1: \documentclass[preprint]{aastex}
2: %\usepackage{psfig}
3: \usepackage{emulateapj5,times,psfig}
4: \newcommand\msun {M$_{\odot}$}
5: \begin{document}
6: \title{Discovery of a kHz QPO in 2S~0918--549}
7:
8: \author{Peter G. Jonker\altaffilmark{1}, Michiel van der
9: Klis\altaffilmark{1}, Jeroen Homan\altaffilmark{1}, Mariano
10: M\'endez\altaffilmark{1,2}, Jan van Paradijs\altaffilmark{1}, Tomaso
11: Belloni\altaffilmark{3}, Chryssa Kouveliotou\altaffilmark{4,5}, Walter
12: Lewin\altaffilmark{6}, Eric C. Ford\altaffilmark{1}}
13:
14: \altaffiltext{1}{Astronomical Institute ``Anton Pannekoek'',
15: University of Amsterdam, and Center for High-Energy Astrophysics,
16: Kruislaan 403, 1098 SJ Amsterdam; peterj@astro.uva.nl,
17: michiel@astro.uva.nl, mmendez@fcaglp.fcaglp.unlp.edu.ar}
18: \altaffiltext{2}{Facultad de Ciencias Astron\'omicas y
19: Geof\'{\i}sicas, Universidad Nacional de La Plata, Paseo del Bosque
20: S/N, 1900 La Plata, Argentina}
21: \altaffiltext{3}{Osservatorio Astronomico di Brera, via Bianchi 46,
22: I-23807 Merate, Italy; belloni@merate.mi.astro.it}
23: \altaffiltext{4}{Universities Space Research Association}
24: \altaffiltext{5}{NASA Marshall Space Flight Center, ES--84,
25: Huntsville, AL 35899}
26: \altaffiltext{6}{Department of Physics and Center for Space Research,
27: Massachusetts Institute of Technology, Cambridge, MA 02138}
28:
29: \begin{abstract}
30: \noindent
31: We report the discovery of a kilohertz quasi--periodic oscillation
32: (kHz QPO) in the low--mass X--ray binary (LMXB) 2S~0918--549. The kHz
33: QPO has a frequency of 1156$\pm$9 Hz, a FWHM of 70$\pm$20 Hz, and a
34: fractional rms amplitude of 18\%$\pm$2\%. We also detected for the
35: first time a type I X--ray burst in this source. The compact object
36: must therefore be a neutron star. Finally, we were able to classify
37: the source as an atoll source exhibiting the full complement of
38: spectral/timing states of this class (island and banana). From the
39: peak burst flux an upper limit on the persistent luminosity can be
40: derived of 0.5\% of the Eddington luminosity, making 2S~0918--549 one of
41: the least luminous LMXBs showing a kHz QPO. We compare the fractional
42: rms amplitudes of the upper kHz QPO across the ensemble of LMXBs. We
43: find a strong anticorrelation with luminosity. In LMXBs with
44: luminosities $\sim$100 times lower than those of Z--sources, the
45: fractional rms amplitude is a factor $\sim$10 larger.
46:
47:
48: \end{abstract}
49:
50: \keywords{accretion, accretion disks --- stars: individual
51: (2S~0918--549) --- stars: neutron --- X-rays: stars}
52:
53: \section{Introduction}
54: \label{intro}
55: \noindent
56: Observations with the {\it Rossi X--ray Timing Explorer} (RXTE)
57: satellite have revealed the presence of several quasi--periodic
58: phenomena in the Fourier power spectra of low--mass X--ray binaries
59: (LMXBs). They occur on timescales similar to the dynamical timescale
60: near a neutron star, i.e. at frequencies around 1000 Hz. Kilohertz
61: quasi--periodic oscillations (kHz QPOs) have been discovered in more
62: than 20 sources (van der Klis et al. 1996a,b; Strohmayer, Zhang, \&
63: Swank 1996; Strohmayer et al. 1996; see for the most recent review van
64: der Klis 2000). In several LMXBs nearly coherent oscillations were
65: discovered during type I X-ray bursts (Strohmayer et al. 1996; for a
66: review see Swank 2000); these burst oscillations presumably occur at
67: frequencies close to the neutron star spin frequency (Strohmayer et
68: al. 1996). Recently, another high frequency quasi-periodic phenomenon
69: was discovered in the power spectra of three LMXBs (Jonker, M\'endez,
70: \& van der Klis 2000). \par
71: \noindent
72: The kHz QPOs are nearly always found in pairs, although in a few
73: sources so far only one kHz QPO has been found (Zhang et al. 1998;
74: Marshall \& Markwardt 1999; Homan \& van der Klis 2000). Their
75: frequencies vary over several hundred Hz on timescales of hours to
76: days; in the best studied sources, it is found that the frequency
77: separation of the twin kHz QPOs, $\nu_2-\nu_1$, decreases
78: monotonically by 50--100 Hz over the observed frequency range as the
79: frequency of the kHz QPO increases (see e.g. van der Klis et al. 1997;
80: M\'endez et al. 1998; M\'endez, van der Klis, \& van Paradijs 1998;
81: M\'endez \& van der Klis 1999); observations of the kHz QPO
82: frequencies in the other sources are consistent with this trend
83: (Jonker et al. 1998, Psaltis et al. 1998). \par
84: \noindent
85: Various models exist for these QPOs. Immediately after their discovery
86: a beat frequency model was proposed (Strohmayer et al. 1996), of which
87: the sonic--point model is the most current (Miller, Lamb, \& Psaltis
88: 1998; revised by Lamb \& Miller 2000). Later, Stella \& Vietri (1999)
89: proposed the relativistic precession model (but see Markovi\'c \& Lamb
90: 2000) and Osherovich \& Titarchuk (1999) introduced the
91: two--oscillator model. In the latter model the QPO at $\nu_1$ (the
92: lower kHz QPO) occurs at the Keplerian frequency of material orbiting
93: the neutron star, whereas in the other two models the QPO at $\nu_2$
94: (the upper kHz QPO) is the one that is Keplerian. Recently, Psaltis \&
95: Norman (2000) proposed the transition layer model in which the twin
96: kHz QPO peaks arise due to disk oscillations at frequencies close to
97: those predicted by the relativistic precession model. \par
98: \noindent
99: In this Letter, we report the discovery of a kHz QPO and a type I
100: X--ray burst in 2S~0918--549. The source turns out to be one of the
101: least luminous kHz QPO sources, however, the fractional rms amplitude
102: of the QPO is quite high. We study the dependence of the fractional
103: rms amplitudes of the upper kHz QPO on source luminosity across the
104: ensemble of LMXBs and show that they are strongly anticorrelated.
105:
106: \section{Observations and analysis}
107: \label{analysis}
108: \noindent
109: We have used observations obtained with the proportional counter array
110: (PCA; Jahoda et al. 1996) onboard the {\it RXTE} satellite (Bradt,
111: Rothschild, \& Swank 1993). A log of the observations and the average
112: source count rates at the time of the observation can be found in
113: Table~\ref{obs_log}. In total $\sim$75 ksec of data were used in our
114: analysis. Data were obtained in four different modes. The Standard 1
115: and 2 modes, which are always operational, respectively provide data
116: with a time resolution of 0.125 seconds in just one energy bin and 16
117: seconds in 129 energy bins covering the effective PCA 2 to 60 keV
118: range. Additionally, data were obtained using the mode providing the
119: highest time resolution ($\sim 1 \mu s$), combined with the full
120: energy resolution {\it RXTE} can provide (256 energy bins covering the
121: 2 to 60 keV; the GoodXenon mode). This mode saturates on count rates
122: higher than $\sim$8000 counts per second.\par
123: \noindent
124: In order to follow spectral variations we created color--color
125: diagrams (CDs) by plotting a soft color vs. a hard color. The soft
126: color was defined as the ratio between the count rates in the 3.5--6.4
127: keV and 2--3.5 keV energy band; the hard color was defined as that
128: between the 9.7--16.0 keV and 6.4--9.7 keV energy band. The data were
129: background subtracted but no dead--time corrections were applied (the
130: dead--time is less than $\sim$2\%). The colors of the 1997
131: observations were corrected for PCA gain changes, using Crab
132: observations obtained in May 1997, and May 2000. Assuming the Crab is
133: constant, observed color changes of the Crab between the observations
134: of May 1997 and May 2000 should reflect changes of the instrument's
135: spectral response. The observed changes in Crab were $\sim$6\% in the
136: soft color and $\sim$1\% in the hard color. The colors were converted
137: to those which would have been detected by the PCA on May 2000 (cf. di
138: Salvo et al. 2000). Residual color errors due to the differences
139: between the spectra of Crab and 2S~0918--549 are less than the
140: statistical errors. \par
141: \noindent
142: We calculated Fast Fourier Transforms with a Nyquist frequency of 4096
143: Hz of data segments of 16 seconds (2--60 keV). The power density
144: spectra were averaged and fitted with a function consisting of
145: Lorentzians (to represent peaks arising in the power spectrum due to
146: QPOs), a constant to account for the power due to Poisson counting
147: statistics, and an exponentially cutoff power law to represent the
148: noise component apparent at low frequencies. Errors on the fit
149: parameters were calculated using $\Delta\chi^2=1.0$ (1$\sigma$ single
150: parameter). The 95\% confidence upper limits were determined using
151: $\Delta\chi^2=2.71$.
152:
153: \section{Results}
154: \label{result}
155: \noindent
156: The color--color diagram of 2S~0918--549 (see Fig.~\ref{cd};
157: left) resembles that of an atoll source. The squares represent the
158: average of 256~s of data from observations 1--10, the filled circles
159: are 128~s averages from observation 11, and the open circles are 128~s
160: averages from observations 12 and 13. Together with the timing
161: properties (described below) it is clear that the squares represent
162: the Island state and the open and filled circles the banana branch of
163: an atoll source.
164:
165: \begin{figure*}
166: \centerline{\psfig{file=total.ps,width=20cm,angle=90}}\figcaption{{\it
167: Left:} Color--color diagram of 2S~0918--549. The data were background
168: subtracted, the colors of the 1997 observations were converted to the
169: May 2000 observations using the Crab as a reference (see text), and
170: the burst was removed. No dead--time corrections have been applied
171: (the dead--time is $\sim$2\%). The squares and circles are 256~s and
172: 128~s averages, respectively. Typical error bars are shown for each
173: category. The pattern resembles the Island state (squares) and the
174: banana branch (open and filled circles) of atoll sources (see Hasinger
175: \& van der Klis 1989). The data in which the kHz QPO was present are
176: represented by filled circles. {\it Right, top:} Rms normalized,
177: Poisson noise subtracted average power spectrum of the observations
178: 1--10 (squares). Strong band limited noise is present, establishing
179: the identification of the Island state. The solid line is the best fit
180: to the data with a cut--off powerlaw. {\it Right, bottom:} Rms
181: normalized, Poisson noise subtracted average power spectrum of the
182: observations 11--13 (circles). Banana state power law noise and a
183: peaked noise component are present. The solid line is the best fit
184: to the data with a powerlaw and a Lorentzian component. \label{cd}}
185: \end{figure*}
186:
187: \noindent
188: The average power spectrum of observations 1--10 combined (see
189: Fig.~\ref{cd}, right, top) shows a strong band limited noise
190: component, which could be described using an exponentially cutoff
191: power law with a fractional rms amplitude of 26\%$\pm$1\% (integrated
192: over 0.1--100 Hz, 2--60 keV), a power law index of 0.47$\pm$0.05, and
193: a cutoff frequeny of 21$\pm$6 Hz. We derived a 95\% confidence upper
194: limit on the presence of a kHz QPO with a full-width at half maximum
195: (FWHM) 100 Hz in the range of 300--1200 Hz of $\sim$30\% (rms). \par
196: \noindent
197: We discovered a kHz QPO at 1156$\pm$9 Hz, with a FWHM of 68$\pm$23 Hz,
198: and a fractional rms amplitude of 18\%$\pm$2\% (2--60 keV) in the
199: average power spectrum of observation 11 (see Fig.~\ref{kHz}). In
200: addition, the average power spectrum of observation 11 (Fig.~\ref{cd},
201: right, bottom) shows a peaked noise component at a frequency of
202: 25$\pm$2 Hz, with a fractional rms amplitude of 17\%$\pm$2\%, and a
203: FWHM of 22$\pm$6 Hz. To investigate the peak noise component and the
204: kHz QPO further, we also calculated power spectra using data in the
205: 4.0--18.0 keV energy band. The values measured in the 4.0--18.0 keV
206: energy band are the same within the errors, although the significance
207: of the detection of the kHz QPO is somewhat higher (6.1$\sigma$
208: compared to 5.7 $\sigma$ in the 2--60 keV band). We subdivided the
209: 4.0--18.0 keV power spectra of observation 11 in two parts. The
210: frequency of the kHz QPO increased from 1126$\pm$9 Hz in the first
211: part of observation 11 to 1218$\pm20$ Hz at the end. The source moved
212: further up the banana branch, from an average hard and soft color of
213: 0.452$\pm$0.003 and 2.49$\pm$0.01 to 0.431$\pm$0.003 and
214: 2.64$\pm$0.01, respectively. The background subtracted count rate in
215: the 4.0--18.0 keV energy band, the FWHM, and the fractional rms
216: amplitude are consistent with being the same in the two parts. The
217: properties of the peaked noise component were consistent with being
218: the same in the two selections. \par
219: \noindent
220: \begin{figure*}
221: \centerline{
222: \psfig{file=0918_pds.ps,width=6cm}}
223: \figcaption{Average 4.0--18.0 keV power density spectrum of observation 11,
224: normalized after Leahy et al. (1983), showing the kHz QPO at
225: 1156$\pm$9 Hz ($6.1\sigma$ single trial). The solid line represents
226: the best fit to the data. \label{kHz}}
227: \end{figure*}
228: \noindent
229: No power spectral components were detected in the average power
230: spectrum of observations 12 and 13 combined. We derived an upper limit
231: on the presence of a 100 Hz wide kHz QPO of $\sim$13\% rms in the
232: range between 300--1200 Hz in both the 2--60 keV and the 4.0--18.0 keV
233: energy band, and an upper limit of 5\%--6\% on the presence of a power
234: law component at frequencies in the range of 0.1 to 100 Hz in both the
235: 2--60 keV and the 4.0--18.0 keV energy band, with a power law index of
236: 1.0. \par
237: \noindent
238: During observation 11 a type I X--ray burst occurred (see
239: Fig.~\ref{lc_burst}). This is the first detection of a burst in
240: 2S~0918--549, and establishes the compact object as a neutron star. As
241: apparent in Fig.~\ref{lc_burst} the burst profile was complex, with a
242: smooth exponential decay with an e-folding time of 8.95$\pm$0.05~s
243: beginning $\sim10$ seconds after the initial rise. Due to the very
244: high count rates of more than 20000 counts per second (3 detector
245: count rate) the GoodXenon mode saturated. The total count rate
246: displayed in Fig.~\ref{lc_burst} is for the 0.125 s Standard 1 data
247: which were not affected by this saturation. During the part of the
248: decay where the count rate was less than 8000 counts per second
249: spectral analysis shows cooling of the blackbody spectral
250: component. Using the 16~s Standard 2 mode we derive a flux for the
251: 16~s bin containing the burst peak of $3.5 \times 10^{-8}$ erg
252: cm$^{-2}$ s$^{-1}$ (2-20 keV). Since the mean count rate in this 16~s
253: time bin is a factor of $\ga$ 2.5 lower than the peak count rate as
254: measured using the 1/8~s bin of the Standard 1 observations, we use
255: this factor to improve our estimation of the peak burst flux. We did
256: not correct for the deadtime effects which become important at the
257: count rates at the peak of the burst ($\sim$25\%), this provides a
258: lower limit on the peak flux. The pre--burst flux is less than $4.2
259: \times 10^{-10}$ erg cm$^{-2}$ s$^{-1}$ (2-20 keV). So, the persistent
260: emission of the source when it is on the lower banana branch is less
261: than 0.5 percent of the Eddington luminosity. Assuming the peak burst
262: flux to be isotropic and at or below the Eddington limit of $2.5
263: \times 10^{38}$ erg s$^{-1}$ we derive an upper limit to the distance
264: of 4.9 kpc.
265:
266: \begin{figure*}
267: \figurenum{3} \centerline{\psfig{file=lc_burst.ps,height=8cm}}
268: \figcaption{Lightcurve showing the profile of the type I X-ray burst
269: occurring during observation 11. The X-axis denotes the time since the
270: start of the burst. Note the complex burst profile and the large
271: increase in count rate during the burst compared with the persistent
272: emission. The count rate is that of the three detectors that were
273: operational only.}
274: \label{lc_burst}
275: \end{figure*}
276:
277:
278:
279: \section{Discussion}
280: \noindent
281: We discovered a kHz QPO at a frequency of $\sim$1156 Hz with a
282: fractional rms amplitude of nearly 20\% (2--60 keV) in the X--ray
283: emission of the LMXB 2S~0918--549. For the first time in this source a
284: type I X--ray burst was detected, establishing the nature of the
285: compact object in 2S~0918--549 as a neutron star. We derived an upper
286: limit to the persistent luminosity when the source is on the lower
287: banana branch of 0.005 $L_{Edd}$ making 2S~0918--549 one of the least
288: luminous LMXBs in which a kHz QPO has been observed. The spectral and
289: timing properties indicate that this source is a low--luminous atoll
290: source, displaying the full complement of spectral and timing states
291: (island state and a full banana branch). Similar to other kHz QPO
292: sources, e.g. M\'endez et al. (1999), the kHz QPO was only observed
293: when the source was on the lower part of the banana branch. Its
294: frequency changed from $1126\pm$8 Hz to 1218$\pm$20 Hz, when the
295: source moved further on the banana branch, similar to other atoll
296: sources (e.g. M\'endez et al. 1999). Its frequency is similar to the
297: frequency of the {\em upper} kHz QPO in other atoll sources in a
298: similar lower--banana state (e.g. di Salvo et al. 2000, M\'endez et
299: al. 1999). The frequency of the lower kHz QPO is usually lower in that
300: part of the CD. Furthermore, when both QPOs of the twin kHz QPOs are
301: detected simultaneously the frequency of the lower kHz QPO has never
302: been found at frequencies larger than $\sim$900 Hz in any LMXB (see
303: van der Klis 2000). Therefore, we conclude that this kHz QPO is most
304: likely the upper peak of the kHz QPO pair which is detected in most
305: other LMXBs. \par
306: \noindent
307: After 4U~0614+109, 2S~0918--549 is another example of a very low
308: luminous LMXB whose kHz QPOs are strong, whereas in the most luminous
309: LMBXs, the Z--sources, the kHz QPOs tend to be weak (see also the
310: review of van der Klis 2000). To further investigate this, we compared
311: the fractional rms amplitudes of the kHz QPO in 2S~0918--549 with
312: those of the upper kHz QPO in other LMXBs. In order to account for the
313: fact that in a given source the fractional rms amplitude of the kHz
314: QPO changes as its frequency changes we measured the fractional rms
315: amplitude at both the lowest and highest frequency at which the upper
316: kHz QPO is found. From extensive studies for several sources it is
317: clear that the fractional rms amplitude of the upper kHz QPO decreases
318: monotonically as the frequency of the upper peak increases
319: (e.g. Wijnands et al. 1997, Jonker et al. 1998, Wijnands et al. 1998,
320: M\'endez, van der Klis, \& Ford 2000). This ensures that the method we used
321: will yield a good indication of the range in fractional rms amplitude
322: spanned in each source. The sample we studied here contains only those
323: sources for which a significant detection of the upper kHz QPO in the
324: 5--60 keV energy range was possible (4U~0614+09; Ford et al. 1996;
325: 4U~1915--05, Boirin et al. 2000; 4U~1728--34, Strohmayer, Zhang, \&
326: Swank 1996; 4U~1608--52, M\'endez et al. 1998; 4U~1702--42, Markwardt
327: et al. 1999; 4U~1636--53, Zhang et al. 1996; 4U~1820--30, Smale et
328: al. 1996; XTE~J2123--058, Homan et al. 1999; KS~1731--260, Wijnands \&
329: van der Klis 1997), or in case of the Z-sources, the 5-60 keV
330: fractional rms amplitude that was reported in the literature
331: (Cyg~X--2, Wijnands et al. 1998; GX~17+2, Wijnands et al. 1997;
332: GX~340+0, Jonker et al. 1998). Sco~X--1 was not included since the
333: fractional rms amplitude of the QPOs can not be calculated accurately
334: at present due to the severe dead-time effects present in this
335: source. We calculated the 5--60 keV fractional rms amplitude in order
336: to avoid as much as possible systematic effects on the derived
337: fractional rms amplitudes of the peaks due to the different $N_H$
338: values towards the various sources.\par
339: \noindent
340: The range in fractional rms amplitude of the upper kHz QPO in each
341: source is plotted in Fig.~\ref{rmsen} as a function of the
342: simultaneously measured luminosity (for the distances and the method
343: used to calculate these luminosities see Ford et al. 2000 and
344: references therein; the distance to 4U~1915--05 was taken from Smale
345: et al. 1988). It is apparent that the upper kHz QPOs detected in
346: low--luminosity sources have relatively large fractional rms
347: amplitudes, when compared with the fractional rms amplitude of the kHz
348: QPO in the bright LMXBs (e.g. the Z-sources). The frequencies of the
349: end--points for each source are given in Table~\ref{freqs}. \par
350: \noindent
351: Van Paradijs \& van der Klis (1994) found that the hardness, defined
352: as the ratio of counts between the 40--80 keV band and the 13--25 keV
353: band, is in neutron star LMXBs correlated with the X--ray
354: luminosity. This suggests that the X--ray flux variability of the
355: source at the kHz QPO frequency is correlated with the hardness in
356: that band (see also Ford et al. 1997). Within a source a similar
357: correlation between the fractional rms amplitude of the upper kHz QPO
358: and the hardness of the energy spectrum has been found; the harder the
359: source the higher the fractional rms amplitude of the upper kHz QPO
360: (e.g. 4U~0614+091, Ford et al. 1997, van Straaten et al. 2000;
361: 4U~1608--52, M\'endez et al. 1999; Aql~X--1, Reig et al. 2000;
362: 4U~1728--34, Di Salvo et al. 2000; Cyg~X--2, Wijnands et al. 1998;
363: GX~17+2, Wijnands et al. 1997; GX~340+0, Jonker et al. 1998).\par
364: \noindent
365: A two--component accretion flow (radially and through a disk, Gosh and
366: Lamb 1979) has been considered to explain that the frequency of the
367: kHz QPO does not depend on source luminosity (Ford et al. 2000). The
368: observed anticorrelation between the fractional rms amplitude of the
369: upper kHz QPO and the source luminosity cannot solely be explained by
370: an increase of an unmodulated part of the accretion flow towards
371: higher source luminosity, since the increase in luminosity is too
372: large compared with the decrease in fractional rms
373: amplitude. Therefore, this cannot explain the findings of Ford et
374: al. (2000). The same conclusion was drawn by M\'endez et al. (2000) on
375: other grounds. It is important to consider the anticorrelation
376: presented in this work and the absence of a correlation reported by
377: Ford et al. (2000) in mechanisms producing kHz QPOs.
378:
379: \acknowledgments This work was supported in part by the Netherlands
380: Organization for Scientific Research (NWO) grant 614-51-002. This
381: research has made use of data obtained through the High Energy
382: Astrophysics Science Archive Research Center Online Service, provided
383: by the NASA/Goddard Space Flight Center. This work was supported by
384: NWO Spinoza grant 08-0 to E.P.J.van den Heuvel. WHGL gratefully
385: acknowledges support by NASA.
386:
387: \begin{thebibliography}{99}
388: \bibitem[Boirin et al. 1999]{boirin} Boirin, L., Barret, D., Olive,
389: J.F., Bloser, P.F., Grindlay, J.E. 2000, A\&A, 361, 121
390: \bibitem[Bradt, Rothschild \& Swank 1993]{bradtetal} Bradt, H.V.,
391: Rothschild, R.E., Swank, J.H. 1993, A\&AS, 97, 355
392: \bibitem[Di Salvo et al. 2000]{disalvo} Di Salvo, T., M\'endez, M.,
393: van der Klis, M., Ford, E.C., Robba, N.R., 2000, in press
394: \bibitem[Ford et al. 1996]{fordetal} Ford, E., Kaaret, P., Tavani, M.,
395: Harmon, B.A., Zhang, S.N., Barret, D., Bloser, P., Grindlay, J. 1996,
396: IAU Circ, No. 6426
397: \bibitem[Ford et al. 1997]{fordetal97} Ford, E., Kaaret, P., Chen, K.,
398: Tavani, M., Barret, D., Bloser, P., Grindlay, J., Harmon, B.A.,
399: Paciesas, W.S., Zhang, S.N. 1997 ApJ, 486, L47
400: \bibitem[Ford \& van der Klis 1998]{ford} Ford, E.C., van der Klis,
401: M. 1998, ApJ, 506, L39
402: \bibitem[Ford et al. 2000]{fordetal2} Ford, E.C., van der Klis, M.,
403: Wijnands, R., Homan, J., Jonker, P.G., van Paradijs, J. 2000, ApJ,
404: 537, 368
405: \bibitem[Hasinger \& van der Klis 1989]{hasinger} Hasinger, G., \&
406: van der Klis, M. 1989, A\&A, 225, 79
407: \bibitem[Homan et al. 1999]{homan} Homan, J., M\'endez, M., Wijnands,
408: R., van der Klis, M., van Paradijs, J. 1999, ApJ, 513, L119
409: \bibitem[Jahoda et al. 1996]{jahoda} Jahoda, K., Swank, J.H., Giles,
410: A.B., Stark, M.J., Strohmayer, T., Zhang, W., Morgan, E.H. 1996,
411: SPIE, 2808, 59
412: \bibitem[Jonker et al. 1998]{jonkeretal} Jonker, P.G., Wijnands, R.,
413: van der Klis, M., Psaltis, D., Kuulkers, E., Lamb, F.K. 1998, ApJ,
414: 499, L191
415: \bibitem[Jonker, M\'endez, \& van der Klis 2000]{jonkersb} Jonker,
416: P.G., M\'endez, M., van der Klis, M. 2000, ApJ, 540, L29
417: \bibitem[Lamb \& Miller 2000]{lamb} Lamb, F.K., Miller, M.C. 2000,
418: ApJ, submitted
419: \bibitem[Leahy et al. 1983]{leahy} Leahy, D.A., Darbro, W., Elsner,
420: R.F., Weisskopf, M.C., Sutherland, P.G., Kahn, S., \& Grindlay, J.E.
421: 1983, ApJ, 266, 160
422: \bibitem[Lee \& Miller 1998]{lee} Lee, H.C., Miller, G.S. 1998, MNRAS,
423: 299, 479
424: \bibitem[Markovi\'c \& Lamb 2000]{markovic} Markovi\'c, D., Lamb,
425: F.K. 2000, submitted to MNRAS
426: \bibitem[Markwardt, Strohmayer, \& Swank 1999]{markwardt} Markwardt,
427: C.B., Strohmayer, T.E., Swank, J.H. 1999, ApJ, 512, L125
428: \bibitem[Marshall \& Markwardt 1999]{marshall} Marshall, F.E.,
429: Markwardt, C.B. 1999, IAU Circ. 7103
430: \bibitem[M\'endez et al. 1998]{mendeza} M\'endez, M., van der Klis,
431: M., van Paradijs, J., Lewin, W.H.G., Vaughan, B.A., Kuulkers, E.,
432: Zhang, W., Lamb, F.K., Psaltis, D. 1998a, ApJ, 494, 65
433: \bibitem[M\'endez et al. 1998]{mendezb} M\'endez, M., van der Klis,
434: M., Wijnands, R., Ford, E.C., van Paradijs, J., Vaughan, B.A. 1998b,
435: ApJ, 505, L23
436: \bibitem[M\'endez et al. 1999]{mendezetal} M\'endez, M., van der Klis,
437: M., Ford, E.C., Wijnands, R., van Paradijs, J. 1999, ApJ, 51, L49
438: \bibitem[M\'endez \& van der Klis 1999]{mendez1999} M\'endez, M.. \&
439: van der Klis, M. 1999 ApJ, 517, 51
440: \bibitem[M\'endez, van der Klis, \& van Paradijs 1999]{mmmvkjvp1999}
441: M\'endez, M.. \& van der Klis, M., van Paradijs, J. 1998, ApJ, 506,
442: L117
443: \bibitem[M\'endez, van der Klis, Ford, Wijnands \& van Paradijs
444: 1999]{mendezetal1999} M\'endez, M.. \& van der Klis, M., Ford, E.C.,
445: Wijnands, R., van Paradijs, J. 1999, ApJ, 511, L49
446: \bibitem[M\'endez, van der Klis, Ford 2000]{mendezklisford} M\'endez,
447: M., van der Klis, M., Ford, E.C. 2000, ApJ, submitted, astro-ph/0006245
448: \bibitem[Miller, Lamb, \& Psaltis 1998]{millersp} Miller, M.C., Lamb,
449: F.K., Psaltis, D. 1998, ApJ, 508, 791
450: \bibitem[Osherovich \& Titarchuk 1999]{osherovich} Osherovich, V.,
451: Titarchuk, L. 1999, ApJ, 522, L113
452: \bibitem[Psaltis et al. 1998]{psaltisetal98} Psaltis, D., M\'endez,
453: M., Wijnands, R., Homan, J., Jonker, P.G., van der Klis, M., Lam,
454: F.K., Kuulkers, E., van Paradijs, J., Lewin, W.H.G. 1998, ApJ, 501, 95
455: \bibitem[Psaltis, Belloni, van der Klis 1999]{pbk} Psaltis, D.,
456: Belloni, T., van der Klis, M. 1999, ApJ, 520, 262
457: \bibitem[Psaltis \& Norman 2000]{psalnor} Psaltis, D., Norman,
458: C. 2000, ApJ, submitted
459: \bibitem[Reig, P. et al. 2000]{reig} Reig, P., M\'endez, M., van der
460: Klis, M., \& Ford, E.C. 2000, ApJ, 530, 916
461: \bibitem[Smale, Zhang, \& White 1996]{smale} Smale, A.P., Zhang, W.,
462: White, N.E. 1996, IAU Circ. No. 6507
463: \bibitem[Stella, \& Vietri (1998)]{stella98} Stella, L., Vietri,
464: M. 1998, ApJ, 492, L59
465: \bibitem[Stella, \& Vietri (1999)]{stella99} Stella, L., Vietri,
466: M. 1999, Phys. Rev. Lett, 82, 17
467: \bibitem[Strohmayer, Zhang, \& Swank 1996]{strohmayeriau} Strohmayer, T.E.,
468: Zhang, W., Swank, J.H. 1996, IAU Circ. 6320
469: \bibitem[Strohmayer et al. 1996]{strohmayer} Strohmayer, T.E.,
470: Zhang, W., Swank, J.H., Smale, A., Titarchuk, L., \& Day, C. 1996,
471: ApJ, 469, L9
472: \bibitem[Swank 2000]{swank} Swank, J.H. 2000, ``Cosmic Explosions'',
473: Proceedings of the 10th October Astrophysics in Maryland, AIP
474: Conf. Proceedings 522, ed. S.S. Holt \& W.W. Zhang, (AIP: Woodbury,
475: N. Y.), 2000, astro-ph 0005498
476: \bibitem[van der Klis et al. 1996a]{michieliau} van der Klis, M., Swank,
477: J., Zhang, W., Jahoda, L., Morgan, E.H., Lewin, W.H.G., Vaughan, B.,
478: van Paradijs, J. 1996a, IAU Circ. 6319
479: \bibitem[van der Klis et al. 1996b]{michiel} van der Klis, M., Swank,
480: J.H., Zhang, W., Jahoda, L., Morgan, E.H., Lewin, W.H.G., Vaughan, B.,
481: van Paradijs, J. 1996b, ApJ, 469
482: \bibitem[van der Klis 1997]{scomichiel} van der Klis, M., Wijnands,
483: R.A.D., Horne, K., \& Chen, W. 1997, ApJ, 481, L97
484: \bibitem[van der Klis 2000]{aaarmichiel} van der Klis, M. 2000,
485: A\&AAR, 38, 717
486: \bibitem[van Paradijs \& van der Klis 1994]{vanparadijs} van Paradijs,
487: J., van der Klis, M. 1994, A\&A, 281, L17
488: \bibitem[van Paradijs et al. 1996]{vanparadijsetal} van Paradijs,
489: J., et al. 1996, IAU Circ. No. 6336
490: \bibitem[van Straaten et al. 2000]{steve} van Straaten, S., Ford,
491: E.C., van der Klis, M., M\'endez, M., Kaaret, P. 2000, ApJ, 540, 1049
492: \bibitem[Wijnands \& van der Klis 1997]{wijnands} Wijnands, R., \& van
493: der Klis, M. 1997, ApJ, 482, L65
494: \bibitem[Wijnands et al. 1997]{rudy} Wijnands, R., Homan, J., van der Klis,
495: M\'endez, M., Kuulkers, E., van Paradijs, J., Lewin, W.H.G., Lamb,
496: F.K., Psaltis, D., Vaughan, B. 1997, ApJ, 490, L157
497: \bibitem[Wijnands et al. 1998]{rudy2} Wijnands, R., Homan, J., van der Klis,
498: M., Kuulkers, E., van Paradijs, J., Lewin, W.H.G., Lamb, F.K.,
499: Psaltis, D., Vaughan, B. 1998, ApJ, 493, L87
500: \bibitem[Zhang et al. 1996]{zhang} Zhang, W., Lapidus, I., White,
501: N.E., Titarchuk, L. 1996, ApJ, 473, L135
502: \bibitem[Zhang et al. 1998]{zhang1} Zhang, W., Jahoda, K., Kelley,
503: R.L., Strohmayer, T.E., Swank, J.H. 1998, ApJ, 495, L9
504: \end{thebibliography}
505:
506: \begin{deluxetable}{ccccc}
507: \tablecaption{Log of the observations of 2S~0918--549 used in this
508: analysis. The count rates in column 5 are the average, background
509: subtracted, 2--60 keV count rates obtained as if 5 PCUs were always
510: operational. \label{obs_log}}
511:
512: \tabletypesize{\normalsize}
513: \tablecolumns{5}
514: \tablewidth{0pc}
515: \tablehead{
516: \colhead{} & \colhead{Observation} & \colhead{Date \&} &
517: \colhead{Amount of data} & \colhead{Average count rate}\\
518: \colhead{} & \colhead{ID} & \colhead{Start time
519: (UTC)} & \colhead{(ksec)} & \colhead{cnts/sec (2--60 keV)}}
520: \startdata
521: 1 & 20071-11-01-02 & 02-05-1997 23:57 & $\sim$4.2 & $\sim$36\\
522: 2 & 20071-11-01-03 & 03-05-1997 02:15 & $\sim$2.1 & $\sim$36\\
523: 3 & 20071-11-01-04 & 03-05-1997 06:42 & $\sim$1.8 & $\sim$45\\
524: 4 & 20071-11-01-05 & 03-05-1997 08:26 & $\sim$1.4 & $\sim$50\\
525: 5 & 20071-11-01-06 & 04-05-1997 08:27 & $\sim$1.5 & $\sim$53\\
526: 6 & 20071-11-01-00 & 04-05-1997 22:41 & $\sim$11.3 & $\sim$47\\
527: 7 & 20071-11-01-01 & 05-05-1997 22:16 & $\sim$6.0 & $\sim$48\\
528: 8 & 20071-11-01-07 & 09-05-1997 10:10 & $\sim$2.4 & $\sim$54\\
529: 9 & 20064-06-01-00 & 15-08-1997 01:38 & $\sim$4.8 & $\sim$84\\
530: 10 & 20064-06-02-00 & 21-09-1997 22:51 & $\sim$5.0 & $\sim$68\\
531: 11 & 50060-01-01-00 & 12-05-2000 12:24 & $\sim$30.0 & $\sim$135\\
532: 12 & 50060-01-01-01 & 14-05-2000 19:43 & $\sim$2.1 & $\sim$187\\
533: 13 & 50060-01-01-02 & 15-05-2000 04:59 & $\sim$2.4 & $\sim$208
534: \enddata
535: \end{deluxetable}
536:
537: \begin{figure*}
538: \figurenum{4}
539: \centerline{\psfig{file=lumi_vs_rms_arrow.ps,height=8cm}}
540: \figcaption{Luminosity versus fractional rms amplitude (5--60 keV) of
541: the upper kHz QPO in various LMXBs. Apparent is the decrease in
542: fractional rms amplitude as the source luminosity increases, both
543: within one source and between sources.}
544: \label{rmsen}
545: \end{figure*}
546:
547: \begin{deluxetable}{lll}
548: \tablecaption{The lowest and highest frequency of the upper kHz QPO
549: for each source used to create Fig. 4. \label{freqs}}
550:
551: \tabletypesize{\normalsize}
552: \tablecolumns{5}
553: \tablewidth{0pc}
554: \tablehead{
555: \colhead{Source name} & \colhead{Lowest frequency} & \colhead{Highest
556: frequency} \\
557: \colhead{} & \colhead{upper kHz (Hz)} & \colhead{upper kHz (Hz)}}
558: \startdata
559: 2S~0918--549 & 1156$\pm$9 & 1156$\pm$9\\
560: 4U~0614+09 & 421$\pm$13 & 1161$\pm$5\\
561: 4U~1915--05 & 542$\pm$14 & 1013$\pm$8\\
562: 4U~1608--52 & 883$\pm$4 & 1091$\pm$8\\
563: 4U~1728--34 & 552$\pm$7 & 1139$\pm$13\\
564: 4U~1702--42 & 1000$\pm$9 & 1059$\pm$11\\
565: 4U~1636--53 & 1152$\pm$4 & 1213$\pm$11\\
566: 4U~1820--30 & 675$\pm$4 & 1050$\pm$4\\
567: XTE~J2123--058& 1129$\pm$8 & 1129$\pm$8\\
568: KS~1731--260 & 1176$\pm$3 & 1176$\pm$3\\
569: Cyg~X--2 & 731$\pm$19 & 1007$\pm$15\\
570: GX~17+2 & 645$\pm$9 & 1086$\pm$8\\
571: GX~340+0 & 535$\pm$70 & 840$\pm$20
572: \enddata
573: \end{deluxetable}
574:
575: \end{document}
576:
577: