1: \documentclass[useAMS,usenatbib]{mn2e}
2: \usepackage{graphicx}
3: % If your system does not have the AMS fonts version 2.0 installed, then
4: % remove the useAMS option.
5: %
6: % useAMS allows you to obtain upright Greek characters.
7: % e.g. \umu, \upi etc. See the section on "Upright Greek characters" in
8: % this guide for further information.
9: %
10: % If you are using AMS 2.0 fonts, bold math letters/symbols are available
11: % at a larger range of sizes for NFSS release 1 and 2 (using \boldmath or
12: % preferably \bmath).
13: %
14: % The usenatbib command allows the use of Patrick Daly's natbib.sty for
15: % cross-referencing.
16: %
17: % If you wish to typeset the paper in Times font (if you do not have the
18: % PostScript Type 1 Computer Modern fonts you will need to do this to get
19: % smoother fonts in a PDF file) then uncomment the next line
20: % \usepackage{Times}
21:
22: %%%%% AUTHORS - PLACE YOUR OWN MACROS HERE %%%%%
23: \newcommand{\swift}{SWIFT J1753.5$-$0127}
24: \newcommand{\xte}{XTE J1118+480}
25: \newcommand{\gx}{GX 339$-$4}
26:
27: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
28:
29: \title[\swift: multiwavelength observations and
30: reactivation]{Multiwavelength spectral and high time resolution
31: observations of \swift: new activity?}
32: \author[Durant et al.]{M. Durant$^{1}$\thanks{E-mail:
33: durant@iac.es}, P. Gandhi$^2$, T. Shahbaz$^1$, H. H. Peralta$^{1}$ and
34: V. S. Dhillon$^3$\\
35: $^1$ Instituto de Astrof\'isica de Canarias, La Laguna, E38205 Tenerife,
36: Spain\\
37: $^2$ RIKEN Institute of Physical and Chemical Research, 2-1 Hirosawa,
38: Wakoshi, Saitama, Japan\\
39: $^3$ Department of Physics and Astronomy, University of Sheffield,
40: Sheffield S3 7RH, UK
41: }
42: \begin{document}
43:
44: \date{~}
45:
46: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2008}
47:
48: \maketitle
49:
50: \label{firstpage}
51:
52: \begin{abstract}
53: We have conducted an extensive observational campaign of
54: \swift\ during June 2007 after its bright outburst episode in
55: 2005. We have performed multi-band optical photometry, optical
56: spectroscopy, X-ray spectroscopy and timing and ULTRACAM optical
57: photometry simultaneously in three bands. Both the optical spectrum
58: and the X-ray spectrum, along with enhanced brightness in broad-band
59: photometry point to recent increased activity. We analyze the
60: different spectral regions, finding a { smooth optical continuum} with a
61: remarkable lack of lines and a very blue component modulated with a period of
62: 3.2\,hr and a hard power-law X-ray spectrum. Both the X-ray and
63: optical power spectra are flat at low frequencies up to the 0.1\,Hz
64: (10\,s) range, then decreasing roughly as a power law consistent with
65: flickering. Furthermore, the optical data show quasi-periodic
66: oscillations (QPOs) near 0.08\,Hz (13\,s). Together with a dynamical
67: and auto-correlation analysis of the light curves we attempt to
68: construct a complete physical picture of this intriguing system.
69: \end{abstract}
70:
71: \begin{keywords}
72: X-rays: binaries, binaries: individual (SWIFT J1753.5-0127), timing analysis.
73: \end{keywords}
74:
75: \section{Introduction}
76: Soft X-ray Transients (SXRTs) are a subset of the low mass X-ray binaries
77: (LMXBs) which are
78: thought to contain a black hole primary, and undergo periodic brightening
79: episodes lasting a few months, with recurrence times of a few years (Tanaka \&
80: Shibazaki, 1996). The outburst
81: episodes typically coincide with spectral and timing changes, for
82: example the X-ray {\em high/soft} state (thermal emission when accretion is
83: high) and the X-ray {\em low/hard} state (coronal/jet or optically
84: thin disc emission when accretion is low; review in Remillard \&
85: McClintock, 2006). Such changes are
86: attributed to varying contributions from the various emission
87: mechanisms e.g., dense disc, jet, low density zone (e.g., van der
88: Klis, 2006). It turns out,
89: however, that such a classification may be
90: simplistic, as some systems have been seen to go into outbursts
91: without changing their X-ray state. It is a challenge to understand, in
92: these cases, what indeed is changing (e.g., Meyer-Hofmeister, 2004).
93:
94: Two main models exist as possible explanations for the low/hard state
95: in accreting binaries. The first and oldest is the Advection-Dominated
96: Accretion Flow (ADAF, e.g., Czerny et al. 2000) model, in which the
97: accretion disc is truncated at a
98: large inner radius, and the relatively inefficient ADAF channels
99: matter to the compact object and to any jets; this region is
100: low-density and responsible for the high-energy emission. In a typical
101: outburst cycle, the accretion disc gains enough mass to overwhelm the
102: ADAF, and fast accretion progresses until the disc is depleted,
103: whereupon the ADAF reforms from small radii to large. The contender
104: model, the Accretion Disc Corona (ADC; e.g., Malzac, 2007) is not
105: dissimilar in its
106: effects, but the mechanism is subtly different. A hot corona of
107: energetic particles may form above and below an accretion disc,
108: and most of the accretion power is carried by strong magnetic
109: fields. When the fields in
110: the corona are strong, the emission from the corona is strong, and a
111: higher proportion of in-falling material is channelled into the
112: jets. When the mass-flow rate increases beyond a critical level,
113: however, enough soft X-ray photons are emittied by the disc to
114: efficiently cool the corona, causing it to condense onto the
115: disc, halting the chanelling of energy and particles away from the
116: core region, which then dominates the emitted spectrum with thermal
117: radiation from small radii. Note that this model was developed
118: initially fpr sources where the central accretion disc is obscured by the
119: outer torus of material, making the ADC relatively more important.
120:
121: In general, it may be
122: assumed that both processes, with their different physical mechanisms,
123: occur in systems either simultaneously or episodically. The most
124: obvious difference between the two is that for the ADC, the accretion
125: disc extends down to very small radii, to the last stable orbit in the
126: case of a black-hole system. The
127: transitions between accretion modes and feedback in the systems is
128: very poorly understood, however. It seems not to be purely a function
129: of accretion rate, and there certainly appears to be hysteresis in all
130: the transient systems, as typically depicted by their path through the
131: X-ray hardness-luminosity diagram (Homan \& Belloni, 2005).
132:
133: One object which has received a lot of attention in recent times,
134: being an out-lier in the SXRT population, is \xte\ (Chaty et
135: al. 2003). This system was
136: thoroughly observed through outburst to quiescence in X-rays, UV,
137: optical and radio (Hynes at al. 2003), and showed interesting breaks
138: and possible QPOs in its periodograms (Shahbaz et al. 2005). Being
139: high above the Galactic plane, it has been
140: suggested that \xte\ belongs to a new population of black hole
141: binaries, with the question of how they can appear in the halo still
142: open (McClintock et al. 2001). \xte\ is one of the few
143: systems which does not
144: follow the typical path for an SXRT through X-ray hardness/luminosity
145: space: it stayed in the low/hard state throughout. It is also the only
146: system for which a high signal-to.noise optical/X-ray
147: cross-correlation has been calculated, which showed an unexpected
148: ``precursor'' anti-correlation with optical leading X-rays (and a
149: stronger positive resonse; Kanbach et al. 2001).
150:
151: The SXRT \swift\ is an X-ray transient system which has been of
152: great interest recently following its outburst episode and
153: detailed observations with the {\em SWIFT} satellite. First discovered
154: by the {\em SWIFT}/BAT (Burst Alert Telescope; Palmer et al., 2005)
155: in 2005, pointed $\gamma$-ray, X-ray, UV,
156: optical and radio observations all detected a new bright source at
157: this location (Morris et al., 2005; Still et al., 2005; Halpern et
158: al., 2005; Fender et al., 2005). {\em Swift}/XRT (X-ray Telescope;
159: Burrows et al. 2005), and {\em RXTE} (Rossi X-ray Timing Explorer;
160: Jahoda et al. 1996)
161: observations detected the existence of
162: a strong 0.6\,Hz quasi-periodic oscillation (QPO; Morgan et al. 2005;
163: Ramadevi \& Seetha, 2005), persistent for some time
164: after the outburst episode. This QPO has so far only been seen in
165: X-ray observations. Zhang et al. (2007) presented the evolution of the
166: QPO frequency with time after the outburst, and its relationship to
167: X-ray hardness. With a high Galactic latitude, comparisons with
168: \xte\ promise to be very interesting.
169:
170: Summarizing the above works very briefly, after a fast initial rise in
171: X-ray flux, the source returned to its previous level, almost
172: undetectable by the RXTE All-sky Monitor, in a few months (rise time
173: of 4.5 days, decay of an exponential time-scale initially $\sim$32\,d,
174: later $\sim$100\,d, peak flux
175: $\sim$0.2\,crab in the RXTE/PCA or SWIFT/XRT band). Throughout,
176: it remained hard in X- and $\gamma$-rays (a rising spectrum in $\nu
177: f_\nu$ out to $\sim$200\,keV), suggesting a hot Comptonizing
178: corona (ADC). The optical counterpart also faded, but the fade slowed, and
179: it remains relatively bright and blue. H$_\alpha$ and He lines were
180: originally observed, double-peaked, but disappeared after the initial
181: outburst (Torres, 2005). The accretion disc inclination is thus
182: non-negligible, but neither is it extreme ($\geq80$\degr), since eclipses
183: are never observed. The H$_\alpha$ line was not particularly strong
184: (equivalent width $\sim3$\,\AA) but very broad
185: (FWHM$\sim$2000\,km\,s$^{-1}$). From the Na{\tt I} doublet and UV
186: observations, the
187: extinction equivalent to $N_H \sim 2\times 10^{21}$\,cm$^{-2}$
188: and a distance of $\sim 6$\,kpc have been estimated. The radio spectrum
189: was variable and consistent with a flat power law ($f_\nu \sim \nu^0$)
190: which lies underneath the optical points, if extrapolated (Fender et
191: al. 2005). This suggests that, initially at least, the optical
192: emission did not appear like a typical synchrotron spectrum.
193:
194: Cadolle Bel et al. (2007; CB07 from hereon) performed follow-up simultaneous
195: multi-wavelength observations. They identify \swift\ as a likely
196: Black-Hole Candidate, which stayed in a low/hard state throughout its
197: outburst and gradual fade. They also detected a QPO in their X-ray
198: data, but weaker, and at the somewhat lower frequency of 0.24\,Hz. Such a
199: reduction in the frequency is consistent with what would be expected
200: for the inner radius of an accretion disc, which expands during the
201: decaying lifetime of the outburst in the ADAF model. The INTEGRAL
202: count-rate during their
203: observation was constant at 43\,cts\,s$^{-1}$ ($\sim205$\,mCrab
204: between 20--320\,keV), with a hardness ratio
205: $HR=f_{20-40\,\textrm{keV}}/f_{40-80\,\textrm{keV}}\sim0.75$. They
206: also found a flat radio spectrum with fluxes of the order
207: 0.65\,mJy.
208:
209:
210: Miller et al. (2006a) observed \swift\ some months after the outburst
211: episode above with XMM-Newton and RXTE, when the source was assumed to have
212: reached quiescence. Their spectral modelling indicated
213: a prominent accretion disc was required by
214: the X-ray spectrum with high significance. This disc is cool ($kT =
215: 0.2$\,keV), and extends to very small radii near the last inner stable
216: orbit. This model would pose problems for models of low-level
217: accretion involving an ADAF region near the central source, and
218: complicate jet-formation scenarios. An ADAF model is commonly invoked
219: to explain the existence of a jet in the low/hard state and its
220: absence in the high/soft state, in systems where both states have been
221: observed (e.g., Fender, 2006).
222:
223:
224: We have conducted a comprehensive observational campaign of \swift,
225: including optical photomety in different filters on five successive
226: nights, with $\sim1$\,min time resolution, contemporary X-ray
227: observations, optical spectroscopy and high time-resolution optical
228: observations with ULTRACAM. Here we present the results of this
229: campaign, particularly timing analyses of the data-sets and
230: comparisons between them. In the companion paper, Durant et al.
231: (2008) analyze the cross-correlation of the X-ray and optical
232: light curves from the simultaneous high-speed observations, finding
233: suprisingly that the optical precedes the X-rays with a broad
234: anticorrelation peak, followed by a weak positive response for the
235: optical lagging the X-rays. The only contemporaneous radio
236: observation of \swift\ we are aware of were taken by Soleri et
237: al. (2008), who did not detect the source with the Westerbrook
238: Synthesis Radio Telescope in July 2007, establishing a 3$\sigma$
239: limiting flux of 1.1\,mJy at 5\,GHz and 8\,Ghz. Zurita et al. (2008)
240: also observed this system in the optical band, but over a much longer
241: time base-line (several months) and they discuss the long-term optical
242: brightness trend and newly discovered $\sim 3.2$\,hr superhump/orbital
243: period.
244:
245: In the following section we describe the various observations and
246: processing. In Section 3, we give the results of this, and further
247: analysis comparing the different data, including timing and spectral
248: analyses. In Section 4, we discuss the results and attempt to draw
249: physical conclusions from them.
250:
251: \section[]{Observations}
252: Table \ref{obslog} lists all the observations of \swift\ that are
253: analyzed here. Of the observations listed there, only the WHT/ISIS
254: spectrum was {\em not} obtained during the same week of 2007 (it is
255: from one year earlier), whereas the radio limits of Soleri et
256: al. (2008) were approximately contemporary with our observations.
257:
258: Further to our own observations below, we retrieved the Rossi All Sky
259: Monitor (AMS; Levine et al., 1996) daily average count rates for
260: \swift, as well as regular monitoring observations by the gamma-ray
261: satellite INTEGRAL's
262: wide-field lower energy-band imaging instrument ISGRI (Integral Soft Gamma-ray
263: Instrument; Lebrun et al. 2003). These are shown in Figure
264: \ref{ASM}. The ASM observes the
265: whole sky, and INTEGRAL surveys the Galactic Centre regularly, and all
266: interesting bright objects are automatically measured. These data are
267: publicly available (the reduction is not described here, see the
268: references above).
269:
270: \subsection{X-ray}
271:
272: \swift\ was observed for 53.6\,min each on 11 Jun 2007 and 13 Jun
273: 2007 with the {\em Rossi X-ray Timing Explorer} (RXTE).
274: RXTE comprises three
275: instruments: the Proportional Counting Array (PCA; Jahoda et al. 1996)
276: for soft-band pointed observations with large effective area, the
277: High-Energy X-ray Timing Experiment (HEXTE; Rothschild et al. 1998)
278: for energies up to 200\,keV and the All-Sky
279: Monitor (ASM, see above), with a very large effective field of view. None are
280: imaging instruments, but provide high temporal resolution. For the
281: first observation, only two of the PCA units were functional, and
282: in the second observation, three. We use only the PCA light curves in
283: our timing analysis, since the count rates are so much higher than the
284: high-energy HEXTE count rates. The latter we only use in the spectral
285: fitting.
286:
287: Events were processed through the standard pipeline for both the PCA
288: and HEXTE data-sets. Standard Good Time Intervals were applied and we
289: selected all events flagged as good for further
290: analysis. We generated light curves by binning the events on a regular
291: grid; uncertainties on each sample are dominated by photon counting
292: noise.
293:
294: When a fluxed spectrum is generated from the data, using the
295: instrument response functions, the HEXTE and PCA parts of the spectrum
296: do not appear to meet: there seems to be a 15\% discrepancy,
297: which corresponds to the effect of dead time. We corrected for this
298: and performed our spectral analysis on these data.
299:
300: \subsection{Optical Photometry}
301: We observed \swift\ with the 2.2\,m Nordic Optical Telescope (NOT; 2007
302: Jun 16, 17) and the 1.5\,m Mercator (2007 Jun 18, 19, 20) telescopes
303: at Roque de los Muchachos Observatory,
304: La Palma, Spain. In each case, we performed CCD-based imaging with
305: exposure times of 60\,s. With the NOT we acquired images in the $U$ and $B$
306: filters, and with the Mercator in the $BVR$ filters. Standards were
307: imaged with the NOT in the $U$ and $B$ filters, and the Mercator
308: photometry was calibrated via images of both the science field and
309: standards fields with the IAC80 telescope at the Teide Observatory,
310: Tenerife. The weather was generally good, especially for the NOT run,
311: with some variability and thin cloud during the Mercator run.
312:
313: Images were in every case first bias-subtracted, then flat-fielded
314: using drift images of the twilight sky. Note that the NOT/Alfosc
315: images were affected by unrepeatable electronic pick-up, and that the
316: photometry was therefore somewhat degraded. Photometry was measured in small
317: apertures relative to a bright local reference in the field. For this,
318: we used the ULTRACAM pipeline, as below with the ULTRACAM data. The
319: local reference was in each case calibrated relative to the Landolt
320: standard list (Landolt, 1992), correcting for a colour term for each filter
321: set. Having other stars in the field of similar brightness
322: to \swift\ enabled us to check that the analysis process was not
323: introducing any systematic signal into the light curves.
324: The uncertainty on each measurement is estimated from the photon
325: statistics of the counts within an aperture and the sky
326: background. Our measurements are source photon noise dominated.
327:
328: \subsection{Optical Spectroscopy}
329: We obtained spectra across the optical range with the
330: Intermediate dispersion Spectrograph and Imaging System
331: (ISIS\footnote{\tt http://www.ing.iac.es/Astronomy/observing/manuals/
332: html\_manuals/wht\_instr/isis\_hyper/isis\_hyper.html}) on the
333: 4.2\,m William Herschel Telescope (WHT) at the Observatorio Roque
334: de los Muchachos, La Palma, Spain. ISIS features twin spectrographs
335: optimized for the red and blue end of the optical, which can be used
336: simultaneously with the use of a dichroic in the main beam. We obtained
337: 1.72\,\AA/pixel mean dispersion in the blue arm, and 1.65\,\AA/pixel
338: in the red, with a mean resolution $\sim3$\,\AA\ throughout, under good
339: conditions. The two spectra were wavelength calibrated and extracted
340: using standard {\tt iraf} tools, and co-added into master spectra. In
341: order not to lose any resolution when co-adding, each spectrum was
342: first re-sampled to 0.5\,\AA.
343:
344: To flux and produce the final spectrum, we observed a spectral
345: standard at a similar airmass, and for both the object and the
346: standard spectrum, summed together the red and blue parts into a
347: single master spectrum (sampled at 0.5\,\AA). Since the dicroic
348: provides a sharp but finite cut-off, calculating the response function
349: for any one arm in the cut-off region is hard, but yet there are
350: enough photons in total at any given wavelength to find the overall
351: sensitivity. The final response curve we used to flux the data was a
352: heavily smoothed version of the ratio of the tabulated flux to the
353: measured counts for the standard.
354: These observations were done one year after the outburst of \swift, during
355: the faintest part of the ASM light curve since detection (Figure
356: \ref{ASM}).
357:
358: We also obtained a spectrum of \swift\ during our June observing
359: campaign with the Focal Reducer and low dispersion Spectrograph
360: (FORS2, Appenzeller et al. 1998) on the 8.2\,m Unit Telescope 1 (Antu)
361: of the VLT, Cerro Paranal, Chile. This observation was
362: simultaneous with one RXTE and one ULTRACAM
363: observation. The spectral scale was 3.17\,\AA/pixel, and resolution
364: $\sim7$\,\AA. Conditions were
365: mediocre, with variable cloud cover, transparency and seeing. Since
366: the whole spectral range is obtained in a single go, the problems
367: associated with matching two spectra (above) were not encountered
368: here, and the extraction was straight-forward.
369:
370: \subsection{ULTRACAM}
371: In addition, \swift\ was
372: also observed with ULTRACAM, mounted on the VLT/3 (Melipal) telescope on the
373: nights of 2007 Jun 12 and 17, for 1.3\,hr and 0.5\,hr
374: respectively. ULTRACAM is an instrument employing dichroic beam
375: splitters, frame-transfer CCDs and a GPS-based timing system in order
376: to be able to make simultaneous multi-wavelength optical light curves at very
377: high time resolution, up to 500\,Hz (Dhillon et al. 2007). We used
378: two small windows on each CCD (one for the
379: source of interest, one for a local standard), with exposure times of
380: 140\,ms for the 12th and 39\,ms for the 17th (and duty cycles of
381: 142\,ms and 41\,ms respectively). The reason for the
382: difference in exposure times was thin cloud on the first of the two
383: nights, giving similar signal-to-noise per image.
384:
385: The object was visible in every frame in the $r'$ and $g'$
386: bands, but not in the $u'$ band. For the latter, therefore, the source
387: could only be
388: detected by co-adding many images, resulting in reduced temporal
389: resolution but better signal-to-noise. On the first night, with
390: the poor conditions, the effect
391: was particularly strong, and we do not attempt to analyze these
392: data. On the 17th, however, it was possible to get reasonable
393: measurements from averages of every 50 images. Thus it was not possible
394: to search for high-frequency variability in these data, but timescales
395: $\geq$4\,s were accessible.
396:
397: Fluxes were extracted by aperture photometry with a variable aperture
398: size scaled to the FWHM of the reference star on each image. This
399: enables some optimization for signal-to-noise under variable
400: conditions. The optimal extraction method (Naylor, 1992) did not yield
401: appreciably different results, since that method is more applicable to the
402: faint, background-dominated regime.
403:
404: The night of the 2007 June 12 was badly affected by
405: transparency variations. In addition, the comparison star chosen was
406: of similar brightness to our source in $r'$ and significantly redder,
407: introducing additional uncertainty in $r'$ and even more in $g'$. This is
408: evidenced by the difference in the
409: $r'$-band and $g'$-band light curves. Only for the 17th can we give
410: reliable average magnitudes, based on the calibrated zero-points of
411: the instrument. These are on the SDSS photometric system
412: (systematic uncertainties here are $\pm$0.03--0.05\,mag, increasing to the
413: blue). By the transformations given in Jester et al.
414: (2005), the SDSS average magnitudes correspond to V$=$16.62(5),
415: B$=$16.92(7), consistent with the slow photometry above. The slow
416: photometry has a better absolute calibration.
417:
418: \begin{table*}[h!]
419: \begin{center}
420: \caption{Observation log of \swift}
421: \label{obslog}
422: \begin{tabular}{lccccc}
423: \hline
424: Date (UT) & Type & Instrument & Filter(s) & Duration (min) \\
425: \hline
426: 2006-06-17 & Spectrum & WHT/ISIS & red, blue &160\\
427: 2007-06-11 & X-ray & RXTE & & 54\\
428: 2007-06-13 & Photometry & ULTRACAM & $u'g'r'$ & 78\\
429: 2007-06-13 & X-ray & RXTE & & 54\\
430: 2007-06-13 & Spectrum & VLT1/FORS2 & & 3\\
431: 2007-06-16 & Photometry & NOT/Alfosc & $UB$ & 434\\
432: 2007-06-17 & Photometry & NOT/Alfosc & $U$ & 420\\
433: 2007-06-17 & Photometry & ULTRACAM & $u'g'r'$ & 30 \\
434: 2007-06-18 & Photometry & Mercator/Merope & $BV$ & 411\\
435: 2007-06-19 & Photometry & Mercator/Merope & $BVR$ & 290\\
436: 2007-06-20 & Photometry & Mercator/Merope & $BVR$ & 432\\
437: \hline
438: \end{tabular}
439: \end{center}
440: \end{table*}
441:
442: \section{Analysis and Results}
443:
444: \subsection{Long-term trend}
445: Figure \ref{ASM} shows the long-term luminosity trend of \swift\, as
446: seen in the X-ray and gamma-ray bands.
447:
448: It appears that, after the initial
449: outburst and fade, the source has been steadily increasing in flux to
450: a peak at the time of our observation, of the order of the flux in the
451: tail of the initial outburst. Note that on the left-hand extreme of
452: Figure \ref{ASM}, one can see the zero level for the default
453: extraction for this source. The ASM count rate at the time of the June
454: 2007 observations is clearly higher than this, near
455: 2\,cts\,s$^{-1}$. This corresponds to a flux of order $F\sim
456: 1\times10^{-9}$\,erg\,s$^{-1}$cm$^{-2}$ in the 1--10\,keV range
457: (estimated using WebPIMMS). Note further, that the INTEGRAL rate
458: ratios in the observations available were similar in each epoch, and
459: the light curve follows that of the ASM,
460: indicating that the high-energy portion of the emission changed little
461: in spectral slope/hardness. Unfortunately, observations closer in time
462: to our campaign are not yet publicly available.
463:
464: \begin{figure*}\includegraphics[width=0.8\hsize]{f1a.eps}
465: \includegraphics[width=0.8\hsize]{f1b.eps}
466: \caption{(upper) RXTE/ASM long-term light curve \swift. The vertical dotted
467: lines show the times of the observations by CB07, Miller et
468: al. (2006) and our RXTE pointed observations in this work, from left
469: to right.
470: (lower) INTEGRAL monitoring average count-rates on the same
471: time-scale. Open circles are counts in the 22--40\,keV range, crosses
472: in the 40--80\,keV range (left scale), and triangles with errors are
473: average count ratios (22--40/40--80, right scale).}\label{ASM}
474: \end{figure*}
475:
476: In the optical, as detailed in Zurita et al. (2008), the counterpart
477: also settled to a steady optical magnitude, which has periodic and
478: aperiodic variability superposed.
479:
480: \subsection{Orbital Modulation}\label{mcp}
481: The light curves from our photometric observations with the NOT and
482: Mercator telescopes are shown in Figure \ref{phot_lc}.
483:
484: \begin{figure*}
485: \begin{center}
486: \large{NOT}\\~\\
487: \includegraphics[width=0.45\hsize]{f2a.eps}
488: \includegraphics[width=0.45\hsize]{f2b.eps}\\
489: \large{Mercator}\\~\\
490: \includegraphics[width=0.45\hsize]{f2c.eps}
491: \includegraphics[width=0.45\hsize]{f2d.eps}
492: \includegraphics[width=0.45\hsize]{f2e.eps}
493: \caption{Optical light curves of \swift. The upper panels are from
494: the NOT, the lower from the Mercator telescope. The five panels
495: are subsequent nights 16--20 Jun 2007 (left to right, top to
496: bottom). { Vertical lines are estimates of the statistical error on
497: each point (but horizontal bars purely mark the point
498: centre).} }\label{phot_lc}
499: \end{center}
500: \end{figure*}
501:
502: In each night and each band ($UBVR$), the $\sim3.2$\,hr modulation
503: reported in the $R$ band by Zurita et al. (2008) is clearly apparent,
504: confirming their result. We are unable, however, to aproach the
505: accuracy of their period determination from our four nights'
506: data. Furthermore, simultaneous light curves in
507: each filter show remarkably similar shapes. (Note that the
508: observations here were not strictly simultaneous: they were made
509: through each filter in turn) This variabilty has a shape typical of
510: superhump modulations (the conclusion reached by Zurita et al), but we
511: cannot exclude eclipsing (of the disc) or emission from the irradiated
512: side of the donor star from these data alone. Clearly, the colours vary
513: very little through the modulation period. The overall brightness does
514: vary during each night and from one night to the next. Table
515: \ref{photvalues} lists the mean magnitudes and colours from each of
516: the five nights. The variability within each night is smooth and at
517: the $\sim10$\% level.
518:
519: \begin{table*}
520: \begin{center}
521: \caption{Slow photometry of \swift.}\label{photvalues}
522: \begin{tabular}{lcccc}
523: \hline
524: Date (start)& $<U>$ & $<B>$ & $<V>$ & $<R>$\\
525: \hline
526: 2007-06-16 & 16.43(15) & 17.05(9) \\
527: 2007-06-17 & 16.51(19) \\
528: 2007-06-18 & & 16.89(15) & 16.49(14)\\
529: 2007-06-19 & & 16.89(12) & 16.49(10) & 16.29(8)\\
530: 2007-06-20 & & 16.94(13) & 16.55(11) & 16.34(11)\\
531: \hline
532: \end{tabular}\\
533: All magnitudes are in the Johnson-Cousins system. Numbers in
534: parentheses indicate the spread in values.
535: \end{center}
536: \end{table*}
537:
538: Interestingly, our average magnitudes do not
539: match those of Zurita et al (2008), which were determined on
540: 2007 June 7, not far in time
541: from our own observations. We find that the object is typically
542: brighter by $\sim0.1$\,mag in each of the bands $BVR$ in our data.
543:
544: \subsubsection{Pulsed Fraction}
545:
546: \begin{figure*}
547: \begin{center}
548: \includegraphics[width=0.7\hsize]{f3.eps}
549: \caption{Spectral energy distribution of \swift\ in the optical
550: region. The upper points are the mean observed flux in the broad-band
551: filters, with uncertainties derived from the scatter between nights. The
552: lower points are the pulsed flux in the 3.2\,hr periodicity, with
553: uncertainties from the error on measuring the pulsed
554: fraction. The points are as measured, uncorrected for
555: reddening.}\label{pulse}
556: \end{center}
557: \end{figure*}
558:
559: The light curves presented also give information about the amount of
560: flux involved in the 3.2\,hr modulation. This can be measured directly
561: from the graphs from the peak-to-peak amplitude. In Figure \ref{pulse}
562: we plot both the total flux (as
563: an average across the observations) and modulated flux in the optical
564: region. Here the uncertainties in the pulsed flux come from the
565: scatter in points on the light curves rather than variability between
566: nights - a caveat for comparing the points. Although there is
567: night-to-night variability, it is clear that the pulsed flux increases
568: with energy, whereas the total flux is flat or turns over.
569: It appears that the pulsing component is hotter
570: than the DC component (or fast-varying component); { specifically, a
571: linear function fitted to the log-log points in Figure \ref{pulse}
572: yields slopes of -0.1(2) and 0.7(2) for the total and modulated
573: components respectively. } We can speculate that modulated emission
574: would dominate in the near-UV. Note that these
575: points have not been de-reddened in order not to add additional
576: uncertainty to the plot. De-reddening the pulsed flux points by
577: $A_V\sim 1$ would yield a spectrum which is roughly consistent
578: with a Wien slope. The dependence of the pulsed fraction on wavelength
579: rules out that this component originates from X-ray reprocessing.
580:
581:
582: \subsection{Spectrum}
583: \subsubsection{X-ray}
584: The spectrum was extracted from the valid photon events, and the
585: nominal background subtracted. We fit the resulting spectrum (using
586: {\tt Xspec}) in the range
587: 2--50\,keV (including both PCA and HEXTE observations) to an absorbed
588: power law, and find a statistically
589: satisfactory fit with $\Gamma=1.548\pm0.005$ ($N_\gamma \propto
590: E^{-\Gamma}$) and normalization
591: $A=0.1057\pm0.0011$\,ph\,keV$^{-1}$cm$^{-2}$s$^{-1}$ -- see Figure
592: \ref{Xspec}. A thermal disc component at low energies and a Gaussian feature at
593: an energy $E\sim6.2$\,keV are consistent with the data and modestly
594: improve the $\chi^2$ statistic of the fit, each with a significance of
595: $\sim2\sigma$, after taking into account the uncertainty on the column
596: density for the former. The broad line
597: might be consistent with a Fe K$_\alpha$ line with a rest-frame energy of
598: $E\approx6.4$\,keV.
599:
600: Specifically, using the hydrogen column found by CB07, we fitted the
601: X-ray spectrum with a model consisting of a power-law continuum as
602: above, summed with a disc thermal model ({\sc diskbb} in {\tt Xspec})
603: and Gaussian emission feature. The best-fit parameters we find are as
604: follows: disc temperature $kT = 0.19\pm0.03$\,keV,
605: normalization\footnote{defined in Xspec as
606: $\left[\frac{(R_{in}/\textrm{km})}{(D/10\,\textrm{kpc})}\right]^2$
607: where $R_{in}$ is an
608: effective inner radius, from Kubota et
609: al. (1998)} $N=4.8(6)\times10^4$, emission peak energy $E=6.15\pm0.17$\,keV,
610: normalization $A=2.63\pm1.3 \times
611: 10^{-4}$\,ph\,cm$^{-2}$s$^{-1}$. { The plotted points and fitting included only
612: photon-counting uncertainties and assume that the pipeline deals
613: correctly with the calibration. It should be noted, that the RXTE/PCA
614: calibration for the low energies of these putative detections is
615: somewhat uncertain, and further complicated by non-negligible
616: background contamination, which increases rapidly to lower energies
617: (Jahoda et al. 2006). The fit values and implied significance of the
618: components should therefore be read cautiously, particularly for the
619: thermal component.} We did, however, include the most recent
620: calibrations available, which fixed previously unknown problems with
621: the background estimation, such as correctly recording the time since
622: last passage through the South-Atlantic Anomaly \footnote{see {\tt
623: http://www.universe.nasa.gov/xrays/programs/rxte/
624: pca/doc/bkg/bkg-2007-saa/}}.
625:
626: We regard the detection of a disc component and of a broad
627: emission feature as suggestive, however the line would fit with a
628: gravitationally red-shifted, velocity-broadened Iron K line, as has
629: been seen for some other black hole accreting systems (Miller, 2007). For the
630: neutral Iron K line at $\approx 6.4$\,keV, the gravitational red-shift
631: would correspond to 3.6 Schwarzschild radii.
632:
633: Ramadevi \& Seetha (2007) had found strong evidence for a very soft thermal
634: component in the X-ray spectrum of \swift\ during the brighter
635: emission near the outburst of 2005. The
636: temperature of this component
637: was $kT\approx0.4$\,keV, whereas Miller et al. (2006a) report a much
638: cooler thermal component of 0.2\,keV at a later time, further after the
639: outburst. The latter was based on XMM-Newton EPIC and
640: grating spectra, which are much more sensitive at low energies than
641: RXTE. Our marginal detection of a thermal component is
642: consistent with the later temperature. A comparison of preferred fit
643: parameters by different authors (at different times since the
644: outburst) are shown in Table \ref{fits}.
645:
646: Note that Ramadevi \& Seetha (2007) reported finding an absorption
647: edge at $E\sim 7$\,keV. In their analysis, this improved the
648: quality of the fit from $\chi^2_{red}\approx5$ to 1.3. This may
649: explain the departure from a power law in our spectrum. The shape
650: around this energy does not look like an absorption edge feature,
651: however.
652:
653: \begin{table*}
654: \begin{center}
655: \caption{Summary of X-ray spectral fits and QPO detections by
656: different authors.\label{fits}}
657: \begin{tabular}{lcccc}
658: \hline
659: Parameter & Ramadevi \& Seetha & Miller et al. & Cadolle-Bel et al. &
660: This work\\
661: & (2007)$^{1}$ & (2006)$^{2}$ & (2007) \\
662: \hline
663: Satellite & RXTE & XMM & RXTE & RXTE \\
664: $T_{obs}$ & July-November & March & August & July\\
665: & 2005 & 2006 & 2005 & 2007 \\
666: $\Gamma$ & $1.76\pm0.014$ & 1.66$\pm$0.01 & &1.548$\pm$0.005\\
667: $kT_{BB}$ (keV) & $0.38\pm0.07$ & 0.21$\pm$0.02 &
668: & 0.19$\pm$0.03 \\
669: $R_{BB}$ $^3$ (km) & 100 & 2--6 & & 170\\
670: $kT_{seed}$ $^4$ & & 0.17$\pm$0.1 & 0.54$\pm0.06$ & \\
671: $\tau$ $^4$ & & 1.03$\pm$0.01 & 1.06$\pm$0.02 & \\
672: $\chi^2_{red}$ & 1.3 & 1.15 & 1.17 & 1.01 \\
673: $N_H$ (cm$^{-2}$) & $2.3\times10^{21}$ & $2.3\pm0.1\times10^{21}$ &
674: $2\times10^{21}$ & $2.3\pm0.2\times10^{21}$\\
675: $f_{QPO}$ (Hz) & 0.891$\pm$0.008 & & 0.241$\pm$0.006 & See \S\ref{pds}\\
676: $\delta f_{QPO}$ (Hz) & $\sim$0.2 & & 0.03$^{+0.02}_{-0.01}$ & \\
677: \%RMS & $\sim$23 & & 5.4$\pm1.8$ & \\
678: \hline
679: \end{tabular}\\
680: Note that each fit is for a different
681: combination of instrument, spectral range and fitted
682: model, and that the hydrogen column is not necessarily variable in
683: the fit. $\chi^2$ values are for the best fit in each case.\\
684: $^1$: Parameters at the start of the outburst.\\
685: $^2$: The power law plus black-body model, and the Comptomized
686: model parameters both given.\\
687: $^3$: At 8\,kpc.\\
688: $^4$: Comptomized models, seed photon energy and corona optical depth.
689: \end{center}
690: \end{table*}
691:
692: The total flux in the spectrum is
693: 1.6$\pm0.1\times$10$^{-9}$\,erg\,s$^{-1}$cm$^{-2}$ in the 2--20\,keV
694: range. This is fainter than the flux reported by CB07,
695: which was measured in the tail of the initial outburst, and
696: brighter than the flux reported by Miller et al. (2006), who performed
697: their observations some months after the outburst (the times of the
698: various follow-up observations are indicated in Figure \ref{ASM}, and
699: the fluxes measured match the ASM curve well).
700:
701: \begin{figure*}
702: \includegraphics[width=0.6\hsize,angle=270]{pgplot.ps}
703: \caption{X-ray spectrum of \swift\ from RXTE for the two
704: observations and two instruments: PCA (left) HEXTE (right). The solid line is
705: the best-fit model in Table \ref{fits}.
706: }\label{Xspec}
707: \end{figure*}
708:
709: \subsubsection{Optical}
710: The signal-to-noise ratio throughout the WHT spectrum (Figure \ref{opspec}) is
711: significantly better than for the later VLT/FORS2 spectrum, the latter
712: of which was taken in the same epoch as the rest of the
713: observations. Some of the features are apparent in both spectra, however, and
714: the equivalent widths for the most significant features is
715: given in Table \ref{lines}. Uncertainties are derived by measuring the
716: standard deviation of the flux in the continuum at either side of the
717: line in question. The continuum is assumed to be a straight line, and determined
718: from the data at either side of each spectral line. Assuming that the
719: continuum is a straight line and that the standard deviation is a
720: measure of the uncertainty on each point is clearly
721: not the case for such a undulating spectrum (in fact, it is an
722: over-estimate). Additional systematic
723: uncertainty due to this is not included in the errors presented in
724: Table \ref{lines}.
725:
726: { The broad-band photometry measurements in Table \ref{photvalues} are
727: equivalent to average fluxes in their respective band-passes (e.g.,
728: the zero points of Bessel, 1995). These fluxes appear to be
729: some 20\% above the continuum level of our spectroscopy.}
730: Note, however, that the flux scale shown in Figure \ref{opspec} is affected by
731: slit losses and poor transparency during the observations. This
732: accounts for the apparent discrepancy between the typical flux shown
733: here and the broad-band magnitudes in Table \ref{photvalues}, which
734: are more reliable. { The photometry values show beyond doubt that the
735: spectrum shown has the correct shape (i.e., it is not affected by
736: broad-band mis-calibration of the continuum) and has reddened
737: signifiacntly since the optical spectrum of CB07.}
738:
739: The Na-I interstellar absorption appears to be somewhat higher than
740: previously measured by CB07. Since this measurement is based on
741: relative fluxes (across the lines, compared to the continuum on either
742: side), and the intrument response functions do not change
743: significantly across the width of a narrow line in the centre of the
744: sensativity range, the statistical
745: uncertainties given should be a fair measure of the accuracy of the
746: EW measurements, affected only by the underlying spectrum and not by
747: the instrument calibration, resolution etc.
748: Here we refer only to the WHT/ISIS
749: equivalent width, which should be more reliable. This implies a higher
750: reddening to the object and therefore correspondingly higher distance;
751: see below. { Since the interstellar extinction should not have changed,
752: we infer that the extinction internal to the binary varies. The
753: difference in the equivalent width between the WHT and VLT spectra is
754: $\Delta EW=0.4\pm0.2$ }
755:
756: We can use the depths of the interstellar lines to estimate the total
757: extincting column to the source. Following the calculation in CB07,
758: $N_H = 0.25\times 5.8\times 10^{21}\times \textrm{EW}_{\textrm{Na\ I}} =
759: 2.45\times 10^{21}$\,cm$^{-2}$. The uncertainty on this from the measurement of
760: the equivalent width alone is $0.13 \times10^{21}$\,cm$^{-2}$, but one
761: must keep in mind the uncertainty in determining the continuum and of
762: the conversion factors. We estimate that our column density is
763: accurate to 10\%.
764:
765: From this we can in principle estimate distance to \swift\ by comparison with
766: field stars of known distance and extinction or H-II
767: clouds. Zurita et al's (2008) distance puts \swift\ well above the Galactic
768: plane, in the halo ($z>1$\,kpc). Unfortunately, at such a distance there are few
769: field stars, H{\sc II} regions or other fiducial distance indicators
770: to compare with. The bulk of the stars along the line of sight are
771: foreground Galactic Plane stars, so it is not possible to find the
772: main sequence or {\em red clump} at large distance in this
773: direction. The hydrogen column we estimate from absorption lines
774: above, larger than CB07's, is
775: consistent with the total Galactic extinction (Still et al. 2006), so
776: the distance $d\sim$7\,kpc estimated by Zurita et al. from the lower column,
777: is consistent with our analysis. { Note that there is no clear way to
778: determine how much of the extinction is internal to the binary
779: system. Hopefully, in a future quiescent state we would be able to
780: determine the donor type spectroscopically, and thus find the distance and
781: extinction independently. }
782:
783: Beyond this, all features remain surprisingly weak, with a notable
784: lack of absorption, similar to the results of CB07, who see only a
785: faint hint of H and He absorption. Early in the outburst, there had been
786: significant H$_\alpha$ and He emission, initially double-peaked
787: (Torres et al. 2005), but there is very little evidence of
788: this in our data. For a line of similar width to the interstellar ones
789: seen, we place a 99\% upper limit on the equivalent width of
790: $EW<0.28$\AA, whereas Torres et al. found $EW(H_\alpha)\sim3$\AA\ (no
791: uncertainty given, except S/N$\sim$30). Finally,
792: the spectrum still appears blue (as seen
793: also from the multi-colour photometry, Section \ref{mcp}), but less so
794: than at the peak of the outburst.
795:
796: The shape of the continuum in the VLT/FORS2 spectrum is qualitatively more
797: similar to the early spectrum in CB07 than to our intermediate
798: spectrum from WHT/ISIS. The blue part of the continuum appears to
799: have recovered somewhat, although the red has stayed rather similar
800: throughout. We note that, although much closer to a power-law shape,
801: the undulations or deviation froma straight line by $\sim5$\%,
802: apparent in the FORS spectrum are real and not an artefact
803: of the flux calibration (the calibration function was calculated for
804: each wavelength bin and smoothed, rather than attempting to fit with a
805: polynomial or other analytic function, which can easily introduce such
806: undulations).
807:
808: To check the evolution of the continuum shape, we performed further
809: optical spectroscopy with the ALFOSC intrument of the 2.5\,m Nordic
810: Optical Telescope (NOT), La Palma, in April 2008. The NOT/ALFOSC
811: spectrum, taken almost a year after our main observation campaign, is
812: also shown in Figure \ref{opspec}. Intriguingly, absorption
813: lines are even less obvious in this spectrum than before, and we do
814: not list them; even the strongest Na-D line seems to have decreased in
815: strength. Using different telescopes, instruments and resolutions, it
816: is hard to say whether these changes are real. If so, they hint at a
817: fair amount of extincting material within the system, the amount of
818: which evolves with time. The undulations in the continuum (i.e.,
819: departure from a power law) are still
820: clearly present, however.
821:
822: \begin{figure*}
823: \begin{center}
824: \includegraphics[width=\hsize]{f5.eps}
825: \caption{WHT/ISIS (upper),( VLT/FORS2 (middle) and NOT/ALFOSC
826: (lower) optical spectra of \swift. A list of the equivalent widths
827: of features in
828: these spectra is given in Table \ref{lines}. The sharpest lines
829: are marked, and atmospheric telluric absorption bands, which have
830: been removed, are marked with ``T''.}\label{opspec}
831: \end{center}
832: \end{figure*}
833:
834: \begin{table}
835: \begin{center}
836: \caption{Optical spectral lines in \swift. All equivalent widths are
837: in absorption. DIL stands for diffuse
838: interstellar line (of unknown, PAH or metallic origin). Limits are
839: at 95\% significance.}
840: \label{lines}
841: \begin{tabular}{lccc}
842: \hline
843: Element & $\lambda$ (\AA) & \multicolumn{2}{c}{Equivalent Width
844: (\AA)}\\
845: & & WHT/ISIS & VLT/FORS2 \\
846: \hline
847: Ca & 3934 & 0.38$\pm$0.02 &<0.3\\
848: Ca & 3969 & 0.43$\pm$0.02 &<0.3\\
849: DIL & 4430 & 0.66$\pm$0.03 & 0.86$\pm$0.10\\
850: DIL &4885 & 1.00$\pm$0.06 & 1.15$\pm$0.11\\
851: DIL & 5780 & 1.09$\pm$0.05 &<0.6\\
852: DIL & 5797 & 0.222$\pm$0.007&<0.5\\
853: Na & 5892 & 1.69$\pm$0.09 & 2.1$\pm$0.2 \\
854: Ca & 6283 & 0.77$\pm$0.05&0.91$\pm$0.05\\
855: DIL &6614&1.20$\pm$0.11&<0.5\\
856: \hline
857: \end{tabular}
858: \end{center}
859: \end{table}
860:
861:
862: Slight undulations or humps are apparent in each of the spectra to a
863: greater or lesser degree. These are on the scale of a few percent in
864: flux and of order $\sim$500\AA\ broad in Figure \ref{opspec},
865: compared to a power law (straight line in these plots); such features
866: have now appeared in each of the four optical spectra since \swift's
867: discovery. We must note that H and He emission lines {\em
868: were} seen in the initial phases of \swift's outburst (Torres,
869: 2005), so there is no lack of hydrogen or helium in the
870: system.
871:
872: {The existance of broad humps throughout the optical spectral range is
873: suggestive of
874: cyclotron emission, but offers
875: little clear evidence; in theory they could be extremely broadened
876: lines or combinations of lines. The lack of sharp lines argues strongly that
877: the optical emission is not thermal emission from a dense disc
878: over-layed by less dense material, as this would result in absorption
879: and/or emission lines, depending on the dominant temperature. A pure
880: black-body or multi-colour dics would have smooth power laws either
881: side of a single peak. Likewise, synchrotron may well exist in the
882: spectrum, but is generally thought to have a $F_\nu\propto\nu^0$ flat
883: spectrum.
884: Charateristic strong cyclotrons humps are not seen, but this
885: is not a similar situation to the more obvious cyclotron emission
886: in the case of Schwope et al. (2003): that is for a
887: polar system, where the emission region has a well-defined magnetic
888: field strength. In our case there could be combined emission
889: from various active patches with various field strengths throughout
890: the disc, not necessarily resulting in a series of clear, well-defined
891: humps.
892:
893: Thus although there is no clear evidence for cyclotron
894: emission, the spectra do not look like what might be expected from
895: other typical emission mechanisms, so cyclotron may be a successful
896: description. }
897:
898:
899: \subsubsection{SED}
900: In Figure \ref{SED} we show the spectral energy distribution of
901: \swift\ in our observations, compared to the comprehensive
902: multi-wavelength campaign of CB07 over a year earlier. We have
903: included limits in the radio from Soleri et al (2008, taken two weeks
904: after our observations), and
905: de-reddened the optical with $A_V=1.05$, the value used by CB07 (so
906: that the comparison is fair).
907:
908: We find the overall
909: luminosity has dropped by a factor $\sim$3, but that the X-ray spectrum has
910: qualitatively the same shape as before (a hard power law). The optical
911: is different: there appears to be a smooth break around the blue
912: such that the V-band flux is significantly fainter, but R and I are
913: consistent, and the slope is redder. One thing is clear: the
914: X-rays and optical require separate emission components, and if the
915: radio is at similar levels to before or lower by a similar factor as
916: the X-rays, then it too requires a separate emission
917: component: synchrotron emission from a jet cannot be dominant in the
918: optical region, if it has a typical flat $F_\nu$
919: spectrum. Intriguingly, the black body component suggested in the
920: X-ray spectrum (and not present in CB07) may connect the softest
921: X-rays with the optical. This is merely speculation, however, as the
922: optical spectrum itself does not appear black-body-like.
923:
924: { Any radio emission, if synchrotron in origin (such as the flat $F_\nu$
925: spectrum in CB07) would be too weak to account for any significant
926: fraction of the optical emission. Synchrotron does not, however, have
927: to be flat: emission from a single optically-thick clump rises as
928: $\nu^3$. Such a spectrum has not, to our knowledge, been seen thus far,
929: although de-convolving the various mechanism possible in the optical
930: is tricky.}
931:
932: \begin{figure*}
933: \begin{center}
934: \includegraphics[width=0.75\hsize]{f6.eps}
935: \caption{Spectral energy distribution of \swift\ around the time of
936: our observations (July 2007) from the radio to gamma rays. The radio
937: points are upper limits. The grey points
938: are those of Cadolle-Bel at al. (2007; reproduced by kind permission)
939: the black points are from our
940: observations, black triangles are $3\sigma$ upper limits from Soleri
941: et al (2008). We have only included X-ray points with good
942: signal-to-noise, and de-reddened the optical by $A_V=1.05$ for a fair
943: comparison with Cadolle-Bel et al.'s points. }\label{SED}
944: \end{center}
945: \end{figure*}
946:
947: \subsection{Timing}
948: We produced light-curves for the X-ray observations of \swift. We used
949: only PCA rates, since the HEXTE
950: is far less sensitive. Note that there were three Proportional
951: Countrer Units (PCUs) active in the
952: second observation and only two in the first. Neither observation
953: shows any general or long time-scale
954: trend in flux.
955:
956: We also produced optical light curves, for further analysis.
957: For the optical light curve on the 17th, the
958: more stable of the two nights, rapid variability is seen, on the order
959: $\sim20$\,s, of a similar magnitude to the 3.2\,hr modulation
960: above. The typical uncertainty on each measurement is 0.03\,mag in
961: the $r'$ band, and 0.05\,mag in the $g'$ and 0.08\,mag in the $u'$ band
962: throughout the observation window.
963:
964: \subsubsection{Power Density Spectra}\label{pds}
965:
966: A comparatively high fraction of flux is involved in short-term
967: ($T<1$\,min) variability in both X-ray and optical --
968: $>$10\%, similar or larger than the orbital-like modulation above.
969: Figures \ref{pgram} to \ref{pgram4} show the power spectra for all the
970: light curves. Note that, for the optical, the
971: Lomb-Scargle method is required, because the sampling is not strictly
972: regular, and some points have null values. A Fourier analysis of these
973: data do give indistinguishable results, if one assumes regular
974: sampling. White noise can be calculated by assuming a standard
975: deviation equal to the Poisson noise
976: ($\sqrt(N)$) on each sample, or obtained by fitting a function to the
977: power spectra which includes a constant term.
978: Fitting the power spectra and calculating from the known
979: count rates give very similar values for the white noise.
980:
981: \begin{figure*}
982: \includegraphics[width=0.6\hsize]{f9e.eps}
983: \includegraphics[width=0.6\hsize]{f9f.eps}
984: \caption{Power spectra of the
985: RXTE/PCA observations of \swift, 2007 June 11 (top) and 13 (bottom). The
986: power spectra have been rebinned and the best fit over plotted
987: (solid line) and its components (dotted lines, see Table
988: \ref{pgramfits}).}\label{pgram1}\label{pgram}
989: \end{figure*}
990:
991: \begin{figure*}
992: \includegraphics[width=0.60\hsize]{f9a.eps}
993: \includegraphics[width=0.60\hsize]{f9b.eps}
994: \caption{Power spectra of the ULTRACAM
995: observations of \swift from 2007 August 13, r' (top) and g'
996: (bottom), displayed as in Figure \ref{pgram1}}\label{pgram2}
997: \end{figure*}
998:
999: \begin{figure*}
1000: \includegraphics[width=0.60\hsize]{f9c.eps}
1001: \includegraphics[width=0.60\hsize]{f9d.eps}
1002: \caption{Power spectra of the ULTRACAM
1003: observations of \swift from 2007 August 18, r' (top) and g'
1004: (bottom), displayed as in Figure \ref{pgram1}}\label{pgram3}
1005: \end{figure*}
1006:
1007: \begin{figure*}
1008: \includegraphics[width=0.60\hsize]{f10.eps}
1009: \caption{Power spectrum of the ULTRACAM
1010: observations of \swift from 2007 August 18, u',
1011: displayed as in Figure \ref{pgram1} with one fitted
1012: Lorentzian.}\label{pgram4}
1013: \end{figure*}
1014:
1015: \begin{table*}
1016: \begin{center}
1017: \caption{Power spectrum best-fit parameters. The power law (PL) exponent
1018: refers only to the high-frequency ($>$0.1\,Hz) part of the
1019: power spectra, whereas the rest of the parameters are for double
1020: zero-centred Lorentzians (see text).} \label{pgramfits}
1021: \begin{tabular}{lccccccc}
1022: \hline
1023: Parameter & \multicolumn{2}{c}{RXTE/PCA} &
1024: \multicolumn{4}{c}{ULTRACAM}\\
1025: & 2007 June 11& 2007 June 13& \multicolumn{2}{c}{2007 June 13} &
1026: \multicolumn{3}{c}{2007~~ June~~ 18} \\
1027: & & & $r'$ & $g'$ & $r'$ & $g'$ & $u'$~$^1$\\
1028: \hline
1029: PL exponent & 1.20 & 1.18 & 0.95 & 0.90 & 0.92 & 0.89 & 1.7\\
1030: \hline
1031: Width 1 (Hz) & 0.102(9) & 0.095(8) & 0.040(4) & 0.033(3) &0.049(6) & 0.051(5) &
1032: 0.087(9)\\
1033: \%RMS & 12.0(7)& 13.1(7) & 9.70(8) & 37.7(3) & 4.00(5) & 15.8(2) &
1034: 4.8(3) \\
1035: Width 2 (Hz) & 1.7(2) & 1.5(2) & 0.9(2) & 1.5(2) & 1.3(2) & 1.3(3) & \ldots\\
1036: \%RMS & 9.3(8) & 9.1(7) & 6.15(15) & 32.2(5) & 2.55(5) &10.7(3)& \ldots\\
1037: QPO? (Hz) &\ldots &\ldots & 0.0789(15) & 0.0791(16) & 0.079(2) & 0.074(12) &
1038: 0.08(3) \\
1039: \%RMS & $<$3 & $<$3 & 4.4(10) & 11(2)& 1.5(6) & 7(2) & $<$10 \\
1040: $\chi^2/dof$ & 145/123 & 116/123 & 136/85 & 101/85 & 92/95 & 94/95 & 55/42\\
1041: \hline
1042: \end{tabular}~\\
1043: Numbers in perentheses are 1-sigma uncertainties in the last digit,
1044: 95\% confidence for limits.\\
1045: $^1$~Due to the smaller number of points available, the fitting
1046: function only included one Lorentzian. \\
1047: \end{center}
1048: \end{table*}
1049:
1050: { Qualitatively, one can see a number of features common to the
1051: power spectra: a power law-like decay at high frequencies and a break
1052: around 0.07\,Hz
1053: becoming flat in the 0.01\,Hz range. Some possible QPOs
1054: can be seen at around 0.08\,Hz in the optical power spectra but not in
1055: the X-ray ones, but with significance only at the
1056: 2--5$\sigma$ level. The optical power spectra are all similar to
1057: one-another in this respect also. }
1058:
1059: For the $u'$ band, we obtained a light curve by co-adding
1060: groups of 50 images, and analyzing these. The temporal resolution was
1061: thus degraded (to about 2\,s), with still significant scatter in the
1062: measurements. This accounts for the different appearence of the $u'$
1063: band power spectrum in Figure \ref{pgram4} (the $u'$-band power spectrum
1064: is the one with the least points).
1065: Many of the same features can be seen in the power spectrum: red noise
1066: power law, break, excess in the 0.01--0.1\,Hz region and a flat power
1067: spectrum at low frequencies. Even a QPO is possibly seen, but not
1068: statistically significant.
1069:
1070: By visual inspection, we find that the power spectrum continua are each
1071: well-fitted by zero-centred Lorentzian functions (as used by CB07, for
1072: example). We have, therefore, fitted each power spectrum accordingly, and
1073: the result of these fits are shown in Table \ref{pgramfits}. Also
1074: shown are what one would find assuming a power-law noise function for
1075: the high-frequency ($>$0.1\,Hz) part of each power spectrum. These numbers are
1076: consistent with {\em flickering} (superposition of discreet,
1077: stochastic flares of various heights and durations), for which one
1078: expects a power-law exponent of 1 (Bruch, 1992). The flattening of the
1079: power spectrum towards lower fequencies does not imply that this
1080: considerable power cannot be produced by micro-flares, but implies
1081: that the micro-flares cannot have arbritarily long durations.
1082:
1083: { Uttley \& McHardy (2001) give arguments against a Lorentzian and/or
1084: power-law power spectrum necessarily being produced by
1085: flickering; specifically, if the average flux and variance of sections
1086: of the light-curve are linearly correlated, without the line passing
1087: through the origin. This is indeed to case here, and implies that
1088: although flickers can occur on a wide range of time-scales as measured
1089: by the power spectra, there is likely an extra component of constant or
1090: very small RMS, perhaps the process which feeds the larger amplitude
1091: variability (e.g., Merloni \& Fabian, 2001).
1092:
1093: Although the power spectra all look rather similar, the numbers
1094: presented in Table \ref{pgramfits} are not. If the ideas of Uttley \&
1095: McHardy are correct, then there seems to be a variable competition
1096: between the variable component and the low RMS/constant component
1097: which depents both on time and spectral range. For the simultanous r'
1098: and g' numbers, the variability is consistently higher for the shorter
1099: wavelength, yet we see generally more variability in the optical than
1100: in the X-rays (where, as already stated, we do not find significant
1101: QPOs). Any micro-flares must thus be rather blue, but damped in their
1102: effect on the X-ray emission.}
1103:
1104: \subsubsection{Auto-correlation functions}
1105:
1106: Figure \ref{autocorr} shows the auto-correlation functions for the
1107: fast timing observations of \swift. This is a measure of how well each
1108: light curve is correlated with itself as a function of time
1109: difference, and therefore of the timescales dominant in the
1110: variability. In this sense, it complements the power spectrum, viewing
1111: the time-series from the point of view of individual events rather
1112: than coherent periodic signals. All the curves are well
1113: approximated by Lorentzian functions. The X-ray autocorrelation
1114: functions are very
1115: narrow (FWHM$\sim 2$\,s), implying that the variability seen at
1116: longer time-scales in the power spectra above are not very
1117: coherent. The wings of the auto-correlation do, however, seem to
1118: extend to large values, with continued structure. Note that the
1119: apparent flat tops of the auto-correlations is due to excising the
1120: zero-lag value, which is contaminated by white noise. After this, the
1121: functions have simply been normalized to 1.
1122:
1123: \begin{figure*}\begin{center}
1124: \includegraphics[width=0.45\hsize]{f11a.eps}~\\
1125: \includegraphics[width=0.45\hsize]{f11b.eps}
1126: \caption{Auto-correlation functions for the different
1127: light curves. The top panel shows the two RXTE observations, the
1128: bottom panel the optical for the two nights, with the
1129: 2007 June 13 data shifted vertically by $+$1. Each
1130: function is normalized to 1 at the peak, and the central point at
1131: $\delta t=0$ has been omitted, as it is strongly affected by white
1132: noise. The $u'$ curve
1133: represents a light curve of much lower time-resolution (by a factor
1134: of 50), and its apparently enhanced width is a result of
1135: this.}\label{autocorr}
1136: \end{center}
1137: \end{figure*}
1138:
1139: The optical auto-correlations in Figure \ref{autocorr} are all
1140: generally broader than their X-ray counterparts (FWHM$\sim$4\,s). One
1141: would normally expect this from a consideration of the energy
1142: scales: higher-energy emission is produced in an optically thin
1143: medium, and can
1144: escape easily. The $u'$ curve is consistent with the others from
1145: the night of the 17th, given its lower sampling rate. The optical
1146: auto-correlation functions also show some structure at larger lags,
1147: which we haven't shown as it would make the central peaks hard to
1148: see.
1149:
1150: In Durant et al. (2008) we present the cross-correlation functions
1151: between various energy bands of the RXTE/PCU and the ULTRACAM
1152: observations. We find that there is a strong anti-correlation with the
1153: optical arriving earlier than the X-rays on timescales of 1--10\,s,
1154: followed by a
1155: much weaker positive response after $\delta t=0$, for the softest X-ray
1156: energies. This is a similar result to the ``precursor'' signal in the
1157: cross-correlation function derived for \xte\ by Kanbach et al (2001).
1158: For medium energies, the correlation function is similar but
1159: narrower, and we find no correlation for high X-ray energies (where
1160: noise dominates). Please refer to this paper for further details.
1161:
1162: \subsubsection{Dynamic analysis}
1163:
1164: We produced dynamic power spectra
1165: (based on successive small sections of each light curve)
1166: from our rapid timing observations.
1167:
1168: At first glance, the dynamic power spectra show the same information
1169: as the average power spectra above: power concentrated in the
1170: 0.01--0.1\,Hz range, with no abrupt changes thoughout the observation
1171: windows. The two sets
1172: of optical dynamis power spectra taken simultanously look identical. For the
1173: X-ray, the fluctuation with time is
1174: totally consistent with noise. For the optical, however, there is a
1175: hint of structure, but again nothing that can be significantly
1176: distiguished from noise.
1177:
1178:
1179: \section{Discussion}
1180:
1181: A summary of our results:-
1182: \swift\ has not returned to a quiescent state, either in X-rays
1183: (ASM data), gamma-rays (INTEGRAL data) or optical; rather, it seems to
1184: have re-brightened somewhat, compared to one year before our
1185: observations. We do not know the
1186: current state of radio emission, only the upper limits of Soleri et
1187: al. (2008). The optical light curves show superhump-like modulations
1188: with a period of 3.2\,hr, increasingly significant towards the blue,
1189: but not detected in X-rays. The X-ray spectrum is well-described by a
1190: single power law of spectral index $\alpha=1.54$ from 2--60\,keV, with
1191: hints of a possible soft disc component and an emission line near
1192: 6\,keV. The optical spectrum shows no sharp features beyond
1193: interstellar ones, only small deviations from a power law.
1194: The SED shows a need for separate components for each wave-band,
1195: in particular, that the optical cannot be a continuation of a flat
1196: synchrotron-dominated spectrum.
1197: In contrast to previous observations, we find no significant QPOs in
1198: power spectra of the X-ray light curves, but we do find $\sim$0.08\,Hz
1199: QPOs in the optical; aside from these, all power spectra are
1200: well-described by two zero-centred Lorentzian functions, with widths
1201: $\sim$0.05\,Hz and $\sim$1.5\,Hz. Autocorrelation functions show a
1202: narrower peak ($\sim$2\,s) for the X-ray emission than the optical
1203: ($\sim3$--4\,s), which are consistent with one-another for all the
1204: $u'g'r'$ bands. The cross-correlation functions (Durant et al. 2008)
1205: show the optical leading by 1--10\,s and strongly anti-correlated with
1206: the X-rays, for the lower energies of the RXTE
1207: range. Dynamic power spectra show features which may be short-lived
1208: QPOs in the 0.02-0.1\,Hz region, more poorly defined in the
1209: X-rays than in the optical.
1210:
1211: From the work of CB07 and Zurita et al. (2008), we suspect that the
1212: system contains a stellar-mass black hole and M2V-type companion
1213: (which has an undetectable contribution to the current total optical
1214: luminosity of the binary), and is
1215: located at a distance of several kpc, significantly above the Galactic
1216: Plane (in the {\em halo}, since its height of order $>1$\,kpc is much
1217: larger than the disc scale height). Our measurement of interstellar
1218: absorption, particularly of Na, strongly supports a large distance but
1219: also suggests an unknown amount of absorbtion internal to the system.
1220:
1221: \subsection{Evolution}
1222: The optical evolution of this source since its 2005 outburst and
1223: discovery is more completely analyzed in Zurita et al. (2008). It is
1224: clear that the compact object in the system, probably a black hole, is
1225: continuing to accrete material, which is powering the observed
1226: luminosity. The system was not observed to leave its low/hard state,
1227: indicating that some low density/high energy region continued to exist
1228: throughout the outburst phase and afterwards. The X-ray spectrum has
1229: changed very little, except for the bolometric luminosity.
1230:
1231: There is apparent evolution in the overall shape of the optical
1232: spectrum between CB07, WHT/ISIS and VLT/FORS2 representing two years
1233: of the system's life since outburst. This is also seen in
1234: the broad-band
1235: photometry values in Table \ref{photvalues} compared to CB07 - these
1236: are a more accurate measure of the change of spectral slope. The
1237: wiggles in the WHT/ISIS spectrum are not present either earlier (CB07)
1238: or later (out VLT/FORS and NOT/ALFOSC) spectra.
1239:
1240: The smooth continuum and the lack of
1241: lines in maintained throughout, although hydrogen (H$_\alpha$) emission
1242: was seen immediately after the outburst (Torres et al, 2005). The red
1243: part of the spectrum changes very little, and indeed the R-band
1244: magnitude has remained remarkably constant (see Zurita et
1245: al. 2008). The blue part of the spectrum has changed, however: the
1246: large peak at $\sim$4000\AA\, dip at $\sim$4500\AA\ and second peak at
1247: $\sim$5000\AA\ seen in CB07 were not evident in 2006, but had recovered
1248: to some extent by 2007. It would appear that the process which had
1249: caused the enhanced emission during the outburst flare is continuing
1250: at a lower level, and that this had maintained the optical emission
1251: through the two-year period. Ongoing activity is also suggested by the
1252: ASM and INTEGRAL light curves.
1253:
1254: In terms of timing, the gross power spectrum has remained qualitatively
1255: similar: flat at low frequencies and flickering at high frequencies. A
1256: QPO was, however, initially detected at 0.6\,Hz by Morgan et
1257: al. (2005) and later at 0.4\,Hz by Ramadevi \& Seetha (2005); an
1258: evolution that was tracked in many RXTE observations by Zhang et al. (2007);
1259: in this work we see such a QPO, relatively weak, but only in the optical
1260: data and not in the X-rays, and at a much lower frequency.
1261:
1262: The slowing of characteristic QPO frequencies and time-scales has been
1263: seen for other transient systems (McClintock
1264: \& Remillard, 2006; Shahbaz et al. 2005, 2003), and is commonly taken to
1265: indicate an expanding inner radius of the accretion disc, inside of
1266: which a low density ADAF region forms. The linking of such a QPO
1267: frequency with Keplarian periods may be misleading, however: it is not
1268: obvious how a rotation rate translates into luminosity variations, and
1269: why there would exist preferred orbits within the disc. If for this
1270: system a disc existed at small radii well after the outburst (as
1271: suggested by Miller et al. 2006a), then the frequency characterised by
1272: the QPOs seen cannot have a Keplerian interpretation.
1273:
1274: More simple
1275: origins for variability might be disc-corona-jet interaction,
1276: turbulence timescales or magnetic field production/migration
1277: timescales; these might depend on factors such as accretion rate,
1278: magnetic/particle energy balance, jet efficiency, disc density profile
1279: etc. Why these in turn should evolve raises further questions;
1280: recurrent outbursting episodes suggested hysteresis in X-ray
1281: transients (e.g., Malzac, 2007), but the maintenance of a low/hard
1282: state in this case would imply only a small change in the mass flow
1283: rate. The re-brightening of the source since 2006, and its associated
1284: bluer optical colours point to somewhat increasing mass flow from 2006
1285: to 2007.
1286:
1287: \subsection{Emission mechanisms}
1288: The X-ray spectrum is typical of Comptonized emission from a
1289: population of energetic particles. This in turn requires a source of
1290: lower energy photons to scatter (possibly seen at the lowest edge of
1291: our X-ray spectrum, and claimed by Miller at al. 2006, and Ramadevi \&
1292: Seetha, 2007) and a replenishable energy reservoir in the
1293: particles. At the same time, we suggested the possibility of a
1294: contribution of cyclotron emission in the optical, requiring
1295: significant energy content in the magnetic field. Note that
1296: Beloborodov (1999) showed that emission from
1297: magnetically driven clouds moving away from the disc would not produce
1298: significant re-processing in cooler material, if the bulk motion was
1299: mildly relativistic.
1300: { We cannot exclude either a static optically
1301: thick syclotron emitting region nor multiple blackbodies where
1302: particular temperatures are favoured. From the cross-correlation, we
1303: can discount simple reprocessing being an important factor in the
1304: optical emission.}
1305:
1306: The case is further complicated by the {\em superhump}
1307: contribution. This accounts for about 10\% of the variation in the
1308: optical region, but
1309: increasing towards the UV. If it is indeed due to a superhump process
1310: (i.e., either tidal resonance between the disc and orbital periods, or
1311: changing area of the disc; Haswell et al. 2001), then the emission is
1312: thermal and very hot with a black-body-like peak blueward of B. We do not
1313: see any lines from this emission, but we do not have good spectral
1314: sensitivity shortward of 400\,nm. The 2006 WHT spectrum showed less of
1315: a blue component. This could also be the reason that no QPO
1316: was seen in the $u'$-band power spectrum - thermal emission may already be
1317: dominating at these energies.
1318:
1319: Finally there is the case of jet emission. Malzac et al. (2007)
1320: presented a model for how jet and disc emission could be coupled
1321: in a black hole system such as this. Jet radio emission was certainly seen
1322: earlier, closer to the onset of the outburst (CB07; Fender et
1323: al. 2005), and Soleri et al's limits are not stringent enough to
1324: exclude even a stronger jet than before. Jet/synchrotron emission
1325: cannot, however, be significant in the optical.
1326:
1327: \subsection{Dynamic behaviour}
1328: In Durant et al. (2008) we suggest that the cross-correlation function
1329: implies that the
1330: optically emitting region was driving rather than responding to the
1331: higher-energy emission. Here we add to this that the optical
1332: auto-correlation functions are {\em broader} than the X-rays, so it
1333: would appear that a relatively slowly building process in the optical
1334: leads to a faster and anti-correlated X-ray response. Furthermore, the
1335: power spectra show that this process occurs as a distribution of small
1336: flares (flickering), with a break at a characteristic time-scale
1337: corresponding to a period of $\sim20$\,s.
1338:
1339:
1340: Although Malzac et al. (2007) produced a model considering the
1341: dynamic interaction of a jet and accretion disc, in an effort to
1342: explain the optical/X-ray cross-correlation function for \xte, they
1343: explicitly consider only optical emission dominated by synchrotron
1344: emission, with some additional X-ray re-processing by dense disc
1345: material. In our case, the optical is not a simple continuation of a
1346: synchrotron spectrum int he radio (which extrapolates to below the
1347: optical emission); nevertheless, the idea of a
1348: magnetic energy reservoir may still be valid, as it can produce the
1349: types of dynamic behaviour and feedback observed here. We would be
1350: interested to see if their model has a parameter space to match the
1351: details given here.
1352:
1353: \subsection{Comparisons}
1354:
1355: It is interesting to note, that \swift\ is the highest Galactic
1356: latitude SXRT after \xte. The latter object shows some of the same
1357: characteristics, which set it apart from the bulk of the SXRT
1358: population: short-period superhump/orbital modulation of 4.1\,hr,
1359: a persistent low/hard state, similar power spectra and unusually bright
1360: optical emission (Hynes et al., 2003; Shahbaz et al. 2005). The X-ray/optical
1361: cross-correlation function for \xte\ was the first to unambiguously
1362: show an optical-leading anti-correlation component (``precognition
1363: peak'', in some sources), although a positive, standard, optical lag
1364: signal was dominant (Kanbach et al. 2001).
1365:
1366: Unlike \swift, \xte\ did settle into quiescence.
1367: In this state \xte\ also showed a power spectrum which could be
1368: described as a power law with break or power law
1369: plus QPO. One interpretation of this is to connect this characteristic
1370: frequency with the inner edge of the
1371: accretion disc: asthe disc-ADAF interface increases in radius,
1372: the system decreased in luminosity (Shahbaz et al., 2005). Can the
1373: changes of state really be explained by variable accretion rate alone?
1374:
1375: A leading anticorrelation was seen once for \gx\ in the past (Motch et
1376: al. 1983). Since this initial and unconfirmed measurement, based on a
1377: very short observation window, \gx\ has changed markedly. In
1378: particular, Gandhi et al. (2008) find the cross-correlation is very
1379: different and weak, in a fainter state, although optical and X-ray
1380: spectral characteristics have not changed much.
1381:
1382: Assuming that the threshold for an X-ray binary to leave the Low-Hard
1383: State (where the X-ray spectrum is predominantly a power law, rather than
1384: thermal) is 1 percent of the Eddington luminosity (McClintock
1385: \& Remillard, 2006), we
1386: can place a limit on the black hole mass in this system, under several
1387: assumptions. If the bolometric luminosity can be scaled from CB07 to
1388: the peak observed luminosity at the time of outburst, i.e., if the
1389: spectrum remained the same throughout the process, then for a distance
1390: $d$\,kpc we can place the limit
1391: \begin{eqnarray*}
1392: 1\%\times
1393: L_{Edd}&<&6\times10^{37}\left(\frac{d}{6\,\textrm{kpc}}\right)^2\times
1394: 3.5 \textrm{\,erg\,cm$^{-2}$s$^{-1}$}\\
1395: M &<& 6.4M_\odot\left(\frac{d}{6\,\textrm{kpc}}\right)^2
1396: \end{eqnarray*}
1397: Zurita et al. (2008) suggest that a stellar-mass black hole
1398: ($M\sim3M_\odot$) and M-type star adequately fit the orbital period and
1399: quiescent (i.e., pre-outburst) luminosity of the system. Only by
1400: obtaining a radial-velocity curve and rotational velocity of the donor
1401: star, combined with ellipsoidal light curves can the black hole mass
1402: be concretely established.
1403:
1404:
1405: \section{Conclusions}
1406: We have conducted optical and X-ray simultaneous spectral and timing
1407: observations of \swift, while the X-ray binary was still in an active
1408: state following its 2005 outburst. We find a superhump-like blue
1409: modulation and remarkably featureless emission in the optical, and
1410: hard power-law
1411: spectrum in the X-ray, with possible thermal and iron K-line
1412: contributions. The power spectra are similar in X-rays and optical,
1413: flat below 0.05\,Hz and decreasing at higher frequencies, except that
1414: the optical power spectra show $\sim$0.08\,Hz
1415: QPOs. Dynamic power spectra show these QPOs to be discreet signals
1416: wandering in frequency and persistent on timescales of
1417: $\sim$10\,min. These, together with the respective auto- and
1418: cross-correlation functions and possible cyclotron emission, suggest
1419: that magnetic processes tie together the disc and high-energy
1420: emission.
1421:
1422: Similar features have been seen for two other interesting SXRTs,
1423: \xte\ and \gx. One suggested reason for the difference of
1424: these sources to normal SXRTs is that the accretion disc extends right
1425: in to the inner-most stable orbit.
1426:
1427: \medskip
1428: \section*{Acknowledgments}
1429: MD and TS are funded by the Spanish Ministry of Science under the
1430: grant AYA\,2004\,02646 and AYA\,2007\,66887.
1431: PG is a Fellow of the Japan Society for the Promotion of Science
1432: (JSPS). Based on observations carried out in ESO programmes 079.D-0535,
1433: and 279.D-5021, and during RXTE Cycle 12.
1434: ULTRACAM was designed and built with funding from PPARC (now
1435: STFC), and used as a visiting instrument at ESO Paranal, and RXTE is
1436: operated by NASA. We are grateful for rapid service observations by NOT.
1437: Partially funded by the Spanish MEC under the
1438: Consolider-Ingenio 2010 Program grant CSD2006-00070: ``First Science
1439: with the GTC'' ({\tt http://www.iac.es/consolider-ingenio-gtc/}).
1440:
1441: Thanks to Cadolle-Bel et al. and Paolo Soleri for use of their data in
1442: our SED plots.
1443:
1444:
1445: \begin{thebibliography}{99}
1446: \bibitem{belo}
1447: Beloborodov, A., 1999, ApJ, 510, L123
1448: \bibitem{phots}
1449: Bessell, M., 1995, IAU Symposium, 167, 175
1450: \bibitem{xte}
1451: Bradt, H., Rothschild, R., Swank, J., 1993, A\&AS, 97, 355
1452: \bibitem{flick}
1453: Bruch, A., 1992, A\&A, 266, 237
1454: \bibitem{xrt}
1455: Burrows, D. N., Hill, J. E., Nousek, J. A., Kennea, J. A., Wells, A.,
1456: Osborne, J. P., Abbey, A. F., Beardmore, A., Mukerjee, K., Short,
1457: A. D. T., Chincarini, G., Campana, S., Citterio, O., Moretti, A.,
1458: Pagani, C., Tagliaferri, G., Giommi, P., Capalbi, M., Tamburelli, F.,
1459: Angelini, L., Cusumano, G., Br\"auninger, H. W., Burkert, W., Hartner,
1460: G. D., 2005, SSRv, 120, 165
1461: \bibitem{cad}
1462: Cadolle Bel, M., Rib\'o, M., Rodriguez, J., Chaty, S., Corbel, S.,
1463: Goldwurm, A., Frontera, F., Farinelli, R., D'Avanzo, P., Tarana, A.,
1464: Ubertini, P., Laurent, P., Goldoni, P., Mirabel, I., 2007, ApJ, 659, 549
1465: \bibitem{xte2}
1466: Chaty, S., Haswell, C., Malzac, J., Hynes, R., Shrader, C., Cui, W.,
1467: 2003, MNRAS, 346, 689
1468: \bibitem{adaf}
1469: Czerny, B., R\'oz\.{a}\'ska, A., Janiuk, A., Z\.{y}cki, P.,
1470: 2000, NewAR, 44, 439
1471: \bibitem{ultra2}
1472: Dhillon, V., Marsh, T., Stevenson, M, Atkinson, D., Kerry, P.,
1473: et al. 2007, MNRAS, 378, 825
1474: \bibitem{me}
1475: Durant, M., Gandhi, P., Shahbaz, T., Fabian, A., Miller, J., Dhillon,
1476: V., Marsh, T., 2008, ApJL submitted
1477: \bibitem{cyclo}
1478: Fabian, A., Guilbert, P., Motch, C., Ricketts, M., Ilovaisky, S.,
1479: Chevalier, C., A\&A, 111, L9
1480: \bibitem{rad}
1481: Fender, R., Garrington, S., Muxlow, T., 2005, ATel, 558
1482: \bibitem{jets}
1483: Fender, R. \& Belloni, T., 2004, ARA\&A, 42, 317
1484: \bibitem{poshak}
1485: Gandhi et al., 2008 in preparation
1486: \bibitem{opt}
1487: Halpern, J., 2005, ATel, 549
1488: \bibitem{superhump}
1489: Haswell, C., King, A., Murray, J., Charles, P., 2001, MNRAS, 321, 475
1490: \bibitem{hom}
1491: Homan, J., Belloni, T., 2005, Ap\&SS, 300, 107
1492: \bibitem{j1118}
1493: Hynes, R., Haswell, C., Cui, W., Shrader, C., O'Brien, K., Chaty, S.,
1494: Skillman, D., Patterson, J., Horne, K., 2003, MNRAS, 345, 292
1495: \bibitem{orbitlag}
1496: Hynes, R., 2005, ASPC, 330, 237
1497: \bibitem{echoes}
1498: Hynes, R., 2006, AIPC, 840, 88
1499: \bibitem{pca}
1500: Jahoda, K., Swank, J. H., Giles, A. B., Stark, M. J., Strohmayer, T.,
1501: Zhang, W., Morgan, E. H., 1996, SPIE, 2808, 59
1502: \bibitem{pcaback}
1503: Jahoda, K., Markwardt, C., Radeva, Y., Rots, A., Stark, M., Swank, J.,
1504: Strohmayer, T., Zhang, W., 2006, ApJS, 163, 401
1505: \bibitem{nat}
1506: Kanbach, G., Straubmeier, C., Spruit, H., Belloni, T., 2001, Nature,
1507: 414, 180
1508: \bibitem{isgri}
1509: Lebrun, F., Leray, J. P., Lavocat, P., Cr\'etolle, J., Arqu\'es, M.,
1510: Blondel, C., Bonnin, C., Bou\'ere, A., Cara, C., Chaleil, T., Daly,
1511: F., Desages, F., Dzitko, H., Horeau, B., Laurent, P., Limousin, O.,
1512: Mathy, F., Mauguen, V., Meignier, F., Molini\'e, F., Poindron, E.,
1513: Rouger, M., Sauvageon, A., Tourrette, T., 2003, A\&A, 411, L141
1514: \bibitem{asm}
1515: Levine, A., Bradt, H., Cui, W., Jernigan, J.., Morgan, E., Remillard,
1516: R., Shirey, R., Smith, D., 1996, ApJ, 469, L33
1517: \bibitem{disc}
1518: Liu, B., Taam, R., Meyer-Hofmeister, E., Meyer, F., 2007, ApJ, 671,
1519: 695
1520: \bibitem{jetmod}
1521: Malzac, J., Merloni, A., Fabian, A., 2004, MNRAS, 351, 253
1522: \bibitem{adc}
1523: Malzac, J., 2007, MmSAI, 78, 382
1524: \bibitem{xtedist}
1525: McClintock, J. E., Garcia, M. R., Caldwell, N., Falco, E. E.,
1526: Garnavich, P. M., Zhao, P., 2001, ApJ, 551, L147
1527: \bibitem{BH}
1528: McClintock, J., Remillard, R., 2006, Chapter 4 of ``Compact Stellar
1529: X-ray Sources'', eds. W. Lewin and M. van der Klis, CUP
1530: \bibitem{thunder}
1531: Merloni, A. \& Fabian, A., 2001, MNRAS, 328, 958
1532: \bibitem{lowhard}
1533: Meyer-Hofmeister, E., 2004, A\&A, 423, 321
1534: \bibitem{smalldisc}
1535: Miller, J., Homan, J, Miniutti, G., 2006, ApJ, 652, L113
1536: \bibitem{ironlines}
1537: Miller, J., 2007, ARA\&A, 45, 441
1538: \bibitem{qpo}
1539: Morgan, E., Swank, J., Markwardt, C., Gehrels, N., 2005, ATel, 550
1540: \bibitem{UV}
1541: Morris, D., Burrows, D., Racusin, J., Roming, P., Chester, M.,
1542: Verghetta, R., Markwardt, C., Barthelmy, S., 2005, ATel, 552
1543: \bibitem{gx}
1544: Motch, C., Ricketts, M., Page, C., Ilovaisky, S., Chevalier, C.,
1545: 1983, A\&A, 119, 171
1546: \bibitem{discover}
1547: Palmer, D., Barthelmey, S., Cummings, J., Gehrels, N., Krimm, H.,
1548: Markwardt, C., Sakamoto, T., Tueller, J., 2005, ATel, 546
1549: \bibitem{rama}
1550: Ramadevi, M., Seetha, S., 2007, MNRAS, 378, 182
1551: \bibitem{review}
1552: Remillard, R. A., McClintock, J. E., 2008, ARA\&A, 44, 49
1553: \bibitem{hexte}
1554: Rothschild, R. E., Blanco, P. R., Gruber, D. E., Heindl, W. A.,
1555: MacDonald, D. R., Marsden, D. C., Pelling, M. R., Wayne, L. R., Hink,
1556: P. L., 1998, ApJ, 496, 538
1557: \bibitem{huaqu}
1558: Schwope, A., Thomas, H-C., Mantel, K-H., Haefner, R., Staude, A.,
1559: 2003, A\&A, 402, 201
1560: \bibitem{tariq1}
1561: Shahbaz, T., Dhillon, V. S., Marsh, T. R., Casares, J., Zurita, C.,
1562: Charles, P. A., Haswell, C. A., Hynes, R. I., 2005, MNRAS, 362, 975
1563: \bibitem{tariq2}
1564: Shahbaz, T., Dhillon, V. S., Marsh, T. R., Zurita, C., Haswell, C. A.,
1565: Charles, P. A., Hynes, R. I., Casares, J., 2003, MNRAS, 346, 1116
1566: \bibitem{sol}
1567: Soleri, P., Altamirano, D., Fender, R., Casella, P., Tudose, V.,
1568: Maitra, D., Mijnands, R., Belloni, T., Miller-Jones, J., Klein-Wolt,
1569: M., van der Klis, M., 2008
1570: \bibitem{gps}
1571: Stil, J., Taylor, A., Dickey, J., Kavars, D., Martin, P., Rothwell,
1572: T., Boothroyd, A., Lockman, F., \& McClure-Griffiths, N., 2006, AJ, 132, 1158
1573: \bibitem{loc}
1574: Still, M., Roming, P., Brocksopp, C., Markwardt, C., 2005, ATel, 553
1575: \bibitem{xrts}
1576: Tanaka, Y., \& Shibazaki, N., 1996, ARA\&A, 34, 607
1577: \bibitem{halpha}
1578: Torres, M., Steeghs, D., Garcia, M., McClintock, J., Miller, J.,
1579: Jonker, P., Callanan, P., Zhao, P., Huchra, J., U, Vivian, Hutcheson,
1580: C., 2005, ATel, 551
1581: \bibitem{RMS}
1582: Uttley, P. \& McHardy, I., 2001, MNRAS, 312, 880
1583: \bibitem{foam}
1584: Uzdensky, D., Goodman, J., 2007, MmSAI, 78, 403
1585: \bibitem{anotherreview}
1586: van der Klis, M., in Compact Stellar X-ray Sources. (eds Lewin
1587: W. H. G., van der Klis M.), Cambridge Univ. Press, p. 39
1588: \bibitem{swiftqpos}
1589: Zhang, G.-B., Qu, J.-L., Zhang, S., Zhang, C.-M., Zhang, F., Chen, W.,
1590: Song, L.-M., Yang, S.-P., 2007, ApJ, 659, 1511
1591: \bibitem{christina}
1592: Zurita, C., Durant, M., Torres, M., Shahbaz, T., Casares, J., 2008,
1593: ApJ, accepted
1594:
1595: \end{thebibliography}
1596:
1597: \bsp
1598:
1599: \label{lastpage}
1600:
1601: \end{document}
1602: