1: % paper on RXTE observations of M82
2: % PK 5/10/07-5/24/07
3:
4: %\documentclass{aastex}
5: \documentclass{emulateapj}
6: %\usepackage{emulateapj5,apjfonts}
7:
8: % format needed for submission
9: %\documentclass[12pt,preprint]{aastex}
10: % preprint2 produces a double-column, single-spaced document:
11: %\documentclass[preprint2]{aastex}
12: %\usepackage{psfig}
13:
14: % You can insert a short comment on the title page using the command below.
15: %\slugcomment{submitted to ApJ}
16:
17: \shorttitle{X-Ray Periodicity from M82}
18: \shortauthors{Kaaret, Feng, \& Lang}
19:
20: \begin{document}
21:
22: \title{Confirmation of the 62 Day X-Ray Periodicity from M82}
23:
24: \author{Philip Kaaret and Hua Feng}
25: \affil{Department of Physics and Astronomy, University of Iowa, Van
26: Allen Hall, Iowa City, IA 52242.}
27:
28: %\thanks{E-mail: philip-kaaret@uiowa.edu}
29:
30: \begin{abstract}
31:
32: Using 400 days of new X-ray monitoring of M82, we confirm the 62~day
33: periodicity previously reported. In the full data set spanning
34: 1124~days, we find a period of $62.0 \pm 0.3$ days and a coherence, $Q =
35: 22.3$, that is consistent with a strictly periodic signal. We estimate
36: that the probability of chance occurrence of our observed signal is $6
37: \times 10^{-7}$. The light curve folded at this period is roughly
38: sinusoidal and has a peak to peak amplitude of $(0.99 \pm 0.10) \times
39: 10^{-11} \rm \, erg \, cm^{-2}\, s^{-1}$. Confirmation of the
40: periodicity strengthens our previous suggestion that the 62~day
41: modulation is due to orbital motion within an X-ray binary.
42:
43: \end{abstract}
44:
45: \keywords{black hole physics -- galaxies: individual: M82 galaxies:
46: stellar content -- X-rays: galaxies -- X-rays: binaries}
47:
48:
49: \section{Introduction}
50:
51: Bright, non-nuclear X-ray sources in external galaxies, the so-called
52: ultraluminous X-ray sources (ULXs), represent either intermediate-mass
53: black holes \citep{Colbert99,Makishima00,Kaaret01} or super-Eddington
54: accretion onto stellar-mass black holes. The brightest X-ray source in
55: the nearby starburst galaxy M82, CXOU J095550.2+694047 = X41.4+60
56: \citep{Kaaret01}, is one of the most extreme ULXs. Assuming isotropic
57: radiation, a black hole mass of at least $500 M_{\sun}$ is required to
58: avoid violating the Eddington limit. The source also shows
59: quasiperiodic oscillations (QPOs) at relatively low frequencies,
60: 50-190~mHz, suggesting a relatively high compact object mass
61: \citep{Strohmayer03,Dewangan06,Mucciarelli06,Kaaret06b}. The relative
62: proximity and brightness of source enables studies that are not feasible
63: for other ULXs.
64:
65: We monitored the X-ray emission from M82 for 240~days in 2004/2005 and
66: detected a period of 62~days \citep{Kaaret06a}. We interpreted the
67: 62~day period as the orbital period of the ULX binary system. For a
68: system in Roche-lobe contact, such a long orbital period implies a low
69: density companion on the giant or supergiant branch. Identification of
70: the evolutionary phase of the companion star represents a significance
71: advance in our knowledge of ULXs. In order to test this interpretation
72: of the 62~day periodicity, we obtained additional monitoring of M82 in
73: 2006/2007. We describe the new X-ray observations in \S~2 and discuss
74: the results in \S~3.
75:
76:
77: \section{Observations}
78:
79: We obtained 187 observations of M82 using the Proportional Counter Array
80: (PCA) on the Rossi X-Ray Timing Explorer (RXTE) covering MJD 53825 to
81: 54223 with observations approximately every other day under RXTE program
82: 92098. The observations were typically less than 1~ks. We also
83: analyzed archival data from RXTE program 90121 which consists of 144
84: observations roughly every other day from MJD 53252 to 53490 and two
85: observations made earlier. In this earlier program, the observations
86: were typically about 2~ks each.
87:
88: \begin{figure*}[tb]
89: \centerline{\includegraphics[width=6.25in]{f1.eps}}
90: \caption{\label{lightcurve} Light curve of M82 in the 2--10~keV band.
91: The plot shows the flux measured using the PCA for each observation
92: versus the observation date in MJD. The flux includes contributions
93: from all X-ray sources within M82.} \end{figure*}
94:
95: We used the RXTE production data and processed the data using HEAsoft
96: version 6.2. We selected good time intervals where the time since the
97: last SAA passage was more than 30 minutes, the electron contamination
98: was less than 0.1, and the pointing was within $0.1\arcdeg$ of the
99: target and at least $10\arcdeg$ above the horizon. We produced spectra
100: using only the top layer in Proportional Counter Unit 2 and estimated
101: the background using the faint source background model. We fitted the
102: spectra using XSPEC 11 in the energy range 2.6--20~keV with a power-law
103: model with an interstellar absorption column density fixed to $3 \times
104: 10^{22} \rm \, cm^{-2}$. The best fit model was used to calculate the
105: absorbed flux in the 2-10 keV band.
106:
107: Fig.~\ref{lightcurve} shows the flux in the 2-10~keV band versus time.
108: We note that, due to the large angular acceptance of the PCA, the
109: measured flux includes contributions from all X-ray sources within M82.
110: Thus, some part of the fluctuations represents sources other than
111: X41.4+60. There is an X-ray flare around MJD 53400 which was discussed
112: in \citet{Kaaret06b}. Points with fluxes above $4 \times 10^{-11} \rm
113: \, erg \, cm^{-2} \, s^{-1}$ were removed in the subsequent analysis.
114: The photon index is generally between 2.0 and 2.7 with an average value
115: of 2.4. The distribution of the measured photon indexes are consistent,
116: within the uncertainties, with a constant value of 2.4.
117:
118: \begin{figure}[tb]
119: \centerline{\includegraphics[width=3.25in]{f2.eps}}
120: \caption{\label{periodogram} Periodogram of the 2--10~keV light curve of
121: M82. The strongest peak is at a period of $62.0 \pm 0.3$~days. The two
122: secondary peaks near the main peak are aliases due to the gap in the
123: monitoring. The powers are calculated using the method of
124: \citet{Horne86} and are normalized by the total variance of the fluxes.}
125: \end{figure}
126:
127: The light curve shows an apparent modulation with a period near
128: 60~days. Fig.~\ref{periodogram} shows a periodogram with the power
129: normalized by the total variance of the data \citep{Horne86}. There is
130: a peak at a period 62.0~days with a power of 77.9. We estimate the 90\%
131: confidence error on the period to be 0.3~days. There are two secondary
132: peaks near the main peak which are aliases due to the gap in the
133: monitoring. Retaining the X-ray flare near MJD 53400 does not shift the
134: peak, but decreases the power (because of the normalization to the total
135: variance of the data) to 70.5. Using fluxes calculated from spectral
136: fits with the photon index fixed to 2.4 does not significantly change
137: the period or the power of the peak.
138:
139: We tested the significance of the observed signal using a red noise
140: background, as is appropriate for an accreting X-ray source
141: \citep{Israel96,Vaughan05}. We fitted the power versus frequency
142: relation for periods in the range 6--280 days to a power-law form and
143: found a spectral index of $-1.04 \pm 0.06$. We generated red noise
144: with a spectral index of $-1.10$ and with mean and variance equal to
145: those of the data using the {\tt rndpwrlc} routine of the {\tt aitlib}
146: IDL subroutine library provided by the Institut f\"ur Astronomie und
147: Astrophysik of the Universit\"at T\"ubingen \citep{Timmer94}. The
148: duration of each generated light curve is longer than the actual data in
149: order to minimize the effects of red noise leakage. Each light curve
150: contains 8192 data points with uniform spacing of 0.66~days and a subset
151: of 330 points from the middle of this set with relative times matching
152: the actual observations are extracted for analysis. These 330 simulated
153: fluxes were processed with the same procedures used to analyze the real
154: data. We generated $2 \times 10^{6}$ trial light curves and searched
155: for cases where the power at periods of 10 to 150 days was greater than
156: or equal to the observed value of 77.9. We found one such case and
157: estimate the probability of chance occurrence of our observed signal to
158: be $5 \times 10^{-7}$. Fitting the distribution of maximum observed
159: powers, we estimate that the probability of chance occurrence of our
160: observed signal is $6 \times 10^{-7}$, in good agreement. This
161: procedure is conservative because it includes the signal in the
162: calculation of the power-law slope and the variance and because the
163: period search range, 10--150 days, extends to significantly lower
164: frequencies than the observed period where the red noise produces high
165: amplitude fluctuations. If we restrict the search range to periods of
166: 62 days or less, then a fit to the distribution of the maximum observed
167: powers indicates that the probability of chance occurrence of our
168: observed signal is $3 \times 10^{-13}$.
169:
170: The coherence or quality value of the peak signal, the period of the
171: peak divided by the full width at half maximum power, is $Q = 22.3$.
172: This is fully consistent with that expected for a periodic process given
173: the observation duration.
174:
175: \begin{figure}[tb]
176: \centerline{\includegraphics[height=3.25in]{f3.eps}}
177: \caption{\label{phase} Flux from M82 folded at a period of 61.98 days.
178: Two periods are shown for clarity. Each point is the average flux of
179: the observations falling within the given phase bin and the error bar is
180: the standard error.}
181: \end{figure}
182:
183: Fig.~\ref{phase} shows the data folded at the best fit period. Each
184: point is the average flux of the observations falling within the given
185: phase bin and the error bar is the standard error, i.e.\ the standard
186: deviation of the fluxes in each bin divided by the square root of the
187: number of fluxes in that bin. The amplitude of the modulation, taken as
188: the maximum average flux in one bin minus the minimum, is $(0.99 \pm
189: 0.10) \times 10^{-11} \rm \, erg \, cm^{-2} \, s^{-1}$.
190:
191:
192: To search for rapid variability, we extracted event files with high time
193: resolution data for the 187 observations. Events in the 2.4-11.9~keV
194: energy band were selected in the good time intervals defined above and
195: split into segments of 256~s. An FFT with a time resolution of 1~s was
196: calculated for each segment. The FFTs within each observation were
197: added incoherently. The resulting total power spectrum was
198: logarithmically rebinned with a bin width of 16\%, equal to the widths
199: of the QPOs previously detected from M82. We searched for individual
200: bins with high powers and calculated the significance taking into
201: account the number of bins in each power spectrum. The highest power
202: was recorded on MJD 54155.148 in a single 256~s interval of data at a
203: frequency of $20 \pm 4$~mHz with a Leahy power of 19.7 corresponding to
204: a chance probability of occurrence of $5.4 \times 10^{-5}$ ($4.0\sigma$)
205: single trial and 0.0012 taking into account the trials for that single
206: observation. Taking into account all observations in program 92098, the
207: chance probability of occurrence is 0.23 indicating that the QPO
208: detection is not significant. The observations in the new RXTE program
209: are too short to provide good sensitivity for QPO detection.
210:
211:
212: \section{Discussion}
213:
214: Using 400 days of new X-ray monitoring of M82, we confirm the 62~day
215: previously reported \citep{Kaaret06a}. The coherence of the signal, $Q
216: = 22.3$, is consistent with a strictly periodic signal. The light
217: curve folded at this period is roughly sinusoidal and has a peak to peak
218: amplitude of $1.0 \times 10^{-11} \rm \, erg \, cm^{-2}\, s^{-1}$. The
219: new observations rule out the possibility that the signal could be
220: red-noise fluctuation. The remaining possible interpretations of the
221: signals are that it represents either the orbital period of the binary
222: system or a superorbital modulation.
223:
224: Superorbital modulations are most often interpreted as due to accretion
225: disk precession \citep{Wijers99,Ogilvie01} and, in this case, would
226: represent variations in our viewing angle of the disk. Indeed, in the
227: one source where both a precessing relativistic jet (thought to be
228: launched perpendicular to the disk) and a superorbital modulation are
229: detected, SS 433, the periods of jet precession and superorbital
230: modulation are the same \citep{Margon89}. If ULXs are beamed sources,
231: then the beaming cone would most naturally be perpendicular to the disk
232: axis. Thus, accretion disk precession would naturally produce
233: modulation of the beam at the superorbital period.
234:
235: \begin{figure}[tb] \centerline{\includegraphics[height=3.25in]{f4.eps}}
236: \caption{\label{superorb} Distribution of superorbital periods of
237: neutron star binaries (diagonally hatched) and black hole candidate
238: binaries (horizontally hatched). The vertical dashed line shows the
239: period measured from M82; it falls in the neutron star binary range.}
240: \end{figure}
241:
242: The distribution of superorbital periods of neutron star and black hole
243: candidate X-ray binaries are plotted in Fig.~\ref{superorb}
244: \citep{Wijers99,Smith02,Corbet03,Rau03}. We include only black hole
245: candidates as defined by \citet{Remillard06}. It is apparent from the
246: figure that there are no black hole binaries with superorbital periods
247: near 62 days. The shortest superorbital period from a dynamically
248: confirmed black hole candidate is the 198~day period of LMC X-3 which is
249: longer by more than a factor of 3. Thus, if the 62 period from M82 is
250: interpreted as a superorbital period, this would suggest that it arises
251: from a neutron star system. The measured flux modulation implies a
252: luminosity (assuming isotropic emission) of $1.6 \times 10^{40} \rm \,
253: erg \, s^{-1}$ for a source in M82 at a distance of 3.63~Mpc. This
254: would exceed the Eddington limit for a $1.4 M_{\odot}$ neutron star by
255: at least a factor of 86. The brightest known neutron star transient is
256: A0538-66 which reached a peak luminosity of $8\times 10^{38} \rm \, erg
257: \, s^{-1}$ \citep{White78}, a factor of 20 lower than X41.4+60.
258: Compared instead to the peak luminosity of flares from X41.4+60, the
259: peak luminosity of A0538-66 is a factor of 95 lower. Thus, a neutron
260: star interpretation for X41.4+60, as would be expected if the 62 day
261: period is a superorbital modulation, is untenable.
262:
263: When monitored over many periods, superorbital modulations show reduced
264: coherence in the form of period or phase shifts. The high coherence
265: measured for the M82 periodicity is inconsistent with all but the most
266: stable of the superorbital modulations, specifically that of Her X-1.
267:
268: Several X-ray binaries produce X-rays modulated at the orbital period
269: including Cygnus X-3 \citep{Elsner80}, LMC X-3 \citep{Boyd01}, 1E
270: 1740.7-2942 and GRS 1758-258 \citep{Smith02}, and GX 13+1
271: \citep{Corbet03}. Thus, the periodicity from M82 may be interpreted as
272: due to orbital modulation. The coherence of the signal we observe from
273: M82 is consistent with that expected for a strictly periodic signal
274: given the observational coverage. This is consistent with
275: interpretation as an orbital modulation. We conclude that the 62~day
276: periodicity most likely indicates the orbital period of an X-ray binary.
277: If the companion fills it Roche-lobe, as expected in a system with a
278: mass accretion rate high enough to produce the observed X-ray flux even
279: with moderate beaming, then the long period indicates that the companion
280: star has a low average density, $5 \times 10^{-5} \rm \, g \, cm^{-3}$,
281: and is therefore a giant or supergiant.
282:
283:
284: \acknowledgements
285:
286: PK and HF acknowledge partial support from NASA Grant NNX06AG77G. PK
287: acknowledges support from a University of Iowa Faculty Scholar Award.
288:
289:
290: %--------------
291:
292: \begin{thebibliography}{}
293:
294: \bibitem[Boyd, Smale, Dolan(2001)]{Boyd01} Boyd, P.T., Smale, A.P.,
295: Dolan, J. F.\ 2001, ApJ, 555, 822
296:
297: \bibitem[Colbert \& Mushotzky(1999)]{Colbert99} Colbert, E.J.M.\ \&
298: Mushotzky, R.F.\ 1999, ApJ, 519, 89
299:
300: \bibitem[Corbet(2003)]{Corbet03} Corbet, R.H.D.\ 2003, ApJ, 595, 1086
301:
302: \bibitem[Dewangan, Titarchuk, \& Griffiths(2006)]{Dewangan06} Dewangan,
303: G.C.\, Titarchuk, L., Griffiths, R.E.\ 2006, ApJ, 637, L21
304:
305: \bibitem[Elsner et al.(1980)]{Elsner80} Elsner, R.F., Ghosh, P.,
306: Darbro, W., Weisskopf, M.C., Sutherland, P.G., Grindlay, J.E.\ 1980,
307: ApJ, 239, 355
308:
309: \bibitem[Horne \& Baliunas(1986)]{Horne86} Horne, J.H.\ \& Baliunas,
310: S.L.\ 1986, ApJ, 302, 757
311:
312: \bibitem[Israel \& Stella(1996)]{Israel96} Israel, G.L.\ \& Stella, L.\
313: 1996, ApJ, 468, 369
314:
315: \bibitem[Kaaret et al.(2001)]{Kaaret01} Kaaret, P.\ et al.\ 2001, MNRAS,
316: 321, L29
317:
318: \bibitem[Kaaret, Simet, \& Lang(2006a)]{Kaaret06a} Kaaret, P., Simet,
319: M.G., Lang, C.C.\ 2006, Science, 311, 491
320:
321: \bibitem[Kaaret, Simet, \& Lang(2006b)]{Kaaret06b} Kaaret, P., Simet,
322: M.G., Lang, C.C.\ 2006, ApJ, 646, 174
323:
324: \bibitem[Makishima et al.(2000)]{Makishima00} Makishima, K.\ et al.\
325: 2000, ApJ, 535, 632
326:
327: \bibitem[Margon \& Anderson(1989)]{Margon89} Margon, B.\ \& Anderson,
328: S.F.\ 1989, 347, 448
329:
330: \bibitem[Mucciarelli et al.(2006)]{Mucciarelli06} Mucciarelli, P.,
331: Casella, P., Belloni, T., Zampieri, L., Ranalli, P.\ 2006, MNRAS, 365,
332: 1123
333:
334: \bibitem[Ogilvie \& Dubus(2001)]{Ogilvie01} Ogilvie, G.I. \& Dubus, G.\
335: 2001, MNRAS, 320, 485
336:
337: \bibitem[Rau, Greiner, McCollough(2003)]{Rau03} Rau, A., Greiner, J.,
338: McCollough, M.L.\ 2003, ApJ, 590, L37
339:
340: \bibitem[Remillard \& McClintock(2006)]{Remillard06} Remillard, R.A.\ \&
341: McClintock, J.E.\ 2006, ARA\&A, 44, 49
342:
343: \bibitem[Smith, Heindl, \& Swank (2002)]{Smith02} Smith, D.M., Heindl,
344: W.A., Swank, J.H. 2002, ApJ, 578, L132
345:
346: \bibitem[Strohmayer \& Mushotzky(2003)]{Strohmayer03} Strohmayer, T.E.\
347: \& Mushotzky, R.F.\ 2003, ApJ, 586, L61
348:
349: \bibitem[Timmer \& K\"onig(1994)]{Timmer94} Timmer, J.\ \& K\"onig, M.\
350: 1994, A\&A, 300, 707
351:
352: \bibitem[Vaughan(2005)]{Vaughan05} Vaughan, S.\ 2005, MNRAS, 431, 391
353:
354: \bibitem[White \& Carpenter(1978)]{White78} White, N.E.\ \& Carpenter,
355: G.F.\ 1978, MNRAS, 183, 11P
356:
357: \bibitem[Wijers \& Pringle(1999)]{Wijers99} Wijers, R.A.M.J.\ \&
358: Pringle, J.E.\ 1999, MNRAS, 308, 207
359:
360: \end{thebibliography}
361:
362: \label{lastpage}
363:
364: \end{document}
365:
366:
367: