1: \documentclass[]{emulateapj}

2: \usepackage{epsfig}

3: \usepackage{graphicx}

4: \usepackage{longtable}

5: \usepackage{rotating}

6: 

7: \renewcommand{\arraystretch}{0.7}

8: %\usepackage{palatino,  multicol}

9: 

10: \begin{document}

11: 

12: \title{Discovery of kilohertz quasi-periodic oscillations and state

13: transitions in the low-mass X-ray binary 1E~1724--3045 (Terzan~2)}

14: 

15: \author{D. Altamirano\altaffilmark{1}, 

16: M. van der Klis\altaffilmark{1},

17: M.  M\'endez\altaffilmark{1,2},

18: R. Wijnands\altaffilmark{1}, \\

19: C. Markwardt \altaffilmark{3,4} \&

20: J. Swank \altaffilmark{3}\\

21: }

22: 

23: \email{d.altamirano@uva.nl}

24: 

25: \altaffiltext{1}{Astronomical Institute, ``Anton Pannekoek'',

26:   University of Amsterdam, and Center for High Energy Astrophysics,

27:   Kruislaan 403, 1098 SJ Amsterdam, The Netherlands.}

28: \altaffiltext{2}{Kapteyn Astronomical Institute, University of

29:   Groningen, P.O. Box 800, 9700 AV Groningen, The Netherlands.}

30: \altaffiltext{3}{Laboratory for High-Energy Astrophysics, NASA Goddard

31:   Space Flight Center, Greenbelt, MD 20771, U.S.A.}

32: \altaffiltext{4}{Department of Astronomy, University of Maryland,

33:   College Park, MD 20742., U.S.A.}

34: 

35: 

36: 

37: 

38: 

39: 

40: 

41: \begin{abstract}

42: We have studied the rapid X-ray time variability in 99 pointed

43: observations with the \textit{Rossi X-ray Timing Explorer} (RXTE)'s

44: Proportional Counter Array of the low-mass X-ray binary 1E~1724--3045

45: which includes, for the first time, observations of this source in its

46: island and banana states, confirming the atoll nature of this

47: source. We report the discovery of kilohertz quasi-periodic

48: oscillations (kHz QPOs). Although we have 5 detections of the lower kHz

49: QPO and one detection of the upper kHz QPO, in none of the

50: observations we detect both QPOs simultaneously.  By comparing the

51: dependence of the rms amplitude with energy of kHz QPOs in different

52: atoll sources, we conclude that this information cannot be use to

53: unambiguously identify the kilohertz QPOs as was previously

54: thought. We find that Terzan~2 in its different states shows 

55: timing behavior similar to that seen in other neutron-star low mass X-ray

56: binaries (LMXBs). We studied the flux transitions observed between

57: February 2004 and October 2005 and conclude that they are due to

58: changes in the accretion rate.

59: 

60: 

61: \end{abstract}

62: \date{}

63:  \keywords{accretion, accretion disks --- binaries: close ---

64: stars: individual (4U 1636--53,4U 1820--30,4U 1608--52,4U 0614+09,4U

65: 1728--34, Terzan 2, 1E 1724-3045) --- stars: neutron --- X--rays: stars}

66: \maketitle

67: 

68: 

69: \section{Introduction}

70: \label{sec:intro}

71: 

72: Low-mass X-ray binaries (LMXBs) can be divided into systems containing

73: a black hole candidate (BHC) and those containing a neutron star

74: (NS). The accretion process onto these compact objects can be studied

75: through the timing properties of the associated X-ray emission

76: \citep[see, e.g., ][for a review]{Vanderklis06}.  \citet{Hasinger89}

77: classified the NS LMXBs based on the correlated variations of the

78: X-ray spectral and rapid X-ray variability properties. They

79: distinguished two sub-types of NS LMXBs, the Z sources and the atoll

80: sources, whose names were inspired by the shapes of the tracks that

81: these sources trace out in an X-ray color-color diagram (CD) on time scales of

82: hours to days. The Z sources are the most luminous, but the atoll

83: sources are more numerous and cover a much wider range in luminosities

84: \citep[e.g. ][and references within]{Ford00}. For each type of source,

85: several spectral/timing states are identified which are thought to

86: arise from qualitatively different inner flow configurations

87: \citep{Vanderklis06}.  In the case of atoll sources, the three main

88: states are the extreme island state (EIS), the island state (IS) and

89: the banana branch, the latter subdivided into lower-left banana (LLB),

90: lower banana (LB) and upper banana (UB) states.

91: The EIS and the IS occupy the spectrally harder parts of the color

92: color diagram and correspond to lower levels of X-ray luminosity

93: ($L_x$).  The associated patterns in the CD are traced out in hours to

94: weeks.  The hardest and lowest $L_x$ state is the EIS, which shows

95: strong \citep[up to 50\% rms amplitude, see][ and references

96: within]{Linares07} low-frequency flat-topped noise also known as

97: band-limited noise (BLN).  The IS is spectrally softer and has higher

98: X-ray luminosity than the EIS.  Its power spectra are characterized by

99: broad features and a dominant BLN component which becomes weaker and

100: generally higher in characteristic frequency as the flux increases and

101: the $>6$ keV spectrum gets softer.  In order of increasing $L_x$ we

102: then encounter the LLB, where twin kHz QPOs are generally first

103: observed, the LB, where 10-Hz BLN is still dominant and finally, the

104: UB, where the $<1$~Hz (power law) very low frequency noise (VLFN)

105: dominates.  In the banana states, some of the broad features observed

106: in the EIS and the IS become narrower (peaked) and occur at higher

107: frequency.

108: %

109: In particular, the twin kHz QPO can be found in the LLB at

110: frequencies higher than 700~Hz \citep[the lower of them with

111: frequencies generally lower than a 1000~Hz, while the upper one up to

112: 1258~Hz, see e.g.,][]{Belloni05a,Jonker07}, only one kHz QPO can be

113: generally found in the LB, and neither of them is detected in the UB

114: \citep[see reviews by][]{Vanderklis00,Vanderklis04,Vanderklis06}.

115: %

116: 

117: 

118: 

119: 

120: 

121: 

122: A small number of weak NS LMXBs 

123: %

124: which do not get brighter than a few times $10^{36}$ ergs s$^{-1}$

125: \citep[usually burst sources and often referred to as 'weak' or 'faint

126: bursters', see e.g,][ and references within]{Muno03}

127: %

128:  resemble atoll sources in the EIS, but in the absence of state

129:  transitions this identification has been tentative \citep[see for

130:    example][]{Barret00a,Vanderklis06}.  An important clue is provided

131:  by the correlations between the component frequencies (and strengths

132:  - see e.g. \citealt{Straaten02,Straaten03,Altamirano08}) which helps

133:  to identify components across sources.

134: %

135: For example, \citet{Straaten02,Straaten03} compared the timing

136: properties of the atoll sources 4U~0614+09, 4U~1608--52 and

137: 4U~1728--34 

138: %

139: (see also \citealt{Altamirano08}, for similar results when

140: the atoll source 4U~1636--53 was included in the sample)

141: %

142: and conclude that the frequencies of the variability

143: components in these sources follow the same pattern of correlations

144: when plotted versus the frequency of the upper kHz QPO ($\nu_u$).

145: %and not versus $\nu_{\ell}$, as in the case of \citet{Psaltis99}.

146: \citet{Straaten03} also showed that low luminosity systems

147: extend the frequency correlations observed for the atoll sources. This

148: last result gave further clues in the link between the atoll and the

149: low luminosity sources.

150: %

151: 

152: 

153: \citet{Psaltis99} found an approximate frequency correlation involving a

154:  low-frequency QPO, 

155: %($L_h$ and $L_{LF}$) 

156:  the lower kHz QPO frequency and two broad noise components

157: interpreted as low-frequency versions of these features.  This

158: correlation spans nearly three decades in frequency, where the Z and

159: bright atoll sources populate the $>100$\,Hz range and black holes and

160: weak NS systems the $<10$\,Hz range. As already noted by

161: \citet{Psaltis99}, because the correlation combines features from

162: different sources which show either peaked  or broad components with

163: relatively little overlap, the data are suggestive but not

164: conclusive with respect to the existence of a single correlation

165: covering this wide frequency range \citep{Vanderklis06}.

166: %

167: 

168: 

169: The low-luminosity neutron star systems can play a crucial role in

170: clearing up this issue. Observations of different source states in

171: such a system could connect the $<10$ and $>100$~Hz regions mentioned

172: above by direct observation of a transition in a single source.  In

173: the case of the pattern of correlations reported by

174: \citet{Straaten03}, low luminosity NS systems extend the frequency

175: correlations observed for ordinary atoll sources down to $\sim100$~Hz.

176: Unfortunately, the low luminosity NS systems are usually observed in

177: only one state (EIS), which makes it difficult to properly link these

178: sources to the atoll sources. However, some of these objects show rare

179: excursions to higher luminosity levels which might correspond to other

180: states. The occurrence of these excursions are usually

181: unpredictable. Therefore, in practice it was not possible until now to

182: check on the frequency behavior of the different variability

183: components as such a source enters higher luminosity states.

184: 

185: 1E~1724--3045 is a classic low luminosity LMXB; a persistent Low-Mass

186: X-ray binary located in the globular cluster Terzan~2

187: \citep{Grindlay80} which is a metal-rich globular cluster of the

188: galactic bulge.  Its distance is estimated to be between 5.2 to 7.7

189: kpc \citep{Ortolani97}.  These values are consistent with that derived

190: from a type I X-ray burst that showed photospheric expansion

191: \citep[see][ but also see

192: \citealt{Kuulkers03,Galloway06}]{Grindlay80}.  The type I X-ray bursts

193: observed from this source also indicate that the compact object is a

194: weakly magnetized neutron star

195: \citep{Swank77,Grindlay80}. \citet{Emelyanov02} have shown, using

196: $\sim30$ years of data from several X-ray satellites, that the

197: luminosity of Terzan~2 increased until reaching a peak in 1997, after

198: which it started to decrease. They suggest that the evolution of the

199: donor star or the influence of a third star could be the cause of this

200: behavior. \citet{Olive98} and \citet{Barret00} have shown that during

201: earlier observations of Terzan~2 its X-ray variability at 

202: frequencies $\gtrsim0.1$~Hz resembled that of black hole

203: candidates. This state was tentatively identified as the extreme

204: island state for atoll sources. Until now, no kilohertz quasi-periodic

205: oscillations have been reported for this source, which was attributed

206: to the fact that the source was always observed in a single intensity

207: state \citep{Barret00a}.

208: 

209: 

210: 

211: Monitoring observations by the All Sky Monitor aboard the Rossi X-ray

212: Timing Explorer showed that the source was weakly variable in X-rays

213: (less than about a factor of 3 on a few day time scale for the first

214: 8 years of the monitoring).  However, recently \citet{Markwardt04} 

215: reported (using PCA monitoring observations of the galactic bulge -

216: \citealt{Swank01}) that during February 2004, 1E~1724--3045 flared up

217: from its relatively steady $\sim20$ mCrab to $\sim66$~mCrab

218: (2--10~keV).

219: %

220: In this paper we report a complete study of the timing variability of

221: the source.

222: %

223: %

224: For simplicity, and since only one bright X-ray source is detected in the

225: globular cluster (See Section~\ref{sec:samesource}), in the rest of

226: this report we will refer to 1E~1724--3045 as Terzan~2.

227: 

228: 

229: 

230: 

231: \section{OBSERVATIONS AND DATA ANALYSIS}

232: \label{sec:dataanalysis}

233: 

234: \subsection{Light curves and color diagrams}\label{sec:lcandccd}

235: We use data from the Rossi X-ray Timing Explorer (RXTE) Proportional

236: Counter Array \citep[PCA; for instrument information

237: see][]{Zhang93,Jahoda06}. There were 534 slew observations until

238: October $30^{th}$ 2006, which are part of the PCA monitoring

239: observations of the galactic bulge \citep{Swank01} and which were

240: performed typically every 3 days. These observations were only used to

241: study the long-term $L_x$ behavior of the source.

242: 

243: There were also 99 pointed observations in the nine data sets we used

244: (10090-01, 20170-05, 30057-03, 50060-05, 60034-02, 80105-10, 80138-06,

245: 90058-06 \& 91050-07), containing $\sim0.8$ to $\sim26$ ksec of useful

246: data per observation.  We use the 16-s time-resolution Standard 2 mode

247: data to calculate X-ray colors.  Hard and soft color are defined as

248: the 9.7--16.0 keV / 6.0--9.7 keV and 3.5--6.0 keV / 2.0--3.5 keV count

249: rate ratio, respectively, and intensity as the 2.0--16.0 keV count

250: rate. The energy-channel conversion was done using the pca\_e2c\_e05v02

251: table provided by the RXTE Team\footnote{see

252: http://heasarc.gsfc.nasa.gov/docs/xte/pca\_news.html}. Channels were

253: linearly interpolated to approximate these precise energy limits.

254: X-ray type I bursts were removed, background was subtracted and

255: deadtime corrections were made.  In order to correct for the gain

256: changes as well as the differences in effective area between the PCUs

257: themselves, we normalized our colors by the corresponding Crab Nebula

258: color values; \cite[see][ see table 2 in \citealt{Altamirano08} for

259: average colors of the Crab Nebula per

260: PCU]{Kuulkers94,Straaten03} that are closest in time but in the same

261: RXTE gain epoch, i.e., with the same high voltage setting of the PCUs

262: \citep{Jahoda06}.

263: 

264: \subsection{Fourier timing analysis and fitting models.}\label{sec:fourier}

265: For the Fourier timing analysis we used either the Event modes

266: E\_125us\_64M\_0\_1s or E\_16us\_64M\_0\_8s, or the Good Xenon data.

267: Leahy-normalized power spectra were constructed using data segments of

268: 128 seconds and 1/8192~s time bins such that the lowest available

269: frequency is $1/128 \approx 8 \times 10^{-3}$~Hz and the Nyquist

270: frequency 4096~Hz.  No background or deadtime corrections were

271: performed prior to the calculation of the power spectra.  We first

272: averaged the power spectra \textit{per observation}.  We inspected the

273: shape of the average power spectra at high frequency ($>2000$~Hz) for

274: unusual features in addition to the usual Poisson noise. None were

275: found.  We then subtracted a Poisson noise spectrum estimated from the

276: power between 3000 and 4000~Hz, using the method developed by

277: \citet{Kleinwolt04} based on the analytical function of

278: \citet{Zhang95}. In this frequency range, neither intrinsic noise nor

279: QPOs are expected based on what we observe in other sources.  The

280: resulting power spectra were converted to squared fractional rms

281: \citep{Vanderklis95b}.  In this normalization the power at each

282: Fourier frequency is an estimate of power density such that the square

283: root of the integrated power density equals the fractional rms

284: amplitude of the intrinsic variability in the source count rate in the

285: frequency range integrated over.

286: 

287: %

288: 

289: 

290: %

291: In order to study the behavior of the low-frequency components usually

292: found in the power spectra of neutron star LMXBs, we needed to improve

293: the statistics. We therefore averaged observations which were close in

294: time and had both similar colors and power spectra \citep[see e.g.][

295: and references within -- see also Appendix in \citealt{Altamirano08}

296: for a discussion on other possible

297: methods]{Straaten02,Straaten03,Straaten05,Altamirano05,Altamirano08}.

298: %

299: %The resulting data selections are labeled from A to R (ordered mainly

300: %in time -- see Table~\ref{table:averages} for details on which

301: %observations were used for each interval and their colors). 

302: %Their

303: %corresponding average power spectra are displayed in

304: %Figure~\ref{fig:pds}.

305: 

306: 

307: 

308: 

309: 

310: To fit the power spectra, we used a multi-Lorentzian function: the sum

311: of several Lorentzian components plus, if necessary, a power law to

312: fit the very low frequency noise (VLFN - see \citealt{Vanderklis06}

313: for a review).  Each Lorentzian component is denoted as $L_i$, where

314: $i$ determines the type of component. The characteristic frequency

315: ($\nu_{max}$ as defined below) of $L_i$ is denoted $\nu_i$.  For

316: example, $L_u$ identifies the upper kHz QPO and $\nu_u$ its

317: characteristic frequency.  By analogy, other components go by names

318: such as $L_{\ell}$ (lower kHz), $L_{\ell ow}$ (lower Lorentzian),

319: $L_{hHz}$ (hectohertz), $L_h$ (hump), $L_b$ (break frequency),

320: $L_{b2}$ (second break frequency) and their frequencies as

321: $\nu_{\ell}$, $\nu_{\ell ow}$, $\nu_{hHz}$, $\nu_h$ and $\nu_b$,

322: $\nu_{b2}$, respectively.  Using this multi-Lorentzian function makes

323: it straightforward to directly compare the characteristics of the

324: different components observed in Terzan~2 to those in previous works

325: which used the same fit function \citep[e.g.,][and references

326:   therein]{Belloni02,Straaten03,Straaten05,Altamirano05,Altamirano08}.

327: 

328: Unless stated explicitly, we only include those Lorentzians in the

329: fits whose single trial significance exceeds $3\sigma$ based on the

330: error in the power integrated from 0 to $\infty$.

331: We give the frequency of the Lorentzians in terms of characteristic

332: frequency $\nu_{max}$ as introduced by \citet{Belloni02}: $\nu_{max} =

333: \sqrt{\nu^2_0 + (FWHM/2)^2} = \nu_0 \sqrt{1 + 1/(4Q^2)}$.  For the

334: quality factor $Q$ we use the standard definition $Q =

335: \nu_0/FWHM$. FWHM is the full width at half maximum and $\nu_0$ the

336: centroid frequency of the Lorentzian. The quoted errors use

337: $\Delta\chi^2 = 1.0$.  The upper limits quoted in this paper

338: correspond to a 95\% confidence level ($\Delta\chi^2=2.7$).

339: 

340: \subsection{Energy spectra}

341: 

342: Since the energy spectra of the quiet state (see Section~\ref{sec:lc}) 

343: of Terzan~2 have already been studied in previous works \citep[see

344: e.g.][]{Olive98,Barret99,Barret00}, in this paper we concentrate on

345: the 14 observations that sample the flaring period (see

346: Section~\ref{sec:lc}). In all 14 cases, we used data of both the PCA

347: and the HEXTE instruments.

348: 

349: %

350: For the PCA, we only used the Standard 2 data of PCU 2, which

351: was active in all observations. The background was estimated using the

352: PCABACKEST version 6.0 (see FTOOLS). We calculated the PCU 2

353: response matrix for each observation using the FTOOLS routine PCARSP V10.1.

354: %

355: For the HEXTE instrument, spectra were accumulated for each cluster

356: separately.  Dead time corrections of both source and background

357: spectra were performed using HXTDEAD V6.0. The response matrices were

358: created using HXTRSP V3.1. 

359: %

360: For both PCA and HEXTE, we filtered out data recorded during, and

361: up to 30 minutes after  passage through the South

362: Atlantic Anomaly (SAA). We only use data when the pointing offset from

363: the source was less than 0.02 degrees and the elevation of the source

364: respect to the Earth was greater than 10 degrees.  We did not perform

365: any energy selection prior to the extraction of the spectra. Finally,

366: we fitted the energy spectra using XSPEC V11.3.2i.

367: 

368: 

369: 

370: 

371: \begin{figure}[!hbtp] 

372: \center

373: \resizebox{1\columnwidth}{!}{\rotatebox{0}{\includegraphics{./f1.eps}}}

374: \caption{ \textit{Above:} PCA count rate obtained from the monitoring

375: observations of Terzan~2 \citep[Monitoring observation of the galactic

376: bulge - see][]{Swank01}. These data was used for the study of the

377: long-term variability (see

378: Sections~\ref{sec:lsp}~\&~\ref{sec:resultslsp}). \textit{Below:}

379: Terzan~2 intensity (Crab normalized - see

380: Section~\ref{sec:dataanalysis}) versus time of all pointed

381: observations. These observations were used for the study of the

382: 0.1--1200~Hz X-ray variability (see Sections~\ref{sec:fourier},

383: \ref{sec:gralpds}~\&~\ref{sec:kHz}) except for the observation

384: 90058-06-04-00, which is marked with an arrow (see

385: Section~\ref{sec:results}). The modified Julian date is defined as

386: $MJD=$ Julian Date$ - 2400000.5$.}

387: \label{fig:lc}

388: \end{figure}

389: 

390: 

391: \begin{table*}[!hbtp]

392: \center

393: \begin{tabular}{|c|c|c|c|c|}\hline 

394: \hline

395:  Time interval  & Duration  & Maximum Count Rate   & Flare & N$^o$. of pointed obs.\\

396:   (MJD)         & (days)      &  (cts/s/5PCU)         & label & sampling the flare\\

397: \hline

398: $\sim53038-53071$ & 33 & $\sim850$  & F1& 7\\

399: $\sim53127-53155$ & 28 & $\sim950$  & F2& 4 \\

400: $\sim53233-53250$ & 17 & $\sim830$  & F3& 2\\

401: $\sim53493-53502$ & 9 & $\sim450$   & F4& 0\\

402: $\sim53566-53602$ & 36 & $\sim1150$ & F5& 1\\

403: $\sim53631-53651$ & 20 & $\sim650$  & F6& 0\\

404: $\sim53934-53972$ & 38 & $\sim865$  & F7& 0\\

405: \hline

406: \end{tabular}

407: \caption{Data on the 6 flares observed until MJD~53667. See

408: Section~\ref{sec:results} for details. (The modified Julian date is

409: defined as $MJD=$ Julian Date$ - 2400000.5$).}

410: \label{table:flares}

411: \end{table*}

412: 

413: 

414: \subsection{Search for quasi-periodic variations over days and months}\label{sec:lsp}

415: 

416: Recently, \citet{Wen06} have performed a systematic search for

417: periodicities in the light curves of 458 sources using data from the

418: RXTE All Sky Monitor (ASM). Terzan~2 was not included in their

419: analysis, probably due to the fact that the ASM

420: source average count rate is low: $2.05\pm0.01$ count/s (the average

421: of the errors -- $1/n \sum err_i$ -- is 0.8 count/s). 

422: 

423: Since the PCA galactic bulge monitoring \citep{Swank01} has observed

424: the source for more than 8 years, and the lowest detected source count

425: rate was $170\pm5$ counts/sec, this new data set provides useful

426: information to search for long term modulations.

427: %

428:  Lomb-Scargle periodograms \citep{Lomb76,Scargle82,Numerical-Recipes}

429: as well as the phase dispersion minimization technique \citep[PDM -

430: see][]{Stellingwerf78} were used. The Lomb-Scargle technique is

431: ideally suited to look for sinusoidal signals in unevenly sampled

432: data.  The phase dispersion minimization technique is well suited to

433: the case of non-sinusoidal time variation covered by irregularly

434: spaced observations.

435: 

436: 

437: 

438: 

439: \section{Results}\label{sec:results}

440: 

441: 

442: \subsection{The light curve}\label{sec:lc}

443: 

444: Figure~\ref{fig:lc} shows both the PCA monitoring lightcurve of the

445: source \citep[see upper panel and][]{Swank01} and the Crab normalized

446: intensity (see lower panel and Section~\ref{sec:lcandccd}) of each

447: pointed observation versus time (in units of modified Julian date $MJD=$

448: Julian Date$ - 2400000.5$).

449: %

450: In the rest of this paper, we will refer as ``quiet period'' to that

451: between MJD 51214 and 52945, and as ``flaring period'' between MJD

452: 52945 and 53666.

453: 

454: During the quiet period, 333 monitoring measurements of the source

455: intensity and 85 pointed observations sample the behavior of the

456: source. The count rate slowly decreases from an average of $\sim300$,

457: to an average of $\sim190$ counts$/$s$/$5PCU at an average rate of

458: $-0.059\pm0.002$ count/s/day.

459: %

460: In the flaring period, 7 flares sampled with 201 monitoring

461: observations were detected with the galactic bulge scan. 14 pointed

462: observations partially sampled parts of 4 of these flares.  In

463: Table~\ref{table:flares} we list approximate dates at which the flux

464: transitions occurred, the flare durations and the maximum count rates

465: detected with the PCA.  As mentioned in Section~\ref{sec:lsp}, the

466: monitoring is done approximately once every three days; additional

467: gaps in the data are present due to visibility windows.  As of course

468: we do not have details of flares that may have occurred during these

469: gaps, the information in Table~\ref{table:flares} is only approximate.

470: In Figure~\ref{fig:lczoom} we show the intensity of the source during

471: the flaring period. We label the different flares F1, F2, F3, F4, F5,

472: F6 and F7 in order of time of occurrence.

473: %

474: 

475: 

476: 

477: We detected three Type I X-ray bursts. One was during the quiet-state

478:  observation 10090-01-01-021 and two during the flaring-state

479:  observations 80138-06-06-00 and 90058-06-02-00. A detailed study of

480:  these X-ray bursts as well as a comparison to bursts observed in

481:  other sources can be found in \citet{Galloway06}.

482: 

483: 

484: 

485: 

486: 

487: \begin{figure}[!hbtp] 

488: \center

489: \resizebox{1\columnwidth}{!}{\rotatebox{0}{\includegraphics{./f2.eps}}}

490: \caption{ \textit{Above:} Intensity in units of the Crab Nebula of the

491:   observations which were performed during the flare state. The

492:   different states are labeled (see Section~\ref{sec:state1}). Arrows

493:   mark the observations in which kHz QPOs were detected (see also

494:   Section~\ref{sec:kHz} and Table~\ref{table:kHz}).

495: %

496: \textit{Below:} Intensity versus time

497: showing the seven flares. The data are the same as those of

498: Figure~\ref{fig:lc}. }

499: \label{fig:lczoom}

500: \end{figure}

501: 

502: 

503: \subsection{Color diagrams; identification of states}\label{sec:state1}

504: Figure~\ref{fig:ccd} (top) shows the color-color diagram of the 99

505: pointed observations. For comparison, we also include the color-color

506: diagram of the atoll source 4U~1608--52, which

507: has been observed in all extreme island, island and banana states

508: (Figure~\ref{fig:ccd}, bottom). The similarity in shape suggests that

509: Terzan~2 underwent state transitions during observations of the

510: flares. Based on Figure~\ref{fig:ccd} we can identify the probable

511: extreme island, island and banana state with the hardest, intermediate

512: and softest colors, respectively. Since partially sampled patterns in

513: the color color diagrams are not necessarily unambiguous, \citep[see

514: review by][]{Vanderklis06}, power spectral analysis (below) is

515: required to confirm these identifications. The extreme island state

516: is sampled with 85 pointed observations which are clumped in 2 regions

517: at similar hard colors but at significantly different soft colors.  We

518: find 3 observations in the island state. They sampled the lowest

519: luminosity sections of flares F2, F3 and F5 (see

520: Figure~\ref{fig:lczoom}).  The banana state is sampled by 11 pointed

521: observations: 7 during F1, 3 during F2 and 1 during F3.

522: %

523: The identifications above are strengthened by the similarities in

524: power spectral shapes between Terzan~2 and those reported in other

525: sources

526: \citep{Straaten03,Disalvo01,Straaten02,Disalvo03,Straaten05,Linares05,Altamirano05,Migliari05,Altamirano08}.

527: In the following sections we describe the power spectra in more

528: detail.

529: 

530: 

531: \begin{figure}[!hbtp] 

532: \center

533: \resizebox{1.0\columnwidth}{!}{\rotatebox{0}{\includegraphics{./f3a.eps}}}

534: \resizebox{0.7\columnwidth}{!}{\rotatebox{0}{\includegraphics[clip]{./f3b.eps}}}

535: \caption{\textit{Top:} Hard color versus soft color normalized to Crab

536: as explained in Section~\ref{sec:dataanalysis}.  Different symbols

537: represent the selections used for averaging the power spectra as

538: explained in Section~\ref{sec:dataanalysis} and shown in

539: Figure~\ref{fig:pds} (See Figure~\ref{fig:lc} for symbols). The arrow

540: marks observation 90058-06-04-00, which was excluded from interval N

541: (see Section~\ref{sec:results}). \textit{Bottom:} Hard color versus

542: soft color normalized to Crab for the NS source 4U~1608--52. This

543: source has been observed in all expected atoll states: extreme island

544: state (EIS), island state (IS), lower left banana (LLB), lower banana

545: (LB) and upper banana (UP). The similarity between both figures

546: suggest that Terzan~2 has been observed in similar states as

547: 4U~1608--52.  }

548: \label{fig:ccd}

549: \end{figure}

550: 

551: 

552: \subsection{kHz QPOs}\label{sec:kHz}

553: 

554: 

555: We searched each averaged observation's power spectrum for the

556: presence of significant kHz QPOs at frequencies $\gtrsim400$~Hz.

557: %

558: As reported in other works \citep{Barret99,Belloni02},

559: during the quiet period the power spectra of single averaged

560: observations show significant power up to $\sim300$~Hz which is fitted

561: with broad ($Q\lesssim0.5$) Lorentzians plus if necessary, one sharp 

562: Lorentzian to account for $L_{LF}$.

563: %

564: During the flaring period, we found that several observations show

565: power excess above 400~Hz. For each observation we fitted the averaged

566: power spectra between 400 and 2000~Hz with a model consisting of one

567: Lorentzian and a constant to take into account the QPO and Poisson

568: noise, respectively.

569: %

570: %

571: In 5 of the 14 observations that sample the flaring period, we detect

572: significant QPOs, with single trial significances up to

573: $6.5\sigma$. In Table~\ref{table:kHz} we present the results of our

574: fits and information on these observations.  As can be seen, the first

575: three observations were performed during the rise of the first flare

576: while the fourth and fifth observations where done during the decay of

577: the second flare (see Figure~\ref{fig:lczoom}).

578: %

579: As can be seen in Table~\ref{table:kHz}, there are significant

580: frequency variations. Unfortunately, due to the sparse coverage of the

581: flares and the fact that we cannot detect the QPO on shorter time

582: scales than an observation, no further conclusions are possible.

583: %

584: 

585: \begin{table*}[!hbtp]

586: \centering

587: 

588: \begin{tabular}{|c|c|c|c|c|c|c|c|c|}\hline 

589: \hline

590:  MJD & ObsId & Used in  & $\nu_0$ (Hz) & FWHM & Q & rms (\%) & Sig.         & c/s \\

591:      &       & interval &              &      &   &          &              & (5PCU) \\

592: \hline

593: 53054.94 & 80138-06-02-00 & M & $731\pm3$ & $33.2\pm7.3$&$ 22.1\pm  4.8$ & $10.4\pm0.8$ & $6.5\sigma$  & $\sim665$\\

594: 53055.00 & 80138-06-02-01 & M & $764\pm2$ & $14.3\pm5.2$&$ 53.4\pm 19.4$ & $6.9\pm0.8$  & $4.3\sigma$  & $\sim690$\\

595: 53056.69 & 80138-06-03-00 & M & $772\pm1$ & $15.1\pm4.2$&$ 51.1\pm 14.2$ & $7.1\pm0.6$  & $5.6\sigma$  & $\sim660$\\

596: 53145.57 & 90058-06-01-00 & N & $559\pm2$ & $17.3\pm6.9$&$ 32.3\pm 12.8$ & $6.6\pm0.7$  & $4.7\sigma$  & $\sim765$\\

597: 53147.21 & 90058-06-02-00 & N & $599\pm1$ & $20.3\pm4.8$&$ 29.5\pm  6.9$ & $7.4\pm0.5$  & $6.6\sigma$  & $\sim610$\\

598: \hline

599: \end{tabular}

600: \label{table:kHz}

601: \caption{Fit results for the 5 averaged observations showing significant

602: kHz QPOs. }

603: 

604: \end{table*}

605: %

606: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

607: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

608: 

609: We do not significantly detect two simultaneous kHz QPOs in any of the

610: 5 observations.  In order to search for a possible second kHz QPO, we

611: used the shift--and--add method as described by \citet{Mendez98a}.  We

612: first tried to trace the detected kilohertz QPO using a dynamical

613: power spectrum \citep[e.g. see figure 2 in ][]{Berger96} to visualize

614: the time evolution of the QPO frequency, but the signal was too weak

615: to be detected on timescales shorter than the averaged observation.

616: %

617: Therefore, for each observation we used the fitted

618: averaged frequency (see Table~\ref{table:kHz})

619: %We used the shift-and-add technique described by \citet{Mendez98a} 

620: % Méndez, van der Klis & van Paradijs (1998b) 

621: to shift each kilohertz QPO to the arbitrary frequency of 770~Hz.

622: Next, the shifted, aligned, power spectra were averaged. The average

623: power spectrum was finally fitted in the range 300--2048~Hz so as to

624: exclude the edges, which are distorted due to the shifting method.  To

625: fit the averaged power spectrum, we used a function consisting of a

626: Lorentzian and a constant to fit the QPO and the Poisson noise,

627: respectively.  We studied the residuals of the fit, but no significant

628: power excess was present apart from the 770~Hz feature. In

629: Figure~\ref{fig:kHz} we show the fitted kHz QPO for observation

630: 80138-06-02-00 (no shift and add was applied) and the

631: shifted-and-added kHz QPO detected with the method mentioned above.

632: % using the full energy range

633: %($\sim2-100$~keV). 

634: %

635: Since it is known that the kHz QPOs become stronger at higher energies

636: \citep[e.g.][and references within]{Berger96,Mendez98,Vanderklis00},

637: we repeated the analysis described above (which was performed on the

638: full PCA energy range), using only data at energies higher than

639: $\sim6$~keV or higher than $\sim10$~keV.  Again, no significant second

640: QPO was present.

641: %

642: It is important to note that this method can produce ambiguous results as we

643: cannot be sure we are always shifting  the same component (either $L_u$

644: or $L_{\ell}$).  We also tried different subgroups, i.e. adding only

645: two to three different observations, but found the same results.

646: 

647: 

648: 

649: 

650: 

651: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

652: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

653: 

654: To investigate the energy dependence of the kHz QPOs, we

655: divided each power spectrum into 3, 4 or 5 energy intervals in order

656: to have approximately the same count rate in all the intervals.  We

657: then produced the power spectrum as described in

658: Section~\ref{sec:dataanalysis} and refitted the data where both

659: frequency and  Q were fixed to the values obtained for

660: the full energy range  (see Table~\ref{table:kHz}).  In

661: Figure~\ref{fig:rmsvse} we show the results for the representative kHz

662: QPOs in flares F1 and F2 (observations 80138-06-03-00 and

663: 90058-06-02-00, respectively). 

664: %

665: Similarly to what is observed in other sources, the fractional rms

666: amplitude of the kHz QPOs increases with energy.

667: %

668: The data show that there is no significant

669: difference in the energy dependence of the kHz QPO at $\sim599$~Hz

670: with that at $\sim772$~Hz.

671: % and then the QPOs are either both $L_u$

672: %or both $L_{\ell}$.

673: %We do not show the other 3 observations since ... WHY WHY WHY????

674: %

675: %

676: 

677: \begin{figure}[!hbtp] 

678: \center

679: \resizebox{1\columnwidth}{!}{\rotatebox{0}{\includegraphics{./f4.eps}}}

680: \caption{Fit  to the  kHz QPO  in observation  80138-06-02-00.  In the

681: subplot on  the top-right of the figure,  we show the fit  to the data

682: after using the shift-and-add method on all 5 observations reported in

683: Table~\ref{table:kHz} where the main peak was set to the arbitrary

684: frequency  $\nu=770$~Hz  (see  Section~\ref{sec:kHz}). No  significant

685: detections of a second kHz QPO were found. }

686: \label{fig:kHz}

687: \end{figure}

688: 

689: 

690: \begin{figure}[!hbtp] 

691: \center

692: \resizebox{1\columnwidth}{!}{\rotatebox{-90}{\includegraphics{./f5.eps}}}

693: \caption{Energy dependence of two representative kHz QPOs of

694: Terzan~2. One in flare F1 (pentagons -- OBSid 80138-06-03-00) and

695: another in flare F2 (downward pointed triangles -- OBSid 90058-06-02-00). For

696: comparison we show the energy dependence of both lower (circles \&

697: open squares) and upper (squares) peak of the atoll source

698: 4U~1608--52 \citep{Berger96,Mendez98a} and the upper peak (upwards pointed

699: triangles) of the atoll source 4U~0614+09

700: \citep{Mendez97,Straaten00}. }

701: \label{fig:rmsvse}

702: \end{figure}

703: 

704: 

705: 

706: \subsection{Averaged power spectrum}\label{sec:gralpds}

707: 

708: As mentioned in Section~\ref{sec:fourier}, in order to study the

709: behavior of the low-frequency components usually found in the power

710: spectra of neutron star LMXBs, we needed to improve the statistics. We

711: therefore averaged observations which were close in time and had both

712: similar colors and power spectra.

713: %

714: % \citep[see e.g.][ and references

715: %  within -- see also Appendix in \citealt{Altamirano08} for a

716: %  discussion on other possible

717: %  methods]{Straaten02,Straaten03,Straaten05,Altamirano05,Altamirano08}.

718: %

719: The resulting data selections are labeled from A to R (ordered mainly

720: in time -- see Table~\ref{table:averages} for details on which

721: observations were used for each interval and their colors).  Their

722: corresponding average power spectra are displayed in

723: Figure~\ref{fig:pds}.

724: 

725: 

726: \begin{table*}[!hbtp]

727: \center

728: {

729: \begin{tabular}{|c|c|c|c|}\hline 

730: \multicolumn{4}{|c|}{Interval A} \\

731: \hline

732:  Observation  & Soft color (Crab) & Hard color (Crab) & Intensity (Crab) \\

733: \hline

734: 10090-01-01-000 & $ 1.3615\pm0.0017$ & $ 1.0760\pm0.0013$ & $ 0.0389\pm0.0001$ \\  

735: 10090-01-01-001 & $ 1.3508\pm0.0019$ & $ 1.0761\pm0.0014$ & $ 0.0384\pm0.0001$ \\  

736: 10090-01-01-00 & $ 1.3619\pm0.0019$ & $ 1.0757\pm0.0015$ & $ 0.0384\pm0.0001$ \\  

737: 10090-01-01-020 & $ 1.3783\pm0.0021$ & $ 1.0781\pm0.0016$ & $ 0.0386\pm0.0001$ \\  

738: 10090-01-01-021 & $ 1.4435\pm0.0019$ & $ 0.9846\pm0.0013$ & $ 0.0468\pm0.0001$ \\  

739: 10090-01-01-022 & $ 1.3638\pm0.0017$ & $ 1.0788\pm0.0013$ & $ 0.0387\pm0.0001$ \\  

740: 10090-01-01-02 & $ 1.3723\pm0.0039$ & $ 1.0801\pm0.0029$ & $ 0.0389\pm0.0001$ \\  

741: \hline

742: \multicolumn{4}{|c|}{Interval B} \\

743: \hline

744: 20170-05-01-00 & $ 1.3501\pm0.0075$ & $ 1.0660\pm0.0058$ & $ 0.0383\pm0.0001$ \\  

745: 20170-05-02-00 & $ 1.3440\pm0.0077$ & $ 1.0492\pm0.0059$ & $ 0.0384\pm0.0001$ \\  

746: 20170-05-03-00 & $ 1.3498\pm0.0076$ & $ 1.0541\pm0.0058$ & $ 0.0389\pm0.0001$ \\  

747: 20170-05-04-00 & $ 1.3449\pm0.0073$ & $ 1.0632\pm0.0056$ & $ 0.0394\pm0.0001$ \\  

748: 20170-05-05-00 & $ 1.3492\pm0.0071$ & $ 1.0703\pm0.0055$ & $ 0.0398\pm0.0001$ \\  

749: 20170-05-06-00 & $ 1.3392\pm0.0072$ & $ 1.0650\pm0.0056$ & $ 0.0394\pm0.0001$ \\  

750: 20170-05-07-00 & $ 1.3465\pm0.0072$ & $ 1.0726\pm0.0055$ & $ 0.0393\pm0.0001$ \\  

751: 20170-05-08-00 & $ 1.3295\pm0.0071$ & $ 1.0605\pm0.0056$ & $ 0.0382\pm0.0001$ \\  

752: 20170-05-09-00 & $ 1.3398\pm0.0070$ & $ 1.0704\pm0.0054$ & $ 0.0389\pm0.0001$ \\  

753: 20170-05-10-00 & $ 1.3163\pm0.0064$ & $ 1.0687\pm0.0051$ & $ 0.0387\pm0.0001$ \\  

754: 20170-05-11-00 & $ 1.3601\pm0.0071$ & $ 1.0714\pm0.0054$ & $ 0.0400\pm0.0001$ \\  

755: 20170-05-12-00 & $ 1.3365\pm0.0074$ & $ 1.0747\pm0.0058$ & $ 0.0399\pm0.0001$ \\  

756: 20170-05-13-00 & $ 1.3532\pm0.0070$ & $ 1.0721\pm0.0053$ & $ 0.0409\pm0.0001$ \\  

757: 20170-05-14-00 & $ 1.3215\pm0.0076$ & $ 1.0768\pm0.0061$ & $ 0.0407\pm0.0001$ \\  

758: 20170-05-15-00 & $ 1.3417\pm0.0070$ & $ 1.0658\pm0.0054$ & $ 0.0394\pm0.0001$ \\  

759: 20170-05-16-00 & $ 1.3398\pm0.0081$ & $ 1.0669\pm0.0062$ & $ 0.0398\pm0.0001$ \\  

760: 20170-05-17-00 & $ 1.3290\pm0.0072$ & $ 1.0832\pm0.0057$ & $ 0.0386\pm0.0001$ \\  

761: 20170-05-18-00 & $ 1.3318\pm0.0086$ & $ 1.0682\pm0.0067$ & $ 0.0392\pm0.0001$ \\  

762: 20170-05-19-00 & $ 1.3336\pm0.0068$ & $ 1.0667\pm0.0053$ & $ 0.0395\pm0.0001$ \\  

763: 20170-05-20-00 & $ 1.3309\pm0.0069$ & $ 1.0599\pm0.0054$ & $ 0.0397\pm0.0001$ \\  

764: \hline

765: \end{tabular}

766: \caption{Observations used for the timing analysis. The colors and

767: intensity are corrected by dead time and normalized to the Crab Nebula

768: (see Section~\ref{sec:dataanalysis}).  The complete table can

769: be obtained digitally from ApJ.}

770: % The complete table is called big_table.tex and can be found at

771: % /home/diego/papers/Terzan2/data/averages/big_table/

772: \label{table:averages}

773: }

774: \end{table*}

775: 

776: 

777: 

778: \begin{figure*}[!hbtp] 

779: \center

780: \resizebox{1.4\columnwidth}{!}{\rotatebox{0}{\includegraphics[clip]{./f6.eps}}}

781: \caption{Power spectra and fit functions in the power spectral density

782:   times frequency representation.  Each plot corresponds to a

783:   different region in the color-color and color-intensity diagrams

784:   (see Figures~\ref{fig:ccd} and~\ref{fig:lc}).  The curves mark the

785:   individual Lorentzian components of the fit.  The re-binning varies

786:   between the panels to balance the frequency resolution with the

787:   signal-to-noise ratio. For a detailed identification, see

788:   Table~\ref{table:data}. }

789: \label{fig:pds}

790: \end{figure*}

791: 

792: \subsubsection{The power spectra of the quiet period}\label{pdsqp}

793: 

794: Intervals A to L are all part of the quiet segment. 4 to 6

795: components were needed to fit all the power spectra of this group,

796: where 11 out of the 12 power spectra showed significant broad components at

797: $\sim150$, $\sim10$, $\sim1$ and at $\sim0.2$~Hz. 

798: %

799: %

800:  Interval G is the

801: exception where the broad component at $\sim150$~Hz was not

802: significantly detected ($8.5$\% rms-amplitude upper limit).  A QPO

803: ($Q\gtrsim2$) at $\sim0.8$~Hz was also significantly

804: detected in 11 out of 12 power spectra, interval B being the exception ($1.4$\%

805: rms-amplitude upper limit).

806: %

807: %Interval F is the only power spectra which shows a peaked (Q$=17.9\pm8.4$)

808: %QPO at $216.1\pm2.2$~Hz. It is 3.1$\sigma$ significant (single trial).

809: Finally, an extra component ($L_{vl}$, where vl stands for ``very

810: low'' ) at frequency $\nu_{vl}\sim0.1$~Hz was detected only in

811: interval C.

812: %

813: 

814: Similar power spectra have been reported in the extreme island state of

815: the atoll sources 4U~0614+09, 4U~1728--34 \citep{Straaten02},

816: XTE~J1118+480, SLX~1735--269 \citep{Belloni02} and 4U~1608--52

817: \citep{Straaten03}. \citet{Olive98} and \citet{Belloni02} analyzed 

818: an early subset of the data we present in this work. Our results are consistent

819: with the frequencies reported in those works.

820: %

821: Following \citet{Straaten02}, \citet{Belloni02} and

822: \citet{Straaten03}, we identify the components as $L_u$, $L_{\ell

823: ow}$, $L_h$, $L_{LF}$ and $L_b$, where $\nu_u\sim150$~Hz, $\nu_{\ell

824: ow}\sim10$~Hz, $\nu_h\sim1$~Hz, $\nu_{LF}\sim0.8$~Hz and

825: $\nu_b\sim0.1$~Hz, respectively.  These identifications are

826: strengthened by the correlations shown in Figure~\ref{fig:nuvsnu}, in

827: which we plot the characteristic frequency of all components versus

828: that of $\nu_u$ for several atoll sources (see Section~\ref{sec:tfp});

829: clearly, we observe the same components as seen in the EIS in other

830: atoll sources. 

831: %

832: In Figure~\ref{fig:syntetic} we provide a modeled power spectra were

833: the main 5 Lorentzians (i.e.  $L_u$, $L_{\ell ow}$, $L_h$, $L_{LF}$

834: and $L_b$) are represented. Components in Figures~\ref{fig:pds}

835: and~\ref{fig:syntetic} are represented with the same line style.

836: 

837: 

838: 

839: 

840: 

841: 

842: \begin{figure}[!hbtp] 

843: \center

844: \resizebox{1\columnwidth}{!}{\rotatebox{0}{\includegraphics[clip]{./f7.eps}}}

845: \caption{Example of the typical power spectra of the quiet period. The

846:   main 5 Lorentzians are $L_u$, $L_{\ell ow}$, $L_h$, $L_{LF}$ and

847:   $L_b$ (see Section~\ref{pdsqp}).}

848: \label{fig:syntetic}

849: \end{figure}

850: 

851: 

852: \begin{figure}[!hbtp] 

853: \center

854: \resizebox{1\columnwidth}{!}{\rotatebox{0}{\includegraphics{./f8.eps}}}

855: \caption{The characteristic frequencies $\nu_{max}$ of the various

856:   power spectral components plotted versus $\nu_{u}$. The grey symbols

857:   mark the atoll sources 4U~0614+09, 4U~1728--34 \citep{Straaten02},

858:   4U~1608--52 \citep{Straaten03}, Aql X-1 \citep{Reig04} and

859:   4U~1636--53 \citep{Altamirano08} and the low luminosity bursters

860:   Terzan~2 (previous results), GS~1826--24 and SLX~1735--269

861:   \citep[][but also see \citealt{Belloni02}]{Straaten05}.  The black

862:   bullets mark our results for Terzan~2. Note that we only plot the

863:   results for Intervals A-N and Q (note that as mentioned in

864:   Section~\ref{sec:tfp}, the points for intervals M and N

865:   ($\nu_u>800$~Hz) were plotted under the assumption that $\nu_u =

866:   \nu_{\ell} + 300$~Hz). For intervals O, P and R no kHz QPOs were

867:   detected (see Section~\ref{sec:gralpds}). }

868: \label{fig:nuvsnu}

869: \end{figure}

870: 

871: 

872: 

873: \subsubsection{The power spectra of the flaring period}\label{sec:flaringsegment}

874: Intervals M, N, O, P, Q and R are part of the flaring period.

875: %

876: For intervals M, N and Q, high frequency ($\gtrsim400$~Hz) single QPOs

877: are seen which can be identified as either  the lower or the upper kHz

878: QPO.

879: %

880: 

881: 

882: 

883: 

884: The kHz QPOs in interval M and N correspond to the averages of

885: significant QPOs observed in single observations. As seen in

886: Table~\ref{table:kHz}, the characteristic frequency of the kHz QPOs

887: averaged in each of the two intervals are within a range of 50~Hz. By

888: our averaging method we are affecting the Q value of the kHz QPOs but

889: improving the statistics for measuring the characteristics of the

890: features at lower frequencies \citep[see Appendix in ][ for a

891: discussion on this issue]{Altamirano08}.

892: 

893: Interval Q is an average of three single observations (see

894: Table~\ref{table:averages}) that individually do not show significant

895: QPOs ( $Q>2$) at high frequencies although low-Q power excess can be

896: measured. $L_h$ in interval Q is only $2.6\sigma$ significant (single

897: trial) but required for a stable fit.

898: 

899: 

900:  Interval O shows a broad component at $30.5\pm6.3$~Hz and a 3.3\% rms

901: low-frequency noise. In this case, the very low frequency noise was fitted

902: with a broad Lorentzian because a power law gave an unstable fit. 

903: The excess of power at

904: $\nu\gtrsim1000$~Hz is less than $3\sigma$ significant.

905: Interval P shows a power spectrum with a power-law  low

906: frequency noise, and two Lorentzian components at frequencies

907: $15.5\pm1.6$ and $30.5\pm6.3$~Hz (see Figure~\ref{fig:pds}). In this

908: case the high $\chi^2/dof = 218 / 163$ reveals that the Lorentzians do

909: not satisfactorily fit the data.  As can be seen in

910: Figure~\ref{fig:pds}, there is a steep decay of the power above

911: $\nu\sim35$~Hz and power excess at $\sim70$~Hz. A fit with three

912: Lorentzians becomes unstable.  To further investigate this, we

913: refitted the power spectrum using instead two Gaussians and one power

914: law to fit the power at $\nu\lesssim40$~Hz, and one Lorentzian to fit

915: the possible extra component at $\sim70$~Hz.  The steeper Gaussian

916: function better fits the steep power decay than the Lorentzians. In

917: this fit, with three more free parameters, we obtain a

918: $\chi^2/dof=188/160$,  and the

919: Lorentzian at $69.1\pm2.6$~Hz becomes $3.4\sigma$ single-trial

920: significant. This power spectrum is very similar to those reported by

921: \citet{Migliari04,Migliari05} and \citet{Migliari05} for the atoll

922: sources 4U~1820--30 and Ser X-1, respectively. Besides the similarity

923: in shape of the power spectrum, it is interesting to note that in both

924: cases these authors found a best fit with components at similar

925: frequencies to the ones we observe in Terzan~2 and

926: %$\lesssim15$, $\lesssim30$ and $\lesssim70$~Hz 

927: which they interpreted as $\nu_b$, $\nu_h$ and $\nu_{hHz}$. This

928: coincidence suggests that the sources were in very similar states. To

929: our knowledge, no systematic study of this state has been reported as

930: yet.

931: 

932: 

933: 

934: Interval R consists of only one observation (90058-06-05-00) of

935: $\sim1.4$ ksec of data.  In addition to a power law with index

936: $\alpha=3.1\pm0.7$, we detect one Lorentzian at $68.5\pm7.2$~Hz. Its

937: frequency is rather high if we compare it with the other power spectra

938: presented in this paper and even when compared with results in other

939: sources (see Figure~\ref{fig:nuvsnu}). This result might be due to 

940: blending of components due to the low statistics present in this power

941: spectrum.

942: 

943: From the 99 pointed observations, only observation 90058-06-04-00

944: was not included in any of the averages described above. The averaged

945: colors of this observation are similar to those of Interval N

946: (observations 90058-06-01-00, 90058-06-02-00) but

947: the power spectrum does not show a significant QPO at $\simeq560$~Hz and

948: can be fitted with a single Lorentzian at $\nu_{max}=23.6\pm3.3$~Hz,

949: $Q=0.5\pm0.2$ and rms$=13.2\pm0.9$\%.  The residuals of the fit show

950: excess power at $\simeq800$~Hz but no significant kHz QPO. 

951: %

952: Although the colors of observation 90058-06-04-00 are similar from

953: those of interval N, the power spectrum is different. Because of this,

954: we refrained from averaging 90058-06-04-00 into interval N.

955: 

956: 

957: 

958: \begin{figure}[!hbtp] 

959: \center

960: \resizebox{1\columnwidth}{!}{\rotatebox{-90}{\includegraphics{./f9.eps}}}

961: \caption{Fractional rms (\%) amplitude versus hard color normalized to the Crab

962: Nebula as explained in Section~\ref{sec:dataanalysis}.  The light-grey

963: squares represent the data of the ``quiet'' state while the dark-grey

964: circles represent the data for the ``flare'' state.  The black circles

965: represent the upper limit ($\Delta\chi^2=2.7$) for the observations

966: 80138-06-05-00 and 80138-06-06-00 which also partially sampled the

967: flare states.}

968: \label{fig:rms1000}

969: \end{figure}

970: 

971: \subsection{Integrated power}

972: 

973: In order to study the rms amplitude dependence on color and intensity,

974: we calculated the average integral power per observation between 0.1

975: and 1000~Hz.  In Figure~\ref{fig:rms1000} we show the 0.01--1000~Hz

976: averaged rms amplitude (\%), of each of the 99 observations, versus

977: its average hard color.  The observations which sample the island and

978: banana state correlate with the averaged hard color.  This type of

979: correlation has been already observed in black hole candidates and

980: neutron star systems \citep[see e.g.][]{Homan01}.  At colors harder

981: than 1.0, there are two clumps (grey squares in

982: Figure~\ref{fig:rms1000}) which can also be seen in the color-color

983: diagram (Figure~\ref{fig:ccd}).  They both correspond to observations

984: of the extreme island state (quiet state) of the source. As we show in

985: Figure~\ref{fig:pds}, the power spectral shape remains approximately

986: the same with time.  However, for a given hard color, the total rms

987: amplitude can change up to $\simeq30$\%.

988: %similar results are observed when the rms

989: %amplitude is ploted versus the soft color or the intensity of the

990: %source. 

991: No correlation with time, intensity or soft color was found and these

992: changes are seen within a small range in intensity (less than 4

993: mCrab). Similar results are also observed when the rms amplitude is

994: calculated in the 0.01--300~Hz range.

995: 

996: 

997: 

998: 

999: 

1000: 

1001: \subsection{Comparing Terzan~2 with other LMXBs}

1002: 

1003: 

1004: 

1005: \subsubsection{The flaring period}\label{sec:tfp}

1006: 

1007: In Figure~\ref{fig:pbk} we plot the frequency correlations between

1008: $L_{LF}$ and $L_{\ell}$ reported by \citet{Psaltis99} and updated by

1009: \citet{Belloni02}. In Figure~\ref{fig:nuvsnu} we plot the

1010: characteristic frequency of all components versus that of $\nu_u$ for

1011: the atoll sources 4U~0614+09, 4U~1728--34, 4U~1608--52, 4U~1636--53

1012: and Aql~X-1 and the two low luminosity bursters GS~1826--24 and

1013: SLX~1735--269 \citep{Straaten02,Straaten03,Reig04,Altamirano08}.

1014: Using these correlations to identify the highest frequencies we

1015: observe (in intervals M, N \& Q) as either $\nu_u$ or $\nu_{\ell}$

1016: presents a problem. Both interpretations give consistent results since

1017: the correlations observed are complex when $\nu_u \gtrsim600$~Hz

1018: \citep{Straaten05}.

1019: 

1020: 

1021: \begin{figure}[!hbtp] 

1022: \center

1023: \resizebox{1\columnwidth}{!}{\rotatebox{-90}{\includegraphics{./f10.eps}}}

1024: \caption{PBK relation after \citet[PBK99 -][]{Psaltis99} and

1025: \citet[BPK02 -][]{Belloni02}. The black squares at $\nu_{LF}<10$~Hz

1026: represent the data of Intervals A and C--L while the black squares at

1027: $\nu_{LF}>20$~Hz represent the data of Intervals M and N. In Interval

1028: B we do not detect $L_{LF}$ and in intervals Q--R we do not detect

1029: $L_{\ell}$.}

1030: \label{fig:pbk}

1031: \end{figure}

1032: 

1033: In recent work, \citet{Barret05a,Barret05b,Barret05c} have

1034: systematically studied the variation of the frequency, rms amplitude

1035: and the quality factor Q of the lower and upper kHz QPOs in the

1036: low-mass X-ray binaries 4U~1636--536, 4U~1608--52 and

1037: 4U~1735--44. Although Q depends on the frequency of the component,

1038: these authors show that $Q_{\ell}$ is always above $\sim30$, while

1039: $Q_u$ is generally below $\sim25$. By comparing our results for the

1040: kHz QPOs with those of \citet{Barret05a,Barret05b,Barret05c}, we find

1041: that the 5 QPOs listed in Table~\ref{table:kHz} (and averaged in

1042: intervals M and N) are relatively high-Q and hence consistent with

1043: being $L_{\ell}$, but also still consistent with being $L_u$. The

1044: quality factor of the kHz QPO found in interval Q is low ($2.4\pm0.6$)

1045: and hence this kHz QPO is probably $L_u$.

1046: 

1047: In Figure \ref{fig:rms} we plot the fractional rms amplitude of all

1048: components (except $L_{LF}$) versus $\nu_{u}$ for the 4 atoll sources

1049: 4U~0614+09, 4U~1728--34, 4U~1608--52, 4U~1636--53. With respect to the

1050: kHz QPO identification, the interpretation that we found $L_{\ell}$ in

1051: interval N is strengthened by the rms amplitudes of the two

1052: low-frequency components found in the averaged power spectrum. If the

1053: QPO we observe is not $L_{\ell}$ but $L_u$, then the low frequency QPO

1054: pairs can be identified as either $L_h$-$L_b$ or $L_b$-$L_{b2}$. For

1055: $\nu_u=586.1\pm6.6$, the pair $L_h$-$L_b$ is not consistent with what

1056: we observe for other atoll sources (Figure~\ref{fig:rms}), since $L_b$

1057: is not seen with rms amplitude as low as $3.9^{+0.8}_{-0.5}$\%. The

1058: pair $L_b$-$L_{b2}$ is not consistent with the data either, since

1059: $L_{b2}$ is always observed at $\nu_u\gtrsim800$~Hz. If the QPO is

1060: $L_{\ell}$, then based on what we observe in other well-studied

1061: sources

1062: \citep{Mendez98a,Mendez99b,Disalvo03,Barret05b,Vanderklis06,Mendez07}

1063: we expect that the frequency difference between kHz QPOs is

1064: %$\nu_u \simeq \nu_{\ell} + 300 \simeq 586 + 300 \simeq

1065: $\Delta_\nu = \nu_u - \nu_{\ell} \simeq 300$ and therefore $\nu_u

1066: \simeq \nu_{\ell} + 300 \simeq 586 + 300 \simeq 886$~Hz.

1067: %

1068:  Under this assumption (i.e. $\Delta_\nu \simeq 300$) and by using the

1069:  same reasoning as above, then only the pair $L_b$-$L_{b2}$ is

1070:  consistent with the data. In the case of interval M, only one

1071:  component is found at low frequencies with an rms amplitude of

1072:  $6.4\pm0.6$\%. This result is consistent with several interpretations

1073:  when compared with the data shown in Figure~\ref{fig:rms}. Therefore,

1074:  for interval M we cannot improve confidence in the identification of

1075:  the kHz QPO using Figure~\ref{fig:rms}.

1076: 

1077: In Figures~\ref{fig:nuvsnu}, \ref{fig:pbk} \& \ref{fig:rms} we have

1078: plotted the data for Terzan~2 based on the identifications above. As

1079: discussed later in this paper, such identifications need to be

1080: confirmed.

1081: 

1082: 

1083: 

1084: 

1085: \begin{figure*}[!hbtp] 

1086: \center

1087: \resizebox{1.5\columnwidth}{!}{\rotatebox{0}{\includegraphics{./f11.eps}}}

1088: \caption{The fractional rms amplitude of all components (except

1089: $L_{LF}$) plotted versus $\nu_{u}$.  The symbols are labeled in the

1090: plot. The data for 4U~1728--34, 4U~1608--52 and 4U~0614+09 were taken

1091: from \citet{Straaten05}. The data for 4U~1636--53 were taken from

1092: \citet{Altamirano08}. Note that for $L_{hHz}$ and $L_{h}$ of

1093: 4U~1608--52, the 3 triangles with vertical error bars which intersect

1094: the abscissa represent 95\% confidence upper limits (see

1095: \citet{Straaten03} for a discussion). The points inside the circle

1096: represent our results for $L_{vl}$ while the points inside the square

1097: represent results for $L_{low}$ (see Section~\ref{sec:discussion}

1098: for a discussion). Note that as mentioned in Section~\ref{sec:tfp},

1099: the points for intervals M and N ($\nu_u>800$~Hz) were plotted under

1100: the assumption that $\nu_u = \nu_{\ell} + 300$~Hz. }

1101: \label{fig:rms}

1102: \end{figure*}

1103: 

1104: 

1105: 

1106: \subsubsection{The quiet period}

1107: 

1108: 

1109: 

1110: In Figure \ref{fig:rms} we show that the rms amplitude of $L_u$,

1111: $L_{\ell ow}$, $L_h$ and $L_b$ in Terzan~2 approximately follow the

1112: trend observed for other sources. Since \citet{Psaltis99} interprets

1113: $L_{\ell ow}$ and $L_{\ell}$ as the same component in different source

1114: states, in Figure~\ref{fig:rms} we plot the data for both components

1115: together. Of course, in Figure~\ref{fig:rms} and

1116: Figure~\ref{fig:nuvsnu} as well, there is the well-known gap between

1117: these two components.

1118: %

1119: Regarding lower frequency components, the point inside the circle

1120: represents our result for $L_{vl}$ in interval C, which is the weakest

1121: component found in the EIS of Terzan~2.  We plotted our results with

1122: those for $L_{b2}$. This component might be related with the VLFN

1123: Lorentzian \citep[see][]{Schnerr03,Reerink05}.

1124: %

1125: \citet{Altamirano08} showed that the rms amplitude of $L_{LF}$ of

1126: several atoll sources does not correlate with $\nu_u$.  There is also

1127: no evidence for a correlation in the case of Terzan~2 (not plotted).

1128: 

1129: 

1130: 

1131: 

1132: 

1133: 

1134: 

1135: 

1136: \subsection{Luminosity estimates from spectral fitting}

1137: 

1138: 

1139: We fitted the PCA and the two cluster's HEXTE spectra simultaneously

1140: using a model consisting of a black body to account for the soft

1141: component of the spectra and a power law to account for the hard

1142: component.  In some cases, it was necessary to add a Gaussian to take

1143: into account the iron K$\alpha$ line \citep[6.4 keV, see

1144: e.g.][]{White86}.  We ignore energies below 2.5 and above 25keV for

1145: the PCA spectra, and below 20 and above 200 keV for the HEXTE spectra

1146: \citep[see e.g.][]{Barret00}.  In most cases, it is not possible to

1147: well constrain the interstellar absorption $n_H$ if we lack spectral

1148: information below 2.5 keV. We therefore opted to fix $n_H$ to the

1149: value $n_H=1.2\cdot10^{22}$ H atoms cm$^{-2}$ (in the Wisconsin cross

1150: section wabs model -- see \citealt{Morrison83} ) based on previous

1151: ASCA/BeppoSAX results \citep[]{Olive98,Barret99,Barret00}.

1152: 

1153: 

1154: Assuming a distance of 6.6~kpc, we found that all 14 observations have

1155: luminosities between $\sim0.4$ and $\sim1.35 \times 10^{37}$ erg

1156: s$^{-1}$ in the energy range 2--20 keV. We also found that at high

1157: energies (20--200 keV) the luminosities of most observations were less

1158: than $0.09 \times 10^{37}$ erg s$^{-1}$. The exceptions are the three

1159: observations which sample the island state, which show 20--200~keV

1160: luminosities of $\sim0.16$, $\sim0.23$ and $\sim0.29 \times 10^{37}$

1161: erg s$^{-1}$ (observations 90058-06-03-00, 90058-06-06-00 and

1162: 91050-07-01-00, respectively).

1163: %

1164: This may be compared with observations of the brightest interval of

1165:  the quiet period of Terzan~2, which have averaged luminosities

1166:  $L_{1-20 keV} = 0.81\times10^{37}$ erg s$^{-1}$ and $L_{20-200 keV} =

1167:  0.48 \times10^{37}$ erg s$^{-1}$ \citep{Barret00}. Clearly, in

1168:  between the flares the luminosity can drop to similarly low values as

1169:  in the quiet period. This is consistent with the lightcurve we show

1170:  in Figure~\ref{fig:lc}. We note that the observations studied by

1171:  \citet{Barret00} correspond to MJDs $50391-50395$ (November 4-8,

1172:  1996) and sample the brightest part of the quiet period of this

1173:  source observed with RXTE (see Figure~\ref{fig:lc}).

1174: 

1175: We are particularly interested in the luminosity at which the kHz QPOs

1176: are detected in Terzan~2 compared with other sources. \citet{Ford00}

1177: have measured simultaneously the properties of the energy spectra and

1178: the frequencies of the kHz QPOs in 15 low-mass X-ray binaries covering

1179: a wide range of X-ray luminosities. The observations of intervals M

1180: and N (see Table~\ref{table:flares}) have average luminosities

1181: $L_{2-50 keV} / L_{edd}$ between 0.025 and 0.04 (where

1182: $L_{edd}=2.5\times10^{38}$ erg s$^{-1}$).  The three observations

1183: that sample the island state (interval Q) have average luminosity

1184: $L_{2-50 keV} / L_{edd} \simeq 0.02$. This means that during the flares,

1185: Terzan~2 shows kHz QPOs at similar luminosities to the atoll sources

1186: Aql X-1, 4U~1608--52, 4U~1702--42 and 4U~1728--34 \citep[see figure 1

1187: in ][]{Ford00}. 

1188: 

1189: \subsection{Lomb Scargle Periodograms}\label{sec:resultslsp}

1190: 

1191: During the quiet period (51214--52945 MJD), we found no significant

1192: periodicities using either  the Lomb-Scargle or the PDM techniques in

1193: the full data set nor in sub-intervals.

1194: 

1195: 

1196: 

1197: As shown in Figure~\ref{fig:lc}, the flares seem to occur every

1198: $\sim60-100$ days. Both Lomb-Scargle and the PDM techniques confirm

1199: this with a significant signal of period $P\sim90.55$ days.  In

1200: Figure~\ref{fig:folded} we show the PCA lightcurve (top) versus a

1201: 20-bin 90.55 days period folded lightcurve (bottom). The folded

1202: lightcurve matches the occurrence of most of the flares. However, it

1203: is clear that the flares are not strictly periodic. For example, F3

1204: seems to occur later and F6 occur earlier than expected. Furthermore,

1205: it is not possible to say if F7 is an early or late flare, or even a

1206: blend of two flares (we observed a small flare which peaked at

1207: $\simeq53926$~MJD, followed by a big one which peaked at

1208: $\simeq53958$~MJD).  Although there are gaps in the data,

1209: Figure~\ref{fig:folded} suggests that some flares do not occur at all

1210: (see arrow in this figure).

1211: \begin{figure}[!hbtp] 

1212: \center

1213: \resizebox{1\columnwidth}{!}{\rotatebox{0}{\includegraphics{./f12.eps}}}

1214: \caption{The PCA monitoring observation lightcurve of Terzan~2 in the

1215: time interval 53013--54050 MJD (\textit{above}) versus a series of

1216: folded light curves with period of 90.55 days (\textit{below}). The

1217: black arrow shows the position of a possible missed flare (see

1218: Section~\ref{sec:lightcurve} }

1219: \label{fig:folded}

1220: \end{figure}

1221: 

1222: 

1223: \section{Discussion}\label{sec:discussion}

1224: 

1225: 

1226: 

1227: 

1228: 

1229: 

1230: 

1231: \subsection{Contamination by a second source in the same field of view?}\label{sec:samesource}

1232: 

1233: 

1234: As shown in Figure~\ref{fig:lc}, the luminosity of the source slowly

1235: decreases with time during the quiet period 51214--52945 MJD. Although

1236: the rms amplitude changes up to 30\% (see Figure~\ref{fig:rms1000}),

1237: the X-ray timing characteristics are very similar (see Interval A to L

1238: in Figure~\ref{fig:pds}). During the 53000--53700 MJD period, the

1239: source shows flares which show different X-ray timing characteristics

1240: consistent with the island and banana states observed in other atoll

1241: sources \citep[see

1242: e.g.][]{Straaten03,Straaten05,Belloni02,Altamirano05,Altamirano08}.

1243: %

1244: A possible mechanism of the observed flux variations in Terzan~2 could

1245: be the emergence of a second X-ray source in this globular

1246: cluster unresolved by the $1^{\circ}$ (FWHM) field of view of the PCA. 

1247: If two sources are observed simultaneously with RXTE, then

1248: we would expect to see power spectra which are a combination of the

1249: intrinsic time variability of both sources. 

1250: %

1251: 

1252: To further investigate this, we compared the absolute rms amplitude

1253: that we observe both in the quiet and flaring states.

1254: Observation 80105-10-01-00 is the last observation performed during

1255: the quiet period from which we measured an average source countrate of

1256: $\sim200$ counts / second. The integrated power between $7.8 \cdot

1257: 10^{-3}$ and $1$~Hz is $(3.4\pm2) 10^{-2}$, which corresponds to a

1258: fractional rms amplitude of $\sim18\pm0.6$\%, i.e. an absolute rms

1259: amplitude of $36\pm1$ counts/second.

1260: 

1261: 

1262: Observation 80138-06-01-00 is the first observation performed during

1263: the flaring state. Its average countrate is $\sim740$ counts / second

1264: and the absolute rms amplitude in the $0.0078-1$~Hz range is

1265: $21.4\pm4.4$ counts/second ($2.9\pm0.6$\% fractional rms amplitude).

1266: Clearly, the absolute rms amplitudes are different when comparing

1267: quiet and flaring periods.  If we repeat the analysis using the second

1268: RXTE observation during the flaring state (80138-06-02-00), the

1269: discrepancy is higher. This observation has an average source

1270: countrate of $\sim405$ counts/second and the upper limit for the

1271: absolute rms amplitude in the $7.8 10^{-3}-1$~Hz frequency range is 5

1272: counts/second.

1273: 

1274: 

1275: 

1276: 

1277: 

1278: Given the characteristics of the power spectra, flares cannot be

1279:  explained by assuming that another source has emerged, unless

1280:  Terzan~2 turned off at the same time that the other X-ray source

1281:  turned on, which is unlikely. Therefore, we conclude that the flux

1282:  transitions are intrinsic to the only low mass X-ray binary detected

1283:  in the globular cluster Terzan~2: 1E~1724--30 \citep{Revnivtsev02}.

1284: 

1285: 

1286: 

1287: 

1288: 

1289: 

1290: \subsection{The kilohertz QPOs, different states and their transitions}\label{sec:kHz2}

1291: 

1292: The results presented in this paper show that the low luminosity

1293: source 1E~1724--3045 in the globular cluster Terzan~2 can be

1294: identified as an atoll source. This is the first time a source

1295: previously classified as weak burst source \citep[see e.g,][ and

1296: references within]{Belloni02,Straaten03} showed other states than

1297: those of the extreme island state, confirming previous suggestions

1298: that these sources are atoll sources. We have identified the new

1299: states as the island and banana states based on comparisons between

1300: color color diagrams of different sources and the characteristics of

1301: the power spectra. We have detected at least one of the the kHz QPOs,

1302: and as explained in

1303: Sections~\ref{sec:flaringsegment}~and~\ref{sec:kHz}, in 5 cases we may

1304: be detecting the lower kHz QPO (intervals M and N -- see also

1305: Table~\ref{table:kHz}) and in one case the upper one (interval Q).

1306: %

1307: No simultaneous twin kHz QPOs were detected within any of the 14

1308: observations that sample the flares. Future observations of flares

1309: will allow us to confirm these identifications and might allow us to

1310: detect both kHz QPOs simultaneously.

1311: 

1312: We found that the frequencies of the various components in the power

1313: spectra of Terzan~2 followed previously reported relations

1314: (Figures~\ref{fig:nuvsnu}~and~~\ref{fig:pbk}).  Terzan~2 is a

1315: particularly important source in the context of these frequency

1316: correlations because it is one of the few neutron star sources

1317: that has been demonstrated to show power spectral features that reach

1318: frequencies as low as $\simeq0.1$~Hz, which is uncommon for

1319: neutron star low mass X-ray binaries, but not for black holes.

1320: %(see Figure~\ref{fig:pbk}

1321: %- \citealt{Psaltis99}).

1322: %

1323: Our results demonstrate that in each of the flares, Terzan~2 undergoes

1324: flux transitions that, if directly observed, would probably allow us

1325: to resolve current ambiguities in the identification of components,

1326: such as the case of $L_{\ell ow}$ component in atoll sources. This

1327: component is interpreted by some authors as a broad lower kHz QPO at

1328: very low frequencies \citep[see][]{Psaltis99,Nowak00,Belloni02} which

1329: becomes peaked at higher frequencies, while other authors interpret

1330: $L_{\ell}$ and $L_{\ell ow}$ as different components \citep[see

1331: e.g. discussion in ][]{Straaten03}. Another example is the

1332: identification of the upper kHz QPO at low

1333: frequencies. \citet{Straaten03} have suggested that the broad

1334: component observed at $\simeq150$~Hz in the EIS of atoll sources

1335: becomes the peaked upper kHz QPO $L_u$. These authors based their

1336: interpretation on the frequency correlations shown in

1337: Figure~\ref{fig:nuvsnu}. Nevertheless, as \citet{Straaten03} argue,

1338: these identifications should be taken as tentative.  One way to

1339: confirm the link between them would be to observe the gradual

1340: transformation from one to another one. 

1341: 

1342: During the time between flares, the source shows intensities similar

1343: to those measured before the quasi-periodic flares started.

1344: Unfortunately there are no observations during those intervals, but we

1345: expect that then Terzan~2 shows X-ray variability similar to that

1346: reported in intervals A--L.  If this is the case, the state transition

1347: between the extreme island state and the island state should be

1348: observable in observations at the beginning or at the end of each

1349: flare. Given the relatively gradual and predictable transitions,

1350: Terzan~2 becomes the best source known up to now to study these

1351: important transitions.

1352: 

1353: 

1354: 

1355: 

1356: 

1357: 

1358: \subsection{On the $\sim90$ days flare recurrence }\label{sec:lightcurve}

1359: 

1360: 

1361: The quasi-periodic variations over days and months observed in some

1362: LMXBs X-ray light-curves are generally associated with the possible

1363: precession period of a tilted accretion disk or alternatively long

1364: term periodic variations in the accretion rate or periodic outbursts

1365: of X-ray transients.

1366: %

1367: Some examples are the $\simeq35$ cycle in Her~X-1 which is thought to

1368: be caused by a varying obscuration of the neutron star by a

1369: tilted-twisted precessing accretion disc; the $\simeq170$ days

1370: accretion cycle of the atoll source 4U~1820--30,

1371: \citep{Priedhorsky84,Simon03}; the 122-125 day cycle in the outbursts

1372: of the recurrent transient Aql X-1 \citep{Priedhorsky84b,Kitamoto93}.

1373: %

1374: Understanding the mechanisms that trigger the long-term variability

1375: associated with variations in the accretion rate of LMXBs can allow us

1376: to better predict, within each source, when the state transitions

1377: occur. This is useful because these transitions are usually

1378: fast and therefore difficult to observe.

1379: 

1380: 

1381: The power spectra of our observations of Terzan~2 during the flaring

1382: confirm that the source undergoes EIS-IS-LLB-LB-UB state transitions,

1383: as observed in other neutron star atoll systems \citep[and not as seen

1384: for Z-sources, see reviews by][ and references

1385: within]{Vanderklis04,Vanderklis06}. As the source increased in X-ray

1386: luminosity, we found that the components in the power spectra

1387: increased in frequency which is consistent with the interpretation

1388: that the accretion disk is moving inwards toward the compact

1389: object. Therefore, the flaring with average 90~days period is most

1390: probably an accretion cycle.

1391: %

1392: We note that the modulation of the light-curve could be related to the

1393: orbital period of the system or set by the precession of a tilted

1394: disk.  However, the mechanisms involved in those interpretations are

1395: very unlikely to affect the frequency of the kHz QPOs.

1396: 

1397:  

1398: If the flares are explained as an accretion cycle, then it is puzzling

1399: why the source underwent a smooth decrease of $L_x$ for

1400: $\simeq8$~years before it started to show the flares. Terzan~2 may not

1401: be the only source that shows this kind of behavior. For example,

1402: KS~1731--260 is a low-luminosity burster that has shown a high $L_x$

1403: phase, during which \citet{Revnivtsev03} reported a possible

1404: $\simeq38$ days period, and a low $L_x$ phase, during which much

1405: stronger variability was observed (which was described as red

1406: noise). After its low $L_x$ phase, KS~1731--260 has turned into

1407: quiescence \citep{Wijnands02a,Wijnands02b}. In Figure~\ref{fig:ks1731}

1408: we show the bulge scan light curve of the source during the low $L_x$

1409: phase. At MJD~$\sim51550$ the source reached very low intensities,

1410: then flared up again for $\gtrsim 250$~days to finally turn into

1411: quiescence. The low luminosities are confirmed by the ASM light curve

1412: (not plotted).

1413: %

1414: Recently, \citet{Shih05} reported that

1415: the persistent atoll source 4U~1636--53 has also shown a period of

1416: high $L_x$ followed by a period of low $L_x$. During high $L_x$, no

1417: long-term periodicity was found, but a highly significant

1418: $\simeq46$~days period was observed after its $L_x$ decline.

1419: 

1420: These  similar patterns of behavior 

1421: might point towards a common mechanism, which then must be unaffected

1422: by the intrinsic differences between these sources.

1423: %

1424: 

1425: For example, while Terzan~2 remained with approximately constant

1426: luminosity in its extreme island state for $\simeq8$ years before

1427: showing long term periodicities, 4U~1636--53 and KS~1731--260 were

1428: observed with variable luminosity and in different states, including

1429: the banana state in which the kHz QPOs were found (see

1430: e.g. \citealt{Wijnands97b} and \citealt{Shih05} for 4U~1636--53 and

1431: \citealt{Wijnands97a} and \citealt{Revnivtsev03} for

1432: KS~1731--260). While Terzan~2 reached a maximum luminosity of

1433: $L_x/L_{Edd}\simeq0.02$ during one of the flares,

1434: % KS~1731--260 has

1435: %turned into quiescence \citep{Wijnands02a,Wijnands02b} and 

1436: 4U~1636--53 shows similar luminosities only at its lowest $L_x$ levels

1437: \citep[while it has reached $L_x/L_{Edd} \gtrsim0.15$ --

1438: see][]{Altamirano08}.  Further differences may be related to whether

1439: these systems are normal or ultra-compact binaries. While 4U~1636--53

1440: is not ultra-compact (see below), \cite{Intzand07} has recently

1441: proposed that Terzan~2 may be classified as ultra-compact based on

1442: measurements of its persistent flux, long burst recurrence times and

1443: the hard X-ray spectra.

1444: %

1445: If the luminosity behavior of these sources is related, the

1446: differences outlined above suggest that the mechanism that triggers

1447: the modulation of the light curve at low $L_x$ may not depend on the

1448: accretion history, the luminosity of the source or even whether the

1449: system is ultra-compact or not.  The modulation period may depend on

1450: these factors.

1451: %

1452: 

1453: 

1454: 

1455: \begin{figure}[!hbtp] 

1456: \center

1457: \resizebox{1\columnwidth}{!}{\rotatebox{0}{\includegraphics{./f13.eps}}}

1458: \caption{The PCA monitoring observation lightcurve of the atoll source

1459:   KS~1731--260 during part of its low $L_x$ period. Unfortunately,

1460:   there are no PCA monitoring observations of the source before to

1461:   MJD~51200. Clearly, the source flares up similarly to Terzan~2

1462:   before it turns into quiescence. Interestingly, the data at MJD

1463:   $\sim51550$ shows that the source had a period of very low

1464:   intensity, followed by a flaring up that lasted for $\sim250$~days

1465:   before the source finally turned into quiescence.  }

1466: \label{fig:ks1731}

1467: \end{figure}

1468: 

1469: 

1470: 

1471: Unfortunately, we cannot compare the orbital periods and the

1472: companions of the three systems, as these are only known for

1473: 4U~1636--53 \citep[$\sim3.8$~hours and

1474: $\simeq0.4$~M$_{\odot}$][]{Casares06}. Nevertheless, with the present

1475: data it is already possible to exclude some mechanisms.

1476: %

1477: For example, mass transfer feedback induced by X-ray irradiation

1478: \citep{Osaki85} is unlikely.  In this model, X-ray radiation from the

1479: compact object heats the companion star surface, causing enhancement

1480: of the mass accretion rate in a runaway instability. However, in

1481: \Citeauthor{Osaki85}'s scenario, it is not clear how the system could

1482: remember the phase of the cycle if one of the flares is missed or if

1483: the size of the flares differs much.  Flares F4 and F5 in Terzan~2,

1484: independently of the other two sources, may already raise an objection

1485: to this model.  Although we miss part of F4 due to a gap in the data,

1486: Figure~\ref{fig:lc} shows that F4 was quite short (less than 9 days),

1487: while F5 was the longest ($\lesssim36$~days) and strongest flare.

1488: 

1489: 

1490: \citet{Shih05} have suggested that the atoll source 4U~1636--53 may

1491: turn into quiescence after its low $L_x$ period, as was observed for

1492: KS~1731--260. Such an observation for 4U~1636--53 as well as for

1493: Terzan~2 would give credibility to the link between these sources.  To

1494: our knowledge, there is no model which predicts such behavior.

1495: 

1496: \subsection{Energy dependence as a tool for kHz QPO identification}

1497: 

1498: \citet{Homan00} discovered a single 695~Hz QPO in the low mass X-ray

1499: binary EXO~0748--676 and identified this QPO as the lower kHz

1500: QPO. These authors based their identification on the fact that at that

1501: time: (i) from the 11 kilohertz QPO pairs found in atoll sources, eight

1502: had ranges of lower peak frequencies that include 695 Hz, which was the

1503: case for only three of the upper peaks and (ii) the upper peaks in atoll

1504: sources generally had Q lower than $\sim14$, while their QPO had

1505: Q$\gtrsim38$, value more common for lower peaks.

1506: %

1507: While from Figure~\ref{fig:nuvsnu} it can be seen that (i) is not

1508: valid anymore, since the upper kHz QPOs have been detected down to

1509: 300--400~Hz (and possibly down to $\lesssim100$~Hz -- see

1510: Section~\ref{pdsqp}),

1511: %

1512: at these low frequencies $L_u$ is usually much broader than they

1513: observed, which confirms their identification (see

1514: Section~\ref{sec:tfp} and \citealt{Barret05a,Barret05b,Barret05c}).

1515: 

1516: \citet{Homan00} also based their identification on the comparison of

1517: the energy dependence of the QPO with that of the two kilohertz peaks

1518: in 4U~1608--52, which have rather different energy dependencies

1519: \citep[the power-law rms-energy relation for $L_{\ell}$ is steeper

1520:   than that for $L_u$, see 

1521: ][]{Berger96,Mendez98a,Mendez01}. Similarly, \citet{Mendez01} use the

1522: same method to strengthen the identification of the single kHz QPO

1523: observed in the atoll source Aql~X-1. To further investigate if this

1524: method could be used to identify the sharp kHz QPOs we report in

1525: Section~\ref{sec:kHz}, in Figure~\ref{fig:rmsvse} we compare the

1526: energy dependence of the kHz QPOs in 4U~1608--52

1527: \citep{Berger96,Mendez98a,Mendez01} and 4U~0614+09 \citep{Mendez97}

1528: with that of Terzan~2.  The data for Terzan~2 seem to fall in between

1529: those for $L_{\ell}$ and $L_u$ of 4U~1608--52 but shows a completely

1530: different behavior than the data of 4U~0614+09.

1531: %

1532: The fact that the rms amplitude of the upper kHz QPO in 4U~0614+09 and

1533: 4U~1608--52 are significantly different (by up to a factor of 3) and

1534: that the data for Terzan~2 fall in between those of $L_{\ell}$ and

1535: $L_u$ in 4U~1608--52 show that the method does not lead to unambiguous

1536: results. Mean source luminosity, instantaneous luminosity and

1537: instantaneous QPO frequency may all affect QPO energy dependence in

1538: addition to QPO type.

1539: %

1540: %

1541: 

1542: 

1543: 

1544: \section{Summary}

1545: 

1546: 

1547: \begin{enumerate}

1548: 

1549: \item We presented a detailed study of the time variability of the

1550: atoll source 1E~1724--3045 (Terzan~2) which includes, for the first

1551: time, observations of this source in its island and banana states

1552: confirming the atoll nature of this source.  We find that the

1553: different states of Terzan~2 show timing behavior similar to that seen

1554: in other NS-LMXBs. Our results for the extreme island state are

1555: consistent with those previously reported in \citet{Belloni02} and

1556: \citet{Straaten03}.

1557: 

1558: \item We report the discovery of kilohertz quasi-periodic oscillations

1559: (kHz QPOs). Although we do not detect two kHz QPOs simultaneously nor

1560: significant variability above 800~Hz, the detection of the lower and

1561: the upper kHz QPOs at different epochs and the power excess found at

1562: high frequencies (such as the case in intervals~O or observation

1563: 90058-06-04-00) suggest that simultaneous twin kHz phenomena as well

1564: as significant variability up to $\sim1100$~Hz (or more) is 

1565: probable.

1566: 

1567: \item By comparing the dependence of the rms amplitude with energy of

1568: kHz QPOs in the atoll sources 4U~1608--52, 4U~0614+09 and Terzan~2, we

1569: show that this dependence appears to differ between sources and

1570: therefore cannot be used to unambiguously identify the kilohertz QPOs

1571: in either $L_u$ or $L_{\ell}$, as previously thought.

1572: 

1573: 

1574: \item We studied the flux transitions or flares observed since

1575: February 2004 and from the source state changes observed we conclude

1576: that they are due to aperiodic changes in the accretion rate.

1577: 

1578: 

1579: 

1580: \item State transitions between the extreme island state and the

1581: island state should be observable in observations at the beginning or

1582: at the end of each flare. Given the relatively gradual and predictable

1583: transitions, Terzan~2 becomes the best source known upto now to study

1584: such transitions.

1585: 

1586: \end{enumerate}

1587: 

1588: 

1589: 

1590: 

1591: 

1592: %

1593: 

1594: \center

1595: \begin{longtable}{|c|c|c|c|}

1596: %{\footnotesize

1597: \hline

1598: Characteristic & Q & rms & Component \\

1599: Frequency (Hz) &   & (\%) & ID \\

1600: \hline

1601: \endfirsthead

1602: \hline

1603: Characteristic & Q & rms & Component \\

1604: Frequency (Hz) &   & (\%) & ID \\

1605: \hline

1606: \endhead

1607: \hline \multicolumn{4}{|r|}{{Continued on next page}} \\ \hline

1608: \endfoot

1609: \hline \hline

1610: \endlastfoot

1611: \hline \multicolumn{4}{|c|}{Interval A} \\ \hline

1612: $   156.9\pm      6.6$ & $     0.25\pm      0.06$ & $    13.9\pm      0.4$ &$L_{u}$\\ 

1613: $    11.5\pm      0.2$ & $     0.04\pm      0.03$ & $    17.6\pm      0.3$ &$L_{low}$\\ 

1614: $     1.05\pm      0.01$ & $     0.47\pm      0.02$ & $    16.3\pm      0.3$ &$L_{h}$\\ 

1615: $     0.820\pm      0.006$ & $     5.76\pm      0.87$ & $     3.3\pm      0.3$ &$L_{LF}$\\ 

1616: $     0.171\pm      0.005$ & $     0.32\pm      0.06$ & $    12.4\pm      0.6$ &$L_{b}$\\ 

1617: %$     0.042\pm      0.012$ & $     --$ & $     5.5\pm      0.9$ &$L_{vl}$\\ 

1618: \hline \multicolumn{4}{|c|}{Interval B} \\ \hline

1619: $   200.1\pm     11.6$ & $     0.55\pm      0.09$ & $    12.6\pm      0.5$ &$L_{u}$\\ 

1620: $    19.1\pm      1.5$ & $     --$                & $    13.2\pm      0.2$ &$L_{low}$\\ 

1621: $     2.59\pm      0.05$ & $     0.40\pm      0.04$ & $    13.6\pm      0.4$ &$L_{h}$\\ 

1622: $     0.40\pm      0.02$ & $     --$              & $    12.2\pm      0.2$ &$L_{b}$\\ 

1623: %$     0.020\pm      0.003$ & $     0.59\pm      0.33$ & $     3.6\pm      0.5$ &$L_{vl}$\\ 

1624: \hline \multicolumn{4}{|c|}{Interval C} \\ \hline

1625: $   148.4\pm      6.8$ & $     0.27\pm      0.05$ & $      13.8\pm      0.3$ &$L_{u}$\\ 

1626: $     9.7\pm      0.1$ & $     --$                & $      18.6\pm      0.1$ &$L_{low}$\\ 

1627: $     0.86\pm      0.01$ & $     0.55\pm      0.02$ & $    15.1\pm      0.3$ &$L_{h}$\\ 

1628: $     0.633\pm      0.007$ & $     6.2\pm      1.2$ & $     3.1\pm      0.3$ &$L_{LF}$\\ 

1629: $     0.158\pm      0.006$ & $     --$              & $    14.5\pm      0.3$ &$L_{b}$\\ 

1630: $     0.106\pm      0.006$ & $     1.94\pm      1.02$ & $   3.4^{+0.8}_{-0.5}$ &$L_{vl}$\\ 

1631: \hline \multicolumn{4}{|c|}{Interval D} \\ \hline

1632: $   130.8\pm      7.4$ & $     0.21\pm      0.07$ & $    13.1\pm      0.3$ &$L_{u}$\\ 

1633: $     8.5\pm      0.20$ & $     --$               & $    18.5\pm      0.1$ &$L_{low}$\\ 

1634: $     0.74\pm      0.01$ & $     0.53\pm      0.03$ & $  14.5\pm      0.3$ &$L_{h}$\\ 

1635: $     0.512\pm      0.004$ & $     6.5\pm      0.9$ & $   3.9\pm      0.2$ &$L_{LF}$\\ 

1636: $     0.131\pm      0.005$ & $     --$              & $  15.4\pm      0.2$ &$L_{b}$\\ 

1637: \hline \multicolumn{4}{|c|}{Interval E} \\ \hline

1638: $   169.2\pm      8.6$ & $     0.31\pm      0.07$ & $    13.6\pm      0.4$ &$L_{u}$\\ 

1639: $    10.8\pm      0.2$ & $     0.01\pm      0.04$ & $    18.2\pm      0.3$ &$L_{low}$\\ 

1640: $     0.92\pm      0.01$ & $     0.45\pm      0.03$ & $    16.6\pm      0.3$ &$L_{h}$\\ 

1641: $     0.750\pm      0.006$ & $     7.8\pm      1.8$ & $     2.8\pm      0.2$ &$L_{LF}$\\ 

1642: $     0.144\pm      0.004$ & $     0.34\pm      0.07$ & $    12.7\pm      0.6$ &$L_{b}$\\ 

1643: %$     0.024\pm      0.016$ & $     --$                  & $     4.39\pm      1.09$ &$L_{vl}$\\ 

1644: \hline \multicolumn{4}{|c|}{Interval F} \\ \hline

1645: $   216.1\pm      2.2$ & $    17.9\pm      8.4$ & $     3.1\pm      0.5$ &$***$\\ 

1646: $   161.4\pm     17.0$ & $     0.14\pm      0.13$ & $    14.2\pm      0.8$ &$L_{u}$\\ 

1647: $    12.1\pm      0.5$ & $     0.10\pm      0.07$ & $    16.6\pm      0.6$ &$L_{low}$\\ 

1648: $     1.21\pm      0.04$ & $     0.44\pm      0.04$ & $    15.6\pm      0.5$ &$L_{h}$\\ 

1649: $     0.93\pm      0.01$ & $     3.5\pm      0.9$ & $     4.4\pm      0.6$ &$L_{LF}$\\ 

1650: $     0.179\pm      0.009$ & $     0.09\pm      0.04$ & $    13.8\pm      0.3$ &$L_{b}$\\ 

1651: \hline \multicolumn{4}{|c|}{Interval G} \\ \hline

1652: $    17.5\pm      2.4$ & $     --$                  & $    19.6\pm      0.5$ &$L_{low}$\\ 

1653: $     1.17\pm      0.07$ & $     0.3\pm      0.1$ & $    16.4\pm      1.1$ &$L_{h}$\\ 

1654: $     0.716\pm      0.008$ & $    11.2^{24.5}_{3.6} $ & $     4.1\pm      0.6$ &$L_{LF}$\\ 

1655: $     0.152\pm      0.016$ & $     --$                  &  $    16.4\pm      0.7$ &$L_{b}$\\ 

1656: \hline \multicolumn{4}{|c|}{Interval H} \\ \hline

1657: $   131.7\pm     15.8$ & $     0.3\pm      0.1$ & $    13.2\pm      0.8$ &$L_{u}$\\ 

1658: $     9.6\pm      0.4$ & $     --$                  & $    18.5\pm      0.2$ &$L_{low}$\\ 

1659: $     0.892\pm      0.026$ & $     0.47\pm      0.04$ & $    14.5\pm      0.4$ &$L_{h}$\\ 

1660: $     0.634\pm      0.006$ & $     4.6\pm      0.8$ & $     4.5\pm      0.4$ &$L_{LF}$\\ 

1661: $     0.133\pm      0.005$ & $     --$                  & $    16.3\pm      0.2$ &$L_{b}$\\ 

1662: \hline \multicolumn{4}{|c|}{Interval I} \\ \hline

1663: $   211.7\pm     40.3$ & $     --$                  & $    19.4\pm      0.9$ &$L_{u}$\\ 

1664: $     9.05\pm      0.39$ & $     0.17\pm      0.07$ & $    18.8\pm      0.6$ &$L_{low}$\\ 

1665: $     0.87\pm      0.02$ & $     0.47\pm      0.06$ & $    17.6\pm      0.7$ &$L_{h}$\\ 

1666: $     0.67\pm      0.01$ & $     8.2\pm      3.3$ & $     3.2\pm      0.4$ &$L_{LF}$\\ 

1667: $     0.136\pm      0.009$ & $     0.13\pm      0.05$ & $    16.2\pm      0.5$ &$L_{b}$\\ 

1668: \hline \multicolumn{4}{|c|}{Interval J} \\ \hline

1669: $    86.4\pm     32.8$ & $     --$                  & $    14.7\pm      1.6$ &$L_{u}$\\ 

1670: $     9.71\pm      1.07$ & $     0.19\pm      0.13$ & $    17.3\pm      1.5$ &$L_{low}$\\ 

1671: $     0.8\pm      0.1$ & $     --$                  & $    17.6\pm      1.6$ &$L_{h}$\\ 

1672: $     0.60\pm      0.02$ & $     1.5\pm      0.4$ & $     8.8^{+2.2}_{-1.2}$ &$L_{LF}$\\ 

1673: $     0.097\pm      0.008$ & $     0.27\pm      0.08$ & $    14.4\pm      0.9$ &$L_{b}$\\ 

1674: \hline \multicolumn{4}{|c|}{Interval K} \\ \hline

1675: $   122.5\pm     18.6$ & $     0.04\pm      0.3$ & $    16.7\pm      1.2$ &$L_{u}$\\ 

1676: $     7.3\pm      0.5$ & $     --$                  & $    18.8\pm      0.4$ &$L_{low}$\\ 

1677: $     0.79\pm      0.03$ & $     0.54\pm      0.06$ & $    14.0\pm      0.6$ &$L_{h}$\\ 

1678: $     0.556\pm      0.006$ & $     5.1\pm      1.5$ & $     4.3\pm      0.5$ &$L_{LF}$\\ 

1679: $     0.134\pm      0.006$ & $     --$                  & $    17.2\pm      0.3$ &$L_{b}$\\ 

1680: %\hline \multicolumn{4}{|c|}{Interval L} \\ \hline

1681: %$   816.823\pm     35.840$ & $     3.516\pm      1.744$ & $     9.757\pm      2.087$ &$L_{u}$\\ 

1682: %$   582.645\pm     11.161$ & $     5.076\pm      2.388$ & $     9.416\pm      1.737$ &$L_{u}$\\ 

1683: %$   126.055\pm     18.384$ & $     1.339\pm      0.815$ & $     7.182\pm      1.254$ &$L_{hHz}$\\ 

1684: %$    26.105\pm      1.333$ & $     0.900\pm      0.220$ & $     9.273\pm      0.955$ &$L_{b}$\\ 

1685: %$     5.743\pm      8.009$ & $     0.236\pm      0.468$ & $     3.946\pm      2.296$ &$L_{b2}$\\ 

1686: \hline \multicolumn{4}{|c|}{Interval L} \\ \hline

1687: $   202.7\pm     94.6$ & $     --$ & $    13.04\pm      2.08$ &$L_{u}$\\ 

1688: $    16.3\pm      1.8$ & $     --$ & $    18.3\pm      0.6$ &$L_{low}$\\ 

1689: $     1.34\pm      0.05$ & $     0.48\pm      0.07$ & $    15.5\pm      0.6$ &$L_{h}$\\ 

1690: $     0.98\pm      0.01$ & $    10.06\pm      8.60$ & $     3.3^{+0.6}_{-0.5}$ &$L_{LF}$\\ 

1691: $     0.19\pm      0.01$ & $     --$ & $    16.1\pm      0.4$ &$L_{b}$\\ 

1692: \hline \multicolumn{4}{|c|}{Interval M} \\ \hline

1693: $   758.9\pm      4.8$ & $    12.2^{+12}_{-4}$ & $     9.1\pm      0.7$ &$L_{\ell}$\\ 

1694: $    39.4\pm      2.5$ & $     1.3\pm      0.6$ & $     6.4\pm      0.6$ &$L_{b}$\\ 

1695: \hline \multicolumn{4}{|c|}{Interval N} \\ \hline

1696: $   586.1\pm      6.6$ & $     5.5\pm      1.24$ & $     10.6\pm      0.8$ &$L_{\ell}$\\ 

1697: $    29.1\pm      1.1$ & $     1.1\pm      0.2$ & $     8.4\pm      0.5$ &$L_{b}$\\ 

1698: $     4.9\pm      1.6$ & $     0.47\pm      0.3$ & $     3.9^{+0.8}_{-0.5}$ &$L_{b2}$\\ 

1699: \hline \multicolumn{4}{|c|}{Interval O} \\ \hline

1700: $    30.5\pm      6.3$ & $     0.28\pm      0.17$ & $     5.8\pm      0.4$ &$***$\\ 

1701: $     0.025\pm      0.002$ & $     0.3\pm      0.1$ & $     3.3\pm      0.1$ &$***$\\ 

1702: \hline \multicolumn{4}{|c|}{Interval P} \\ \hline

1703: $    30.9\pm      0.4$ & $     2.02\pm      0.21$ & $    12.6\pm      0.6$ &$***$\\ 

1704: $    15.5\pm      1.6$ & $     1.005\pm      0.156$ & $     7.9\pm      0.9$ &$***$\\ 

1705: \hline \multicolumn{4}{|c|}{Interval Q} \\ \hline

1706: $   565.0\pm     24.5$ & $     2.4\pm      0.6$ & $    15.10\pm      1.44$ &$L_{u}$\\ 

1707: $   148.7\pm     23.3$ & $     0.7\pm      0.4$ & $    12.4\pm      1.8$ &$L_{hHz}$\\ 

1708: $    28.02\pm      2.37$ & $     1.06\pm      0.56$ & $     9.4^{+2.7}_{-1.8}$ &$L_{h} ^{ **} $\\ 

1709: $     8.1\pm      1.4$ & $     0.10\pm      0.08$ & $    15.1\pm      1.2$ &$L_{b}$\\ 

1710: \hline \multicolumn{4}{|c|}{Interval R} \\ \hline

1711: $    68.5\pm      7.2$ & $     1.8\pm      1.1$ & $     5.5\pm      0.9$ &$***$\\ 

1712: \hline

1713: \caption{Characteristic frequencies $\nu_{max}$, $Q$ values

1714: ($\nu_{max} \equiv \nu_{central}/FWHM$), fractional rms (in the full

1715: PCA energy band) and component identification (ID) of the Lorentzians

1716: fitted for Terzan~2.  The quoted errors use $\Delta\chi^2 =

1717: 1.0$. Where only one error is quoted, it is the straight average

1718: between the positive and the negative error. ``$--$'': Broad

1719: Lorentzians whose quality factor was fixed to 0. \\ $^{**}$: This

1720: component is not $3\sigma$ significant. However, without this

1721: component the fit becomes unstable (see Section~\ref{sec:gralpds}). \\ $***$:

1722: For this component, no clear identification was possible. }

1723: \label{table:data}

1724: %the end of small font

1725: %}

1726: \end{longtable}

1727: 

1728: %\begin{table}[!hbtp]

1729: %\center

1730: %\begin{tabular}{|c|c|c|c|}\hline 

1731: %Obsid          & $L_x$ (2--20 kev)  & $L_x$  (20--200 kev)             & $L_x$    (2--50 kev)         \\

1732: %               &  ($10^{37}$ ergs s$^{-1}$)      & ($10^{37}$ ergs s$^{-1}$)       & ($10^{37}$ ergs s$^{-1}$)       \\

1733: %\hline

1734: %80138-06-01-00 &   1.24            &    0.06             &  1.28             \\

1735: %80138-06-02-00 &   0.89            &    0.04             &  0.92             \\

1736: %80138-06-02-01 &   0.92            &    0.04             &  0.95             \\

1737: %80138-06-03-00 &   0.89            &    0.03             &  0.91             \\

1738: %80138-06-04-00 &   1.01            &    0.04             &  1.04             \\

1739: %80138-06-05-00 &   1.06            &    0.02             &  1.07             \\

1740: %80138-06-06-00 &   1.14            &    0.04             &  1.17             \\

1741: %90058-06-01-00 &   1.01            &    0.09             &  1.07             \\

1742: %90058-06-02-00 &   0.82            &    0.06             &  0.86             \\

1743: %90058-06-03-00 &   0.45            &    0.16             &  0.53             \\

1744: %90058-06-04-00 &   0.49            &    0.09             &  0.54             \\

1745: %90058-06-05-00 &   1.34            &    0.06             &  1.39             \\

1746: %90058-06-06-00 &   0.52            &    0.29             &  0.66             \\

1747: %91050-07-01-00 &   0.40            &    0.23             &  0.50             \\

1748: %\hline

1749: %\end{tabular}

1750: %\caption{ALMOST SURE I ERASE YOU!}

1751: %\label{table:spectra}

1752: %\end{table}

1753: 

1754: 

1755: 

1756: \clearpage

1757: 

1758: %\bibliographystyle{aa}

1759: %\bibliography{../../Most_complete_bib}

1760: 

1761: \begin{thebibliography}{77}

1762: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi

1763: \expandafter\ifx\csname url\endcsname\relax

1764:   \def\url#1{{\tt #1}}\fi

1765: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi

1766: 

1767: \bibitem[{{Altamirano} et~al.(2005){Altamirano}, {van der Klis}, {M{\'e}ndez}

1768:   et~al.}]{Altamirano05}

1769: {Altamirano} D., {van der Klis} M., {M{\'e}ndez} M., et~al., 2005, \apj, 633,

1770:   358

1771: 

1772: \bibitem[{{Altamirano} et~al.(2008){Altamirano}, {van der Klis}, {M{\'e}ndez}

1773:   et~al.}]{Altamirano08}

1774: {Altamirano} D., {van der Klis} M., {M{\'e}ndez} M., et~al., Jun. 2008, ArXiv

1775:   e-prints, 806

1776: 

1777: \bibitem[{{Barret} et~al.(1999){Barret}, {Grindlay}, {Harrus}, \&

1778:   {Olive}}]{Barret99}

1779: {Barret} D., {Grindlay} J.E., {Harrus} I.M., {Olive} J.F., 1999, \aap, 341, 789

1780: 

1781: \bibitem[{{Barret} et~al.(2000{\natexlab{a}}){Barret}, {Olive}, {Boirin}

1782:   et~al.}]{Barret00}

1783: {Barret} D., {Olive} J.F., {Boirin} L., et~al., 2000{\natexlab{a}}, \apj, 533,

1784:   329

1785: 

1786: \bibitem[{{Barret} et~al.(2000{\natexlab{b}}){Barret}, {Olive}, {Boirin}

1787:   et~al.}]{Barret00a}

1788: {Barret} D., {Olive} J.F., {Boirin} L., et~al., 2000{\natexlab{b}}, Advances in

1789:   Space Research, 25, 383

1790: 

1791: \bibitem[{{Barret} et~al.(2005{\natexlab{a}}){Barret}, {Klu{\'z}niak}, {Olive},

1792:   {Paltani}, \& {Skinner}}]{Barret05a}

1793: {Barret} D., {Klu{\'z}niak} W., {Olive} J.F., {Paltani} S., {Skinner} G.K.,

1794:   2005{\natexlab{a}}, \mnras, 357, 1288

1795: 

1796: \bibitem[{{Barret} et~al.(2005{\natexlab{b}}){Barret}, {Olive}, \&

1797:   {Miller}}]{Barret05b}

1798: {Barret} D., {Olive} J.F., {Miller} M.C., 2005{\natexlab{b}}, \mnras, 361, 855

1799: 

1800: \bibitem[{{Barret} et~al.(2005{\natexlab{c}}){Barret}, {Olive}, \&

1801:   {Miller}}]{Barret05c}

1802: {Barret} D., {Olive} J.F., {Miller} M.C., 2005{\natexlab{c}}, Astronomische

1803:   Nachrichten, 326, 808

1804: 

1805: \bibitem[{{Belloni} et~al.(2002){Belloni}, {Psaltis}, \& {van der

1806:   Klis}}]{Belloni02}

1807: {Belloni} T., {Psaltis} D., {van der Klis} M., 2002, \apj, 572, 392

1808: 

1809: \bibitem[{{Belloni} et~al.(2005){Belloni}, {M{\'e}ndez}, \&

1810:   {Homan}}]{Belloni05a}

1811: {Belloni} T., {M{\'e}ndez} M., {Homan} J., 2005, \aap, 437, 209

1812: 

1813: \bibitem[{{Berger} et~al.(1996){Berger}, {van der Klis}, {van Paradijs}

1814:   et~al.}]{Berger96}

1815: {Berger} M., {van der Klis} M., {van Paradijs} J., et~al., 1996, \apjl, 469,

1816:   L13+

1817: 

1818: \bibitem[{{Casares} et~al.(2006){Casares}, {Cornelisse}, {Steeghs}

1819:   et~al.}]{Casares06}

1820: {Casares} J., {Cornelisse} R., {Steeghs} D., et~al., Dec. 2006, \mnras, 373,

1821:   1235

1822: 

1823: \bibitem[{{Di Salvo} et~al.(2001){Di Salvo}, {M{\'e}ndez}, {van der Klis},

1824:   {Ford}, \& {Robba}}]{Disalvo01}

1825: {Di Salvo} T., {M{\'e}ndez} M., {van der Klis} M., {Ford} E., {Robba} N.R.,

1826:   2001, \apj, 546, 1107

1827: 

1828: \bibitem[{{Di Salvo} et~al.(2003){Di Salvo}, {M{\' e}ndez}, \& {van der

1829:   Klis}}]{Disalvo03}

1830: {Di Salvo} T., {M{\' e}ndez} M., {van der Klis} M., 2003, \aap, 406, 177

1831: 

1832: \bibitem[{{Emelyanov} et~al.(2002){Emelyanov}, {Revnivtsev}, {Aref'ev}, \&

1833:   {Sunyaev}}]{Emelyanov02}

1834: {Emelyanov} A.N., {Revnivtsev} M.G., {Aref'ev} V.A., {Sunyaev} R.A., 2002,

1835:   Astronomy Letters, 28, 12

1836: 

1837: \bibitem[{{Ford} et~al.(2000){Ford}, {van der Klis}, {M{\' e}ndez}

1838:   et~al.}]{Ford00}

1839: {Ford} E.C., {van der Klis} M., {M{\' e}ndez} M., et~al., 2000, \apj, 537, 368

1840: 

1841: \bibitem[{{Galloway} et~al.(2006){Galloway}, {Muno}, {Hartman}

1842:   et~al.}]{Galloway06}

1843: {Galloway} D.K., {Muno} M.P., {Hartman} J.M., et~al., 2006, ArXiv Astrophysics,

1844:   astro-ph/0608259

1845: 

1846: \bibitem[{{Grindlay} et~al.(1980){Grindlay}, {Marshall}, {Hertz}

1847:   et~al.}]{Grindlay80}

1848: {Grindlay} J.E., {Marshall} H.L., {Hertz} P., et~al., 1980, \apjl, 240, L121

1849: 

1850: \bibitem[{{Hasinger} \& {van der Klis}(1989)}]{Hasinger89}

1851: {Hasinger} G., {van der Klis} M., 1989, \aap, 225, 79

1852: 

1853: \bibitem[{{Homan} \& {van der Klis}(2000)}]{Homan00}

1854: {Homan} J., {van der Klis} M., 2000, \apj, 539, 847

1855: 

1856: \bibitem[{{Homan} et~al.(2001){Homan}, {Wijnands}, {van der Klis}

1857:   et~al.}]{Homan01}

1858: {Homan} J., {Wijnands} R., {van der Klis} M., et~al., 2001, \apjs, 132, 377

1859: 

1860: \bibitem[{{in't Zand} et~al.(2007){in't Zand}, {Jonker}, \&

1861:   {Markwardt}}]{Intzand07}

1862: {in't Zand} J.J.M., {Jonker} P.G., {Markwardt} C.B., 2007, \aap, 465, 953

1863: 

1864: \bibitem[{{Jahoda} et~al.(2006){Jahoda}, {Markwardt}, {Radeva}

1865:   et~al.}]{Jahoda06}

1866: {Jahoda} K., {Markwardt} C.B., {Radeva} Y., et~al., 2006, \apjs, 163, 401

1867: 

1868: \bibitem[{{Jonker} et~al.(2007){Jonker}, {in't Zand}, {M{\'e}ndez}, \& {van der

1869:   Klis}}]{Jonker07}

1870: {Jonker} P.G., {in't Zand} J.J.M., {M{\'e}ndez} M., {van der Klis} M., 2007,

1871:   \mnras, 378, 1187

1872: 

1873: \bibitem[{{Kitamoto} et~al.(1993){Kitamoto}, {Tsunemi}, {Miyamoto}, \&

1874:   {Roussel-Dupre}}]{Kitamoto93}

1875: {Kitamoto} S., {Tsunemi} H., {Miyamoto} S., {Roussel-Dupre} D., 1993, \apj,

1876:   403, 315

1877: 

1878: \bibitem[{{Klein-Wolt}(2004)}]{Kleinwolt04}

1879: {Klein-Wolt} M., 2004, PhD.Thesis

1880: 

1881: \bibitem[{{Kuulkers} et~al.(1994){Kuulkers}, {van der Klis}, {Oosterbroek}

1882:   et~al.}]{Kuulkers94}

1883: {Kuulkers} E., {van der Klis} M., {Oosterbroek} T., et~al., 1994, \aap, 289,

1884:   795

1885: 

1886: \bibitem[{{Kuulkers} et~al.(2003){Kuulkers}, {den Hartog}, {in't Zand}

1887:   et~al.}]{Kuulkers03}

1888: {Kuulkers} E., {den Hartog} P.R., {in't Zand} J.J.M., et~al., 2003, \aap, 399,

1889:   663

1890: 

1891: \bibitem[{{Linares} et~al.(2005){Linares}, {van der Klis}, {Altamirano}, \&

1892:   {Markwardt}}]{Linares05}

1893: {Linares} M., {van der Klis} M., {Altamirano} D., {Markwardt} C.B., Dec. 2005,

1894:   \apj, 634, 1250

1895: 

1896: \bibitem[{{Linares} et~al.(2007){Linares}, {van der Klis}, \&

1897:   {Wijnands}}]{Linares07}

1898: {Linares} M., {van der Klis} M., {Wijnands} R., May 2007, \apj, 660, 595

1899: 

1900: \bibitem[{{Lomb}(1976)}]{Lomb76}

1901: {Lomb} N.R., 1976, \apss, 39, 447

1902: 

1903: \bibitem[{{M{\' e}ndez} \& {van der Klis}(1999)}]{Mendez99b}

1904: {M{\' e}ndez} M., {van der Klis} M., 1999, \apjl, 517, L51

1905: 

1906: \bibitem[{{M{\' e}ndez} et~al.(2001){M{\' e}ndez}, {van der Klis}, \&

1907:   {Ford}}]{Mendez01}

1908: {M{\' e}ndez} M., {van der Klis} M., {Ford} E.C., 2001, \apj, 561, 1016

1909: 

1910: \bibitem[{{Markwardt} \& {Swank}(2004)}]{Markwardt04}

1911: {Markwardt} C.B., {Swank} J., 2004, The Astronomer's Telegram, 237, 1

1912: 

1913: \bibitem[{{M{\'e}ndez} \& {Belloni}(2007)}]{Mendez07}

1914: {M{\'e}ndez} M., {Belloni} T., Oct. 2007, \mnras, 381, 790

1915: 

1916: \bibitem[{{M{\'e}ndez} et~al.(1997){M{\'e}ndez}, {van der Klis}, {van Paradijs}

1917:   et~al.}]{Mendez97}

1918: {M{\'e}ndez} M., {van der Klis} M., {van Paradijs} J., et~al., 1997, \apjl,

1919:   485, L37+

1920: 

1921: \bibitem[{{M{\'e}ndez} et~al.(1998{\natexlab{a}}){M{\'e}ndez}, {van der Klis},

1922:   {van Paradijs} et~al.}]{Mendez98}

1923: {M{\'e}ndez} M., {van der Klis} M., {van Paradijs} J., et~al.,

1924:   1998{\natexlab{a}}, \apjl, 494, L65+

1925: 

1926: \bibitem[{{M{\'e}ndez} et~al.(1998{\natexlab{b}}){M{\'e}ndez}, {van der Klis},

1927:   {Wijnands} et~al.}]{Mendez98a}

1928: {M{\'e}ndez} M., {van der Klis} M., {Wijnands} R., et~al., 1998{\natexlab{b}},

1929:   \apjl, 505, L23+

1930: 

1931: \bibitem[{{Migliari} et~al.(2004){Migliari}, {Fender}, {Rupen}

1932:   et~al.}]{Migliari04}

1933: {Migliari} S., {Fender} R.P., {Rupen} M., et~al., 2004, \mnras, 351, 186

1934: 

1935: \bibitem[{{Migliari} et~al.(2005){Migliari}, {Fender}, \& {van der

1936:   Klis}}]{Migliari05}

1937: {Migliari} S., {Fender} R.P., {van der Klis} M., 2005, \mnras, 363, 112

1938: 

1939: \bibitem[{{Morrison} \& {McCammon}(1983)}]{Morrison83}

1940: {Morrison} R., {McCammon} D., 1983, \apj, 270, 119

1941: 

1942: \bibitem[{{Muno} et~al.(2003){Muno}, {Baganoff}, \& {Arabadjis}}]{Muno03}

1943: {Muno} M.P., {Baganoff} F.K., {Arabadjis} J.S., Nov. 2003, \apj, 598, 474

1944: 

1945: \bibitem[{{Nowak}(2000)}]{Nowak00}

1946: {Nowak} M.A., 2000, \mnras, 318, 361

1947: 

1948: \bibitem[{{Olive} et~al.(1998){Olive}, {Barret}, {Boirin} et~al.}]{Olive98}

1949: {Olive} J.F., {Barret} D., {Boirin} L., et~al., 1998, \aap, 333, 942

1950: 

1951: \bibitem[{{Ortolani} et~al.(1997){Ortolani}, {Bica}, \& {Barbuy}}]{Ortolani97}

1952: {Ortolani} S., {Bica} E., {Barbuy} B., 1997, \aap, 326, 614

1953: 

1954: \bibitem[{{Osaki}(1985)}]{Osaki85}

1955: {Osaki} Y., 1985, \aap, 144, 369

1956: 

1957: \bibitem[{{Press et al.}(1992)}]{Numerical-Recipes}

1958: {Press et al.}, 1992, Numerical Recipes: The Art of Scientific Computing,

1959:   Cambridge University Press, Cambridge (UK) and New York, 2nd edn.

1960: 

1961: \bibitem[{{Priedhorsky} \& {Terrell}(1984{\natexlab{a}})}]{Priedhorsky84}

1962: {Priedhorsky} W., {Terrell} J., 1984{\natexlab{a}}, \apjl, 284, L17

1963: 

1964: \bibitem[{{Priedhorsky} \& {Terrell}(1984{\natexlab{b}})}]{Priedhorsky84b}

1965: {Priedhorsky} W.C., {Terrell} J., 1984{\natexlab{b}}, \apj, 280, 661

1966: 

1967: \bibitem[{{Psaltis} \& {Chakrabarty}(1999)}]{Psaltis99}

1968: {Psaltis} D., {Chakrabarty} D., 1999, \apj, 521, 332

1969: 

1970: \bibitem[{{Reerink} et~al.(2005){Reerink}, {Schnerr}, {van der Klis}, \& {van

1971:   Straaten}}]{Reerink05}

1972: {Reerink} T.J., {Schnerr} R.S., {van der Klis} M., {van Straaten} S., 2005,

1973:   A\&A submitted

1974: 

1975: \bibitem[{{Reig} et~al.(2004){Reig}, {van Straaten}, \& {van der

1976:   Klis}}]{Reig04}

1977: {Reig} P., {van Straaten} S., {van der Klis} M., 2004, \apj, 602, 918

1978: 

1979: \bibitem[{{Revnivtsev} \& {Sunyaev}(2003)}]{Revnivtsev03}

1980: {Revnivtsev} M., {Sunyaev} R., 2003, \aap, 399, 699

1981: 

1982: \bibitem[{{Revnivtsev} et~al.(2002){Revnivtsev}, {Trudolyubov}, \&

1983:   {Borozdin}}]{Revnivtsev02}

1984: {Revnivtsev} M.G., {Trudolyubov} S.P., {Borozdin} K.N., 2002, Astronomy

1985:   Letters, 28, 237

1986: 

1987: \bibitem[{{Scargle}(1982)}]{Scargle82}

1988: {Scargle} J.D., 1982, \apj, 263, 835

1989: 

1990: \bibitem[{{Schnerr} et~al.(2003){Schnerr}, {Reerink}, {van der Klis}

1991:   et~al.}]{Schnerr03}

1992: {Schnerr} R.S., {Reerink} T., {van der Klis} M., et~al., 2003, \aap, 406, 221

1993: 

1994: \bibitem[{{Shih} et~al.(2005){Shih}, {Bird}, {Charles}, {Cornelisse}, \&

1995:   {Tiramani}}]{Shih05}

1996: {Shih} I.C., {Bird} A.J., {Charles} P.A., {Cornelisse} R., {Tiramani} D., 2005,

1997:   \mnras, 361, 602

1998: 

1999: \bibitem[{{Simon}(2003)}]{Simon03}

2000: {Simon} V., 2003, \aap, 405, 199

2001: 

2002: \bibitem[{{Stellingwerf}(1978)}]{Stellingwerf78}

2003: {Stellingwerf} R.F., 1978, \apj, 224, 953

2004: 

2005: \bibitem[{{Swank} \& {Markwardt}(2001)}]{Swank01}

2006: {Swank} J., {Markwardt} K., 2001, in ASP Conf. Ser. 251, New Century of X-ray

2007:   Astronomy, eds. H. Inoue \& H. Kunieda (San Francisco: ASP), 94

2008: 

2009: \bibitem[{{Swank} et~al.(1977){Swank}, {Becker}, {Boldt} et~al.}]{Swank77}

2010: {Swank} J.H., {Becker} R.H., {Boldt} E.A., et~al., 1977, \apjl, 212, L73

2011: 

2012: \bibitem[{{van der Klis}(1995)}]{Vanderklis95b}

2013: {van der Klis} M., 1995, Proceedings of the NATO Advanced Study Institute on

2014:   the Lives of the Neutron Stars, held in Kemer, Turkey, August 19-September

2015:   12, 1993. Editor(s), M. A. Alpar, U. Kiziloglu, J. van Paradijs; Publisher,

2016:   Kluwer Academic, Dordrecht, The Netherlands, Boston, Massachusetts, 301

2017: 

2018: \bibitem[{{van der Klis}(2000)}]{Vanderklis00}

2019: {van der Klis} M., 2000, \araa, 38, 717

2020: 

2021: \bibitem[{{van der Klis}(2004)}]{Vanderklis04}

2022: {van der Klis} M., 2004, in ``Compact Stellar X-ray Sources'', eds. W.H.G.

2023:   Lewin and M. van der Klis, in press.

2024: 

2025: \bibitem[{{van der Klis}(2006)}]{Vanderklis06}

2026: {van der Klis} M., 2006, in Compact Stellar X-Ray Sources, ed. W. H. G. Lewin

2027:   \& M. van der Klis (Cambridge: Cambridge Univ. Press), in press

2028: 

2029: \bibitem[{{van Straaten} et~al.(2000){van Straaten}, {Ford}, {van der Klis},

2030:   {M{\'e}ndez}, \& {Kaaret}}]{Straaten00}

2031: {van Straaten} S., {Ford} E.C., {van der Klis} M., {M{\'e}ndez} M., {Kaaret}

2032:   P., 2000, \apj, 540, 1049

2033: 

2034: \bibitem[{{van Straaten} et~al.(2002){van Straaten}, {van der Klis}, {di

2035:   Salvo}, \& {Belloni}}]{Straaten02}

2036: {van Straaten} S., {van der Klis} M., {di Salvo} T., {Belloni} T., 2002, \apj,

2037:   568, 912

2038: 

2039: \bibitem[{{van Straaten} et~al.(2003){van Straaten}, {van der Klis}, \& {M{\'

2040:   e}ndez}}]{Straaten03}

2041: {van Straaten} S., {van der Klis} M., {M{\' e}ndez} M., 2003, \apj, 596, 1155

2042: 

2043: \bibitem[{{van Straaten} et~al.(2005){van Straaten}, {van der Klis}, \&

2044:   {Wijnands}}]{Straaten05}

2045: {van Straaten} S., {van der Klis} M., {Wijnands} R., 2005, \apj, 619, 455

2046: 

2047: \bibitem[{{Wen} et~al.(2006){Wen}, {Levine}, {Corbet}, \& {Bradt}}]{Wen06}

2048: {Wen} L., {Levine} A.M., {Corbet} R.H.D., {Bradt} H.V., 2006, \apjs, 163, 372

2049: 

2050: \bibitem[{{White} et~al.(1986){White}, {Peacock}, {Hasinger} et~al.}]{White86}

2051: {White} N.E., {Peacock} A., {Hasinger} G., et~al., 1986, \mnras, 218, 129

2052: 

2053: \bibitem[{{Wijnands} et~al.(2002{\natexlab{a}}){Wijnands}, {Guainazzi}, {van

2054:   der Klis}, \& {M{\'e}ndez}}]{Wijnands02b}

2055: {Wijnands} R., {Guainazzi} M., {van der Klis} M., {M{\'e}ndez} M.,

2056:   2002{\natexlab{a}}, \apjl, 573, L45

2057: 

2058: \bibitem[{{Wijnands} et~al.(2002{\natexlab{b}}){Wijnands}, {Heinke}, \&

2059:   {Grindlay}}]{Wijnands02a}

2060: {Wijnands} R., {Heinke} C.O., {Grindlay} J.E., 2002{\natexlab{b}}, \apj, 572,

2061:   1002

2062: 

2063: \bibitem[{{Wijnands} \& {van der Klis}(1997)}]{Wijnands97a}

2064: {Wijnands} R.A.D., {van der Klis} M., 1997, \apjl, 482, L65+

2065: 

2066: \bibitem[{{Wijnands} et~al.(1997){Wijnands}, {van der Klis}, {van Paradijs}

2067:   et~al.}]{Wijnands97b}

2068: {Wijnands} R.A.D., {van der Klis} M., {van Paradijs} J., et~al., 1997, \apjl,

2069:   479, L141+

2070: 

2071: \bibitem[{{Zhang} et~al.(1993){Zhang}, {Giles}, {Jahoda} et~al.}]{Zhang93}

2072: {Zhang} W., {Giles} A.B., {Jahoda} K., et~al., 1993, In: Proc. SPIE Vol. 2006,

2073:   p. 324-333, EUV, X-Ray, and Gamma-Ray Instrumentation for Astronomy IV,

2074:   Oswald H. Siegmund; Ed., 324--333

2075: 

2076: \bibitem[{{Zhang} et~al.(1995){Zhang}, {Jahoda}, {Swank}, {Morgan}, \&

2077:   {Giles}}]{Zhang95}

2078: {Zhang} W., {Jahoda} K., {Swank} J.H., {Morgan} E.H., {Giles} A.B., 1995, \apj,

2079:   449, 930

2080: 

2081: \end{thebibliography}

2082: 

2083: 

2084: 

2085: \end{document} 

2086: 

2087: 

2088: 

2089: 

2090: 

2091: 

2092: 

2093: