to make a reliable
134: periodicity analysis. Note that the second observation was longer than the
135: first by a factor of about 1.8, so the photon flux in the 0.3-10keV decreased
136: by a factor of about 18 between these two observations.
137:
138: Photons in the 0.3-8 keV band are used to perform spectral fits for both
139: spectra, as shown in Figure 2. The CIAO thread {\tt acisspec} was used to
140: extract the spectra, and {\tt corrarf} was used to correct the continuous
141: degradation in the ACIS quantum efficiency. To get good $\chi^2$ statistics, we
142: grouped the first spectrum (ObsID 354) requiring a minimum of 20 photons per
143: bin. For the second spectrum (ObsID 1622), we required a minimum of 5 photons
144: per bin to get enough bins. We tried absorbed (wabs) power-law (powerlaw),
145: absorbed multi-color disk (diskbb) and absorbed blackbody (bb) models on each
146: spectrum. In Table 1 we list the best fit parameters, the errors for the
147: best-fit parameters are for the 90\% confidence level. For the first spectrum,
148: an absorbed blackbody model is not acceptable, but an absorbed power-law model
149: is acceptable, with a power-law photon index $\Gamma=1.25^{+0.24}_{-0.22}$; and
150: an absorbed multi-color disk model is also acceptable, with an inner edge
151: temperature $kT_{in}=2.10^{+0.63}_{-0.40}$ keV. To test whether there is an
152: additional component due to a neutron star solid surface, we also fit an
153: absorbed (multi-color disk + 2 keV blackbody) model to the first spectrum. With
154: one more free parameter, this model improves the spectral fit marginally as
155: compared to an absorbed multi-color disk model, and results in a smaller inner
156: edge temperature, $kT_{in}=0.82^{+1.57}_{-0.30}$ keV. However, this additional
157: component does not make a better model than a simple absorbed power-law model.
158:
159: For the second spectrum, an absorbed power-law model gives a power-law photon
160: index of $\Gamma>2.6$, an absorbed multi-color disk model gives an inner edge
161: temperature of $kT_{in}=0.29\pm0.09$ keV, while an absorbed blackbody
162: model is not acceptable. As compared with the first spectrum of higher photon
163: flux, the second spectrum is much softer, and the best-fit absorbing column
164: density is higher too. For simple absorbed power-law models, we switched the
165: photon indices of the two spectra, and fitted a frozen $\Gamma=2.6$ to the
166: first spectrum, and a frozen $\Gamma=1.25$ to the second spectrum, and found
167: that both models can be accepted only on the levels $<10^{-2}$. Thus we
168: conclude that the first spectrum is truly harder than the second spectrum.
169: Absorbed power-law models with smaller frozen $n_H$ values were also fitted to
170: the second spectrum and gave worse fits, indicating that the high column
171: density in the second spectrum is real.
172:
173: \section{Discussion}
174:
175: Periods are not uncommon in Galactic X-ray point sources, such as cataclysmic
176: variables (CVs), low mass X-ray binaries (LMXBs) and high mass X-ray binaries
177: (HMXBs). HMXBs have periods of more than a few days, because they are usually
178: widely separated systems. A two hour period is common for CVs, where the
179: accretor and the donor are white dwarfs. However, CV systems have typical
180: temperatures under $10^6$ K, with most of the radiation emitted in the
181: ultraviolet region and with X-ray luminosities below $10^{34}$ erg/s. LMXBs
182: usually have periods of 10+ hours, but they can have periods as short as one
183: hour, for example, 1624-490 has a period of 0.69 hours. The source studied
184: here, with a period of two hours and a temperature of $\sim1$ keV, is most
185: similar to the low mass X-ray binaries.
186:
187: The LMXB picture of ULXs does not contradict its location in a star forming
188: region. While observations show that ULXs preferentially occur in active star
189: forming regions (e.g.,Zezas et al. 2002), indicating their short lifetimes,
190: they nevertheless could be a short stage in the long lifetime of LMXBs. Indeed,
191: some ULXs have also been found outside of star forming regions, for example, in
192: gas poor elliptical galaxies such as NGC 1399 (Angelini et al. 2001), and in a
193: globular cluster on the outskirt of the bulge in edge-on spiral galaxy NGC 4565
194: (Wu et al. 2002).
195:
196: If X-7 is a LMXB with an orbital period of about 2 hours, there are a number of
197: implications. The secondary star would be filling its Roche Lobe and
198: undergoing mass loss to the primary, and from this picture, we can estimate the
199: mass of the secondary, $M_1$. For a mass ratio of $q=M_1/M_2<0.8$, the Roche
200: Lobe radius is $R_{cr}=0.46a({M_1 \over M_1+M_2})^{1/3}$ (Paczynski 1967), in
201: which $a$ is the separation between the donor and the accretor, and $M_2$ is
202: the mass of the accretor. Combined with Kepler's 3rd law, this leads to
203: $P_{orb}=8.9(R_1/R_\odot)^{3/2}(M_\odot/M_1)^{1/2}$ hours. For a late-type low
204: mass star, if we adopt an analytical mass-radius relation
205: $R_1/R_\odot=M_1/M_\odot$ (Verbunt 1993), we obtain the mass of the donor to be
206: $M_1\approx0.23M_\odot$. We also tried to obtain the mass of the donor by
207: adopting computed mass-radius relation from the Geneva stellar models (Lejeune
208: et al. 2001). We extended the star mass down to 0.1 $M_\odot$ by extrapolating
209: computed lgR-lgM relation for low mass stars ($0.8<M/M_\odot<0.9$). These
210: computed stellar models give an equivalent analytical relation of
211: $R_1/R_\odot\approx(M_1/M_\odot)^{1.2}$, and the resulted donor mass is about
212: 0.3 $M_\odot$. Such a low mass secondary in M51 is impossible to identify
213: even with the Hubble Space Telescope.
214:
215: The mass of the accretor cannot be constrained by the period under our
216: assumptions. The accretor is most likely a neutron star, or a black hole , as a
217: compact object is needed to produce the observed hard spectra. One can refine
218: this suggestion further by comparing the observed spectrum with models of
219: compact objects, as we try here. Neutron star LMXBs and black hole LMXBs at
220: high luminosities might be distinguished by their spectral features (cf. Tanaka
221: \& Lewin 1995), since high luminosity neutron star LMXBs exhibit a multi-color
222: disk component and a blackbody component with a color temperature of $\sim2$
223: keV, which is interpreted as emission from the neutron star surface, while
224: black hole LMXBs do not show such a single-color blackbody component at any
225: luminosity level. As the luminosity decreases further, the spectrum becomes a
226: single power-law, and distinction between black hole and neutron star LMXBs
227: diminishes. For X-7, the model with an additional 2 keV thermal component
228: improves the spectral fit marginally as compared to an absorbed multi-color
229: disk model, which may be very weak evidence that the accretor is a neutron
230: star. However, such a marginal improvement may be due to the additional freedom
231: of parameters, and does not really reflect the preference of a thermal
232: component. Also, its trend of spectral transition from a hard high state to a
233: soft low one has some Galactic parallels, such as Nova Muscae 1991 (Ebisawa et
234: al. 1994), whose system is comprised of a low mass companion and a black hole
235: of $\sim6$ $M_\odot$. Such spectral transitions, as noted by Terishima \&
236: Wilson (2002), are in contradiction to most Galactic LMXBs which switch between
237: soft high and hard low states. Thus it is premature to identify it as a neutron
238: star or a black hole with current observations.
239:
240: One concern about this picture is that mass transfer from the less massive
241: donor to the accretor will widen the orbit and enlarge the Roche lobe with
242: respect to the donor, and thus prevent further mass transfer via Roche lobe
243: overflow. This can be avoided if the orbit angular momentum can be dissipated
244: effectively by gravitational radiation, with the rate $-(\dot{J}/J)_{GR}={32G^3
245: \over 5c^5}M_1M_2(M_1+M_2)/a^4$. The separation $a$ evolves according to the
246: equation $\dot{a}/a = 2\dot{J}/J-2(1-M_1/M_2)\dot{M_1}/M_1$ under conservative
247: mass transfer via Roche lobe overflow(Verbunt 1993). In order for the orbit
248: not to widen, the mass transfer rate via Roche lobe overflow must be less than
249: a value of about $10^{-9}-10^{-8}$ $M_\odot/yr$ for a primary of 3-100
250: $M_\odot$. In a steady-state, accretion disks with these accretion rates
251: generate luminosities of $10^{37}-10^{38}$ erg/s regardless of the mass of the
252: accretor, less than $10^{39}$ erg/s we observed for X-7. The larger observed
253: energy would require an additional phenomenon, such as a transient outburst or
254: relativistic beaming. For example, X-ray novae can have substantial outbursts,
255: but usually with a duty cycle of less than 10\%. However, this source has been
256: brighter than $10^{39}$ erg/s during the past two decades (Figure 3), in
257: conflict with expectations for dwarf novae. Therefore, the more likely
258: explanation for the high luminosities is by relativistic beaming, as suggested
259: by King et al. (2001).
260:
261: An alternative model for the binary and the periodicity is that the period
262: arises due to a hot spot in the accretion disk around the primary. This
263: picture removes the restrictions on the mass of the secondary and its
264: separation from the primary. If we assume Keplerian motion of the hot spot
265: around the accretor, this 2 hour period corresponds to a distance of $\sim1-2$
266: R$_\odot$ for an accretor mass of $\sim3-10$ M$_\odot$. At such distances, the
267: multi-color disk temperature is normally $10^4$ K. A hot spot at such distances
268: fails to explain the fluctuation amplitude for hard photons ($>1.5$ keV).
269:
270: Another possibility is that the period is not real and results from the limited
271: sampling of a randomly variable object. If the period reflects that of orbital
272: motion, it should be highly reproducible, provided that the source is
273: sufficiently bright. If it is due to the motion of a hot spot on a disk, one
274: might expect the hot spot to fade with time (then the period would vanish while
275: the source might still be luminous) or the spot might move inward in the disk,
276: leading to a shorter period. Additional observations that could distinguish
277: between these expectations would be greatly beneficial in clarifying the nature
278: of the variation seen in M51 X-7.
279:
280: \acknowledgements
281:
282:
283: We are grateful for the service of Chandra Data Archive. We would like to thank
284: Renato Dupke, Eric Miller, Rick Rothschild, and Steven Murray for helpful
285: discussions. We gratefully acknowledge support for this work from NASA under
286: grants HST-GO-09073.
287:
288: \begin{thebibliography}{}
289:
290: \bibitem[]{286} Aldcroft, T.L., Karovska, M., et al. 2000, Proc. SPIE, 4012, 650
291:
292: \bibitem[]{288} Angelini, L., Loewenstein, M., and Mushotzky, R.F. 2001, \apj, 557,
293: L35
294:
295: \bibitem[]{291} Begelman, M.C., 2002, \apj, 568, L97
296:
297: \bibitem[]{293} Colbert, E. J. M. and Mushotzky, R. F. 1999, \apj, 519, 89
298:
299: \bibitem[]{295} Ebisawa, K., Ogawa, M. et al. 1994, PASJ, 46, 375
300:
301: \bibitem[]{297} Fabbiano, G. 1989, ARA\&A, 27, 87
302:
303: \bibitem[]{299} King, A. R., Davies, M. B., Ward, M. J., Fabbiano, G. and Elvis, M.
304: 2001, \apj, 552, L109
305:
306: \bibitem[]{} King, A. R., 2002, MNRAS, 335, L13
307:
308: \bibitem[]{302} Kubota, A., Mizuno, T., Makishima, K. et al. 2001, ApJ, 547, L119
309:
310: \bibitem[]{304} Lejeune, T., and Schaerer, D., 2001, A\&A 366, 538
311:
312: \bibitem[]{306} Makishima, K., Kubota, A. Muzuno, T. et al. 2000, \apj, 535, 632
313:
314: \bibitem[]{308} Mirabel, I. F. and Rodriguez, L.F., 1999, ARA\&A, 37, 409
315:
316: \bibitem[]{310} Marston, A.P., Elmegreen, D., and Elmegreen, B. 1995, \apj, 438,663
317:
318: \bibitem[]{312} Paczynski, B. 1967, {\it Acta Astron. } 17, 287
319:
320: \bibitem[]{314} Roberts, T. P. and Warwick, R. S. 2000, MNRAS, 315, 98
321:
322: \bibitem[]{316} Sugiho, M., Kotoku, J., Makishima, K., Kubota, A., Mizuno, T.,
323: Fukazawa, Y., and Tashiro, M. 2001, \apj, 561, L73
324:
325: \bibitem[]{319} Tanaka, Y. and Lewin, W.H.G. 1995, in {\it X-ray Binaries}, eds.
326: Lewin, W.H.G, van Paradijs, J. and van den Heuvel, E.P.J. P.265
327:
328:
329: \bibitem[]{323} Terishima, Y. and Wilson, A. 2002, astro-ph/0204321
330:
331: \bibitem[]{325} Verbunt, F. 1993, ARA\&A, 31, 93
332:
333: \bibitem[]{327} Wu, H., Xue, S.J., Xia, X.Y., Deng, Z.G., and Mao, S.D. 2002,
334: astro-ph/0204359
335:
336: \bibitem[]{330} Zezas, A., Fabbiano, G., Rots, A.H., and Murray, S.S. 2002,
337: astro-ph/0203175
338:
339: \end{thebibliography}
340:
341: \clearpage
342:
343: \begin{deluxetable}{lllllllll}
344: \tablecaption{Spectral Fits for M51 X-7 in 2 ACIS observations}
345: \tabletypesize{\tiny}
346: \tablehead{
347: \colhead{ObsID} & \colhead{model} & \colhead{$N_H$} & \colhead{$\Gamma$} &
348: \colhead{$kT_{in}$} & \colhead{$kT$} & \colhead{$\chi_\nu^2$/dof} & \colhead{absorbed $L_X$\tablenotemark{a}} & \colhead{unabsorbed $L_X$\tablenotemark{a}} \\
349: \colhead{}& \colhead{}& \colhead{($10^{22}$ $cm^{-2}$)} & \colhead{} &
350: \colhead{(keV)} & \colhead{(keV)} &\colhead{} & \colhead{($10^{38}$erg/s)} & \colhead{($10^{38}$erg/s)}
351: }
352:
353: \startdata
354: & wabs*powerlaw & $0.06^{+0.07}_{-0.06}$ & $1.25^{+0.24}_{-0.22}$ & \nodata & \nodata & 0.95/21 & 18.6 & 19.8 \\
355: 354 & wabs*diskbb & $<0.03$ & \nodata & $2.12^{+0.71}_{-0.42}$ & \nodata & 1.06/21 & 17.1 & 17.1 \\
356: & wabs*bb & \nodata & \nodata & \nodata & \nodata & 3.47/21 & \nodata & \nodata\\
357: & wabs*(bb+diskbb) & $0.02^{+0.09}_{-0.02}$ &\nodata&$0.82^{+1.57}_{-0.30}$ & 2.0 (frozen)& 1.01/20 & 18.8 & 19.2 \\
358: \hline
359: & wabs*powerlaw & $0.70^{+1.02}_{-0.55}$ & $>2.6$ & \nodata & \nodata & 0.82/5 & 0.39 & 9.34\\
360: 1622& wabs*diskbb & $0.27^{+1.09}_{-0.27}$ & \nodata & $0.29\pm0.09$ & \nodata & 0.89/5 & 0.30 & 0.77 \\
361: & wabs*bb & \nodata & \nodata & \nodata & \nodata & 4.61/7 & \nodata & \nodata \\
362: \enddata
363: \tablenotetext{a}{The X-ray luminosities are in 0.3-8 keV.}
364: \end{deluxetable}
365:
366: \clearpage
367:
368: \begin{figure}
369: \plotone{f1.eps}
370: \caption{Light curve of M51 X-7 in ACIS observation ObsID354. The ULX flux
371: varies periodically by an amplitude of more than 50\%. The overlaid light curve
372: labeled by squares is the scaled background, which shows variation of less
373: than 10\%. Photons within 0.3-10 keV were used for the light curves.}
374:
375: \end{figure}
376:
377: \begin{figure}
378: \epsscale{0.8}
379: \plotone{f2.eps}
380: \caption{Spectral fits of M51 X-7 in two ACIS observations. In the first
381: observation (ObsID 354) the ULX was in a high hard state, while in the second
382: observation (ObsID 1622, model and $\chi^2$ in thick lines) it changed to a low soft state.}
383:
384: \end{figure}
385:
386: \begin{figure}
387: \plotone{f3.eps}
388: \caption{Long term variability of M51 X-7. Extracted from the archive were
389: 1 EINSTEIN HRI observation (triangles), 1 ROSAT PSPC observation (pentagons), 5
390: ROSAT HRI observations (squares) and 2 Chandra ACIS observations (circles). The
391: instrumental count rates were converted to flux assuming absorbed power-law
392: models (a) if in the high-hard state (open symbols), and (b) if in the low-soft
393: state (filled symbols). A distance of 7.7 Mpc was assumed to convert fluxes to
394: luminosities. }
395:
396: \end{figure}
397:
398: \end{document}
399:
400: