astro-ph0001134/ms.tex
1: %\documentstyle[12pt,aasms4]{article} 
2: %\documentstyle[11pt,aaspp4]{article} 
3: \documentstyle[aas2pp4]{article} 
4:  
5: \def\rem#1{{\bf #1}} 
6: \def\simless{\mathbin{\lower 3pt\hbox 
7:      {$\rlap{\raise 5pt\hbox{$\char'074$}}\mathchar"7218$}}}   %< or of order 
8: \def\simmore{\mathbin{\lower 3pt\hbox 
9:      {$\rlap{\raise 5pt\hbox{$\char'076$}}\mathchar"7218$}}}   %> or of order 
10: \def\Msun{{\rm M}_\odot}                                       % solar masses 
11: \def\msun{{\rm M}_\odot}                                       % solar masses 
12:  
13: \received{} 
14: \accepted{} 
15: %\journalid{337}{15 January 1989} 
16: %\articleid{11}{14} 
17: \slugcomment{Submitted to Ap.J, 8 Jan 2000} 
18:  
19: \lefthead{Reig et al.} 
20: \righthead{Lags in GRS 1915+105} 
21:  
22: \begin{document} 
23:  
24: \title{Phase lag variability associated with the 0.5--10 Hz QPO \\
25: in GRS~1915+105} 
26:  
27: \author{ 
28: P.~Reig\altaffilmark{1,2},
29: T. Belloni\altaffilmark{3},
30: M.~van~der~Klis\altaffilmark{4}, 
31: M. M\'endez\altaffilmark{4,5},
32: N. D. Kylafis\altaffilmark{1,2},
33: E. C. Ford\altaffilmark{4}
34: }
35:  
36: \altaffiltext{1}{Foundation for Research and Technology-Hellas, 711 10  
37: Heraklion, Crete, Greece} 
38:  
39: \altaffiltext{2}{Physics Department, University of Crete, 710 03 Heraklion, 
40: Crete, Greece} 
41: 
42: \altaffiltext{3}{Osservatorio Astronomico di Brera, Via E. Bianchi 46, I-23807 
43: Merate (LC), Italy}
44:  
45: \altaffiltext{4}{Astronomical Institute ``Anton Pannekoek'', 
46:        University of Amsterdam and Center for High-Energy Astrophysics, 
47:        Kruislaan 403, NL-1098 SJ Amsterdam, the Netherlands} 
48:        
49: \altaffiltext{5}{Facultad de Ciencias Astron\'omicas y Geof\'{\i}sicas, 
50: Universidad Nacional de La Plata, paseo del Bosque S/N, 1900 La Plata, 
51: Argentina}
52:  
53: \begin{abstract} 
54:  
55: We have used {\em Rossi  X-ray  Timing  Explorer}  data to measure the
56: lags between  soft (2--5  keV) and hard  (5--13  keV)  photons  and to
57: study the aperiodic variability of the superluminal black hole candidate
58: GRS 1915+105 during low-flux states.  The power density spectra exhibit 
59: quasi-periodic oscillations (QPO) whose frequency increases with
60: increasing count rate and varies in the  frequency  range 0.6--8 Hz.  A 
61: correlation  between the QPO frequency  and the phase lag spectra is
62: reported  for the first time.  This correlation is found for both the
63: phase lag continuum and the phase lag at the QPO  frequency.  We find that
64: as the QPO  frequency  moves to higher  values the phase  lags  reverse 
65: sign  from  positive  to  negative.  The absolute value of the lag always
66: increases with photon energy.  The negative  (soft) lags are  associated 
67: with a softer energy spectrum,  whereas the positive (hard) lags are seen
68: when the source is harder.  We describe a possible scenario that may
69: account for the change in the sign of the lags.
70: 
71: \end{abstract} 
72:  
73: \keywords{accretion --- accretion disks --- stars:  neutron --- stars: 
74: individual (GRS 1915 +105) --- X-rays:  stars} 
75:  
76: \section{Introduction} 
77:  
78: The  X-ray  source  GRS~1915+105  was  discovered  in 1992  with the 
79: WATCH instrument   onboard  GRANAT   (Castro-Tirado  et  al.  1992). 
80: Since  {\em RXTE} established  continuous  coverage in 1996 February, 
81: GRS~1915+105 has never switched  off  completely.  It  was  the  first 
82: galactic  object  to  show superluminal  radio  expansion  (Mirabel  \& 
83: Rodr\'{\i}guez  1994).  It is located at a distance between 6 to 12.5 kpc
84: (Mirabel \& Rodr\'{\i}guez  1994; Fender  et al.  1999).  Due to high 
85: extinction  in the  line of  sight  no optical  study  of GRS  1915+105 
86: has been  possible  and a near infrared counterpart  has been  suggested 
87: (Mirabel  et al.  1994; Bo\"er et al. 1996).  No binary mass function or
88: orbital period are known.
89: 
90: The  source  is  believed  to  host  a  black  hole  on  the  basis  of its
91: similarities  to the  dynamically  proven  black hole  system GRO  1655--40
92: (Orosz  \&  Bailyn  1997)  and its  X-ray  luminosity,  which is above  the
93: Eddington limit for a neutron star.  The X-ray spectrum of  GRS~1915+105 is
94: also  typical  of black  hole  candidates,  requiring  two main  components
95: (Belloni et al.  1997a):  a power-law  component which accounts for the high
96: energy part of the  spectrum  and a thermal  component  (multi-color  black
97: body) which  dominates at energies below 10 keV (Mitsuda et al.  1984).  At
98: times, only the power law  component is observed  (Trudolyubov,  Churazov, \&
99: Gilfanov 1999; Belloni et al.  2000).
100: 
101: GRS~1915+105  displays  a  remarkable  richness  in  variability  (see
102: e.g. Greiner, Morgan \& Remillard 1997).  Belloni et al.  (1997a,b) 
103: modeled the main type of  variations  observed  in this source as due to
104: the onset of a thermal-viscous  instability.  In this framework, the
105: intervals of low flux correspond  to the  non-observability  of  the 
106: innermost  portion  of  the accretion  disk.   During three long (30-100 
107: days) periods in 1996 and 1997, no strong flux  variability  was observed
108: from  GRS~1915+105.  On the basis of an X-ray color  analysis of the {\em
109: RXTE}/PCA  data, Belloni et al. (2000)  interpreted  these quiet periods
110: as the result of the recovery from an  instability  that  affected  a
111: larger  portion of the  accretion  disk, combined  with a relatively 
112: lower value of the  accretion  rate.  Detailed timing  analysis  of part
113: of  these  observations  has  been  presented  by Trudolyubov et al.
114: (1999).
115: 
116: 
117: Three different types of QPOs have been identified in GRS~1915+105  (Morgan,
118: Remillard \& Greiner 1997).  At higher frequencies a sharp ($Q \approx$ 20)
119: QPO with  constant  centroid  frequency  at 67 Hz and whose  rms  amplitude
120: increases with photon energy has been detected.  This QPO has been proposed
121: to arise in the  inner  accretion  disk and its  frequency  is  thought  to
122: reflect  the  properties  of the  black  hole.  At much  lower  frequencies
123: (0.001--0.1  Hz),  the  source  occasionally  shows  high-amplitude  QPOs or
124: brightness  "sputters".  These  variations  probably  correspond  to a disk
125: instability.  The third class of QPOs  observed  in  GRS~1915+105  comprises
126: those  detected in the frequency  range 0.5--10 Hz.  They are linked to the
127: properties  of the  accretion  disk  since  their  centroid  frequency  and
128: fractional  rms have been  reported  to  correlate  with the  thermal  flux
129: (Markwardt,  Swank, \& Taam 1999;  Trudolyubov  et al.  1999) and  apparent
130: temperature  (Muno, Morgan, \& Remillard  1999) of the disk.  These QPOs are
131: observed only during the low-flux intervals when, according to the model of
132: the  thermal-viscous  instability, a portion of the inner accretion disk is
133: not observable.
134: 
135: A phase-lag  analysis of the observation  which showed the strongest signal
136: at 67 Hz was  presented  by Cui (1999):  the 67 Hz QPO shows a marked  hard
137: lag,  while  the low  frequency  (well  below  1 Hz) QPO  shows  a  complex
138: hard/soft lag structure.
139: 
140: In this  paper,  we  concentrate  on the  phase  lags  associated  with  the
141: properties  of the  0.5--10  Hz QPO during two of the long quiet  intervals
142: mentioned above.
143: 
144: 
145: \section{Observations} 
146:  
147: The data used in this  work  were  retrieved  from the  public  {\em 
148: RXTE} archive and  correspond to two different  sets of  observations. 
149: The first set took place  between  1996 July 11 and 1996 August 18 and
150: consists of 27 observation intervals (OBI).  The second set  comprises 16 
151: observation intervals covering the period between  1997   September  29 
152: and  1997  October  24.  These  two  sets  of observations  correspond to
153: low-flux  intervals in which the source did not display  structured 
154: variability  either  in the  PCA  or in the  ASM  data (Belloni et al. 
155: 2000).  The journal of the {\em RXTE}  observations  is given in
156: Table~\ref{observ}.  Due to its low-Earth orbit, {\em RXTE} data consist
157: of a  number  of  contiguous  data  intervals   (typically  1  hour 
158: long) interspersed with  observational gaps produced by Earth occultations
159: of the source and passages of the satellite  through the South  Atlantic 
160: Anomaly. Each observation interval   was  analysed   separately  and 
161: appear  as  different entries  in  Table~\ref{observ}.  A suffix 
162: $a, b, c ,...$  has been added to the data set identifiers accordingly.
163: 
164: \section{Analysis and Results} 
165: 
166: \subsection{Power density spectrum}
167:  
168: In order to study the  source  variability,  we
169: divided the 2--60 keV PCA light  curves (no energy  selection  was made) of
170: each  observation  into 16 s segments  and  calculated  the  Fourier  power
171: spectrum of each  segment.  The Nyquist  frequency was 256 Hz for the first
172: set of  observations  and 64 Hz for the  second.  The  contribution  by the
173: photon counting noise was computed and subtracted from each power spectrum.
174: The power  spectra  were  normalized  such that  their  integral  gives the
175: squared  fractional rms variability  (Belloni \& Hasinger 1990).  The power
176: spectra were then averaged  together to produce one power spectrum for each
177: observation interval.  
178: 
179: In all power spectra a strong  band-limited  noise  component  was present,
180: together with one or more QPO peaks.  We fitted the power  spectra  using a
181: model  that  consisted  of a broken  power law  and two  Lorentzian  peaks,  
182: representing  the band-limited noise
183: continuum  and the QPO plus its second  harmonic,  respectively.  While all
184: power  spectra  clearly  showed  one QPO peak, in some  cases the  second
185: harmonic was not required.  Generally, the power spectra are  characterized
186: by a flat power-law below the QPO frequency ($\alpha$=0.1--0.3) and a steep
187: power-law   above  the  QPO   ($\alpha$=   2.7--3.0).  The   slope  of  the
188: low-frequency  power-law  tends to increase with  increasing QPO frequency.
189: While the QPO  frequency  increases  with count rate, its  amplitude  (rms)
190: decreases  (Fig~\ref{cs_qpo_rms}).  Trudolyubov  et al.  (1999)  found  the
191: same trend  during  the 1996  November--1997  April  low-flux  period.  The
192: results of the power spectral fitting are given in Table~\ref{observ}.
193: 
194: \subsection{Phase lags}
195: 
196: For each  observation,  we produced a phase-lag  spectrum.  We divided 
197: the data into two energy bands (soft:  2--5 keV; hard:  5--13 keV) and
198: produced a cross  spectrum  for  every  16 s.  The  cross  spectrum  is 
199: defined  as $C(j)=X_1^*(j) X_2(j)$,  where  $X_1$  and  $X_2$  are the 
200: complex  Fourier coefficients  for  the  two  energy  bands  at  a 
201: frequency  $\nu_j$,  and $X_1^*(j)$  is the  complex  conjugate  of 
202: $X_1(j)$  (van  der Klis et al. 1987).  The phase lag  between  the 
203: signals  in the two  bands at  Fourier frequency  $\nu_j$ is 
204: $\phi_j=arg[C(j)]$  (the position angle of $C(j)$ in the complex plane)
205: and the corresponding time lag $\phi_j/2\pi  \nu_j$.  We calculated an
206: average cross vector $C$ by averaging the complex values over multiple
207: adjacent 16 s spectra,  and then finding the final value of $\phi$ vs. 
208: frequency.  The error in $\phi$ is  computed  from the observed variance
209: of $C$ in the real and imaginary  directions.  We limited our  phase lag
210: analysis  to  frequencies  below 10 Hz, well  below the  region  where
211: binning  effects  are  important  (Crary  et al.  1998).  In all  phase
212: lag spectra,  positive  lag values  mean that hard  photons  are  lagging 
213: soft photons.  No  correction  for  dead-time  effects in the phase lag 
214: spectra (cross-talk  between energy channels) was done since this effect
215: was found to be negligible.
216: 
217: In order to study the  variability  of the phase lags and its 
218: relationship with the  properties  of the QPO we obtained the mean phase
219: lag  spectrum  by averaging the cross-correlation vectors in four
220: different QPO frequency ranges, $\nu_{QPO} < 1$  Hz,  $1  \leq  \nu_{QPO} 
221: < 2$  Hz, $2  \leq  \nu_{QPO}  < 3$ Hz  and $\nu_{QPO}  \geq 3$ Hz. 
222: Figure~\ref{pds_lag}  shows  these  four phase lag spectra  together with
223: a  representative  power spectrum of the group.  One can see  that  both
224: the lag at the QPO  frequency,  and the  characteristic shape of the lag
225: continuum vary  drastically.  A common  feature of the lag spectra is the
226: presence of a broad dip at around the QPO frequency.  As the QPO
227: frequency  increases,  its position relative to the dip changes.  The
228: frequency of the dip also varies although not as fast: at low QPO
229: frequencies (below 2 Hz) the QPO occurs slightly before the dip minimum,
230: at QPO frequencies above 2 Hz it occurs at around the dip minimum. At 
231: QPO frequencies $\simmore 3.5$ Hz a second dip at lower Fourier
232: frequency develops in the phase lag spectra.
233: 
234: At low values of the QPO  frequency, the continuum of the phase lag
235: spectrum is curved and positive  whereas at high and intermediate QPO
236: frequencies the continuum is flatter and close to zero lag.  Averaging
237: over the 0.06--0.6 Hz range, that is, below any of the QPOs,  for  the 
238: four  cases  shown  in  Figure~\ref{pds_lag}  we find  the following 
239: values of the phase lags:  0.215$\pm$0.004  rad for  $\nu_{QPO}$ $<$  1 
240: Hz,   0.159$\pm$0.004   rad  for  1  $\leq   \nu_{QPO}   <$  2  Hz,
241: 0.071$\pm$0.004 rad for 2 $\leq \nu_{QPO} <$ 3 Hz and --0.024$\pm$0.003
242: rad for $\nu_{QPO} \geq$ 3 Hz.
243: 
244: We also derived phase lags of the  continuum and at the QPO  frequency 
245: for each  observation,  by measuring the average vector in the frequency 
246: range 0.1--5 Hz and $\nu_{QPO} \pm  \frac{1}{2} \Delta \nu$,
247: respectively.  $\Delta \nu$ is the FWHM of the QPO peak.  The results 
248: were  plotted as a function of  QPO  frequency   and  are  shown  in  
249: Figure~\ref{freq_lag}.  A  tight correlation  between  these  quantities 
250: is  found.  As the  QPO  frequency (or count rate) increases the lag
251: decreases, changing sign at around 2 Hz. The opposite behavior was
252: reported for GX 399--4 by Nowak et al. (1999). These authors found the
253: shortest time delays at the lowest flux observation.  The increase above 5
254: Hz observed in Figure~\ref{freq_lag} (bottom) is due  to the dip moving
255: toward higher frequencies.  This  change  in the sign of the lag means 
256: that at a certain point soft photons lag hard photons.  Similar results
257: were obtained for the lags measured at the theoretical harmonics of the
258: QPO frequency.  The frequency at which the lags change from positive to
259: negative are consistent with 4, 6 and 8 Hz, that is, 2, 3 and 4 times that
260: of the QPO (2 Hz).
261: 
262: In  Figure~\ref{lag_ener},  the  dependence  of the phase  lags (at the QPO
263: frequency)  on energy is plotted.  Different  symbols  correspond  to three
264: different values of the QPO frequency:  squares for  $\nu_{QPO}$=0.999  Hz,
265: dots for  $\nu_{QPO}$=2.316  Hz and  stars  for  $\nu_{QPO}$=4.68  Hz.  The
266: energy bands shown are 2.0--5.0, 5.4--6.9, 7.2--9.4 and 9.8--13.1 keV.  The
267: phase lag was obtained by taking the softest band as the reference band and
268: by computing a  cross-power  spectrum  between each of the energy bands and
269: the reference  band.  At low  frequencies,  the phase lags are positive and
270: increase with energy; at high  frequencies  the phase lags are negative and
271: decrease  with energy; at  intermediate  frequencies  the lags are zero and
272: roughly constant. 
273: 
274: 
275: We also calculated  background subtracted light curves corresponding to
276: the following  bands in  pulse-height  channels:  0--13  (2.0-5.0 keV,
277: band A), 14--35  (5.0-13.0  keV,  band B) and  36-100  (13.0-60.0  keV, 
278: band C) and produced hard and soft colors as HR1 = B/A and HR2 = C/A,
279: respectively.  In Figure~\ref{hr1_lag}  we have plotted the  phase lags of
280: the continuum (0.1--5 Hz) and the QPO frequency as a function of HR1.  The
281: negative lags occur when the source is softer (HR1  $\simless$ 1.2), 
282: whereas  the hard lags are seen when the source  spectrum  is harder (HR1
283: $\simmore$ 1.3).  A similar trend is obtained if the hard color HR2 is
284: used.  The anti-correlation between the QPO frequency and HR1 is a
285: consequence of the anti-correlation of the colors with count rate. 
286: 
287: It is worth noting that no soft component is required to fit the energy
288: spectrum in the observations presented here with the possible exception of
289: the ones with the highest QPO frequency. During low-flux states in GRS
290: 1915+105 the X-ray spectra are dominated by the power-law component.
291: Therefore, variations in HR1 correspond to variation in the power-law
292: index. For a more detailed   interpretation  of X-ray colors in terms of
293: spectral models, see Belloni et al.  (2000).
294: 
295: \section{Discussion} 
296:  
297: We have analysed {\em RXTE} data of the superluminal source GRS~1915+105 during
298: two quiescent periods and measured lags of the 5--13 keV photons  relative
299: to the 2--5 keV photons. We have found that both positive and negative
300: lags are present. Belloni et al.  (1997a,b) have shown that the complex
301: spectral and temporal variability  of GRS~1915+105 may be  explained  by
302: the rapid  removal  and subsequent  refilling  of the  inner  region  of
303: the  accretion  disk.  The quiescent  state  corresponds to the absence of
304: this inner region and it is characterized  by a relatively low count rate
305: ($\simless$ 20000 {\em RXTE} PCA c/s), power-law dominated X-ray spectrum and
306: the presence of 0.5--10 Hz QPO.   
307: 
308: In this state both the timing (QPO frequency, lags) and spectral
309: parameters (hardness ratios, temperature of disk) correlate (or
310: anticorrelate) with count rate. Thus a selection according to QPO
311: frequency implicitly implies a selection on spectral hardness or count
312: rate. Of special interest here are the properties of the 0.5--10 Hz QPO
313: (centroid, amplitude), which are strongly  correlated  with the changes of
314: spectral and timing  parameters and can be summarized as follows:  as the
315: frequency  of the QPO  increases  {\em i)} the  flux of the soft 
316: component (disk) increases (Markwardt et al.  1999), {\em ii)} the
317: temperature of the accretion disk increases (Muno et al.  1999), {\em
318: iii)} the fractional rms amplitude of the QPO decreases (Trudolyubov et
319: al.  1999).  In this work we find that {\em iv)} as the  frequency of the
320: QPO  increases the lag between hard and soft  photons  decreases, 
321: changing  sign for  $\nu_{QPO} \simmore 2$ Hz (Fig. ~\ref{freq_lag}), {\em
322: v)} negative lags occur when the power-law  spectrum  is soft and 
323: positive  lags  when  it is hard  (Fig. \ref{hr1_lag}),  {\em  vi)} the 
324: 0.5--10  Hz QPO  disappears  when  the PCA intensity is high ($\simmore$
325: 20000 counts per second). 
326: 
327: 
328: The X-ray  spectra of black hole systems  consist of a power-law 
329: component, thought to be caused by Comptonization, plus a soft thermal 
330: component, which is modeled as a multi-temperature blackbody coming from
331: an optically thick accretion disk.  The relative  strength of these
332: components varies with  X-ray  flux in the 1--10 keV band. In the standard
333: Comptonization model only hard, i.e. positive lags, are expected. The
334: initially low-frequency photons scatter off energetic electrons and gain
335: energy. The larger the number of scatterings the higher the energy of the
336: escaping photons  and also the longer the time they spend in the
337: Comptonizing region.
338: 
339: In what follows we discuss  a simple scenario which might, at least {\em
340: qualitatively}, account for the soft, i.e negative lags.  Let us consider
341: a  Comptonizing  region in which the  Comptonization  process becomes 
342: more  efficient  closer to the black  hole. One  such situation  could 
343: be, for example, a disk consisting of cold electrons with a Keplerian 
344: distribution of velocities (bulk Comptonization) or a hot corona with a
345: temperature gradient, the inner parts being hotter than the outer parts
346: (thermal Comptonization). Let us examine the Keplerian case. Near the
347: black hole, the rotational  velocity of the electrons is close to the 
348: speed of light and those  photons  that find themselves in this
349: region get up-scattered due to bulk-motion  Comptonization. Note than in
350: this model the Comptonizing region is the accretion disk itself. The
351: source of soft photons can be inside (the disk itself) or outside the
352: disk. What is needed is that some photons find their way to the
353: high-velocity parts of the disk. Preliminary results of this type of
354: Comptonization have been reported by Kylafis \& Reig (1999).
355: 
356: If the optical depth is small, the majority of the input photons  escape
357: after a few scatterings. In this case the photons that undergo some
358: scatterings (hard photons) lag soft photons.  But if the optical depth is
359: large, the soft input  photons undergo a random-walk  through the medium
360: prior to escape and gain energy. Some of these now hard photons find 
361: themselves  in the outer parts of the Comptonizing  region where the
362: velocity of the electrons is low and they begin to give away energy by
363: means of the direct Compton effect. We expect then that soft photons lag
364: hard photons. In order for such  downward  scattering  to occur the
365: optical depth must be large and the disk must be geometrically thick so
366: that the photons  sample a large fraction of the  Comptonizing  medium.
367: 
368: If we accept that the optical depth gives a measure of the mass accretion
369: rate and that, in turn, larger accretion rates imply higher count rates,
370: then this simple model may also explain the dependence of the lag sign on
371: intensity. Positive lags occur when the count rate is low, whereas
372: negative lags are detected for high count rates (this observational fact
373: can be inferred from Fig~\ref{cs_qpo_rms} and \ref{freq_lag}). Low
374: intensity, that is low mass accretion rate, implies low optical depth, hence
375: up-scattering only or equivalently, positive lags. High intensity implies
376: large accretion rates or large optical depths and therefore diffusion plus
377: down-scattering, opening up the posibility of negative lags.
378: 
379: Given the relatively short time lags (of the order of a few milliseconds)
380: the Comptonizing region does not need to be very large (a few tens of
381: gravitational radii, $r_g=GM/c^2$). Note also, that the negative time lags
382: are typically an order of magnitude shorter than the positive lags.
383: 
384: 
385: \acknowledgements
386: 
387: This work was  supported  by the  Netherlands  Foundation  for  research
388: in astronomy  (ASTRON) under grant  781-76-017,  by the  Netherlands 
389: Research School for Astronomy  (NOVA), and the NWO Spinoza grant 08-0 to
390: E.P.J.  van den Heuvel.  One of us (PR)  acknowledges  support from the
391: European  Union through the Training  and  Mobility  of Researchers 
392: Network  Grant  ERBFMRX/CT98/0195.  This research has made use of data
393: obtained through the High Energy Astrophysics Science   Archive  
394: Research  Center  Online   Service,   provided  by  the NASA/Goddard Space
395: Flight Center.
396: 
397: \clearpage 
398:  
399: \begin{thebibliography}{} 
400: 
401: \bibitem{} Belloni, T. \& Hasinger, G. 1990, A\&A, 230, 103
402: \bibitem{} Belloni, T., M\'endez, M., King, A.R., van der Klis, M. \& van
403: Paradijs,  J. 1997a, \apj, 479, L145 
404: \bibitem{} Belloni, T., M\'endez, M., King, A.R., van der Klis, M. \& van
405: Paradijs,  J. 1997b, \apj, 488, L112 
406: \bibitem{} Belloni, T., Klein-Wolt, M., M\'endez, M., van der Klis, M. 
407:     \& van Paradijs, J. 2000, A\&A, in press, astro-ph/0001103
408: \bibitem{} Bo\"er, M., Greiner, J. \& Motch, C. 1996, A\&A, 305, 835
409: \bibitem{} Castro-Tirado, A.J., Brandt, S. \& Lund, N. 1992, IAU Circ., 5590
410: \bibitem{} Crary, D. J., Finger, M. H., Kouveliotou, C., van der Hooft, F.,
411: van der Klis, M., Lewin, W. H. G. \& van Paradijs, J. 1998, \apj, 493, L71
412: \bibitem{} Cui W. 1999, \apj, 524, L59.
413: \bibitem{} Fender, R. P.; Garrington, S. T., McKay, D. J., Muxlow, T. W. B.,
414: Pooley, G. G., Spencer, R. E., Stirling, A. M. \& Waltman, E. B. 1999, MNRAS, 
415: 304, 865 
416: \bibitem{} Greiner, J., Morgan, E.H. \& Remillard, R.A. 1997, \apj, 473, L107
417: \bibitem{} Kylafis, N.D. \& Reig, P., 1999 in {\em The Neutron star--Black Hole 
418: Connection}, NATO ASI, Elounda, Crete 7-18 June 1999, in press.
419: %\bibitem{} Liang,  E.P., Physics Reports, 302, 67
420: \bibitem{} Morgan, E.H., Remillard, R.A. \& Greiner, J. 1997, \apj, 482, 993
421: \bibitem{} Markwardt, C.B., Swank, J.H. \& Taam, R.E. 1999, \apj, 513, L37  
422: \bibitem{} Mirabel, I. F., Duc, P. A., Rodriguez, P. A., Teyssier, R., Paul, J.,
423:  Claret, A., Auriere, M.,  Golombek, D. \& Marti, J. 1994, A\&A, 282, L17
424: \bibitem{} Mirabel F. \& Rodr\'{\i}guez L. 1994, Nature, 371, 46 
425: %\bibitem{} Mirabel F. \& Rodr\'{\i}guez L., 1998, Nature, 392, 673
426: %\bibitem{} Mirabel F. \& Rodr\'{\i}guez L., 1999, ARA\&A, 37 
427: \bibitem{} Mitsuda, K, Inoue, H., Koyama, K.,et al. 1984, PASJ,  36, 741 
428: \bibitem{} Muno, M. P., Morgan, E. H. \& Remillard, R. A. 1999, \apj, 527, 321 
429: \bibitem{} Nowak, M.A., Wilms, J. \& Dove, J.B. 1999, \apj, 517, 355
430: \bibitem{} Orosz, J.A. \& Bailyn, C.D. 1997, \apj, 477, 876
431: %\bibitem{} Reig, P.,  Kylafis, N. \& Spruit, H.C. 2000, in preparation
432: \bibitem{} Trudolyunov, S., Churazov, E. \& Gilfanov, M. 1999, Astro-ph/9811449 
433: \bibitem{} van der Klis, M., Hasinger, G., Stella, L., Langmeier, A.,
434: van Paradijs, J. \& Lewin, W. H. G. 1987, \apj, 319, L13
435:  
436: \end{thebibliography} 
437:  
438:  
439: \clearpage 
440: %---------------------------------------------------------------------------- 
441: \begin{deluxetable}{lccccccc} 
442: \scriptsize 
443: \tablecolumns{10} 
444: \tablecaption{RXTE observations of GRS 1915+105 \label{observ}} 
445: \tablewidth{0pt} 
446: \tablehead{} 
447: \startdata 
448: OBI	& MJD	&PCA$^a$    &$\nu_{QPO}$  & rms$^b$ (\%)  &rms$^b$ (\%) &lag (rad)  
449: & lag (ms)  \\
450: 	&	&(c/s)	&(Hz) 	      &(0.01-20 Hz)&(QPO) &at $\nu_{QPO}$ 
451: & at $\nu_{QPO}$ \\
452: \hline 
453: &&&&&&&\\
454: \multicolumn{8}{c}{1996 July 11 -- August 18} \\ 
455: &&&&&&&\\
456: \hline 
457:  10408-01-22-00a     &50275.0905   &9778.3 &  3.468$\pm$0.011     &9.58$\pm$0.15 
458:    &9.17$\pm$0.25   &	 -0.082$\pm$0.015  & -3.8$\pm$0.7  \\
459:  10408-01-22-00b     &50275.1072   &9789.0 &  3.475$\pm$0.005     
460: &11.09$\pm$0.07   &9.31 $\pm$0.15  &	 -0.096$\pm$0.009  & -4.4$\pm$0.4\\
461:  10408-01-22-01      &50275.2238   &9282.4 &  2.768$\pm$0.006     
462: &10.45$\pm$0.10   &10.68 $\pm$0.13 &	 -0.045$\pm$0.006  & -2.6$\pm$0.3\\
463:  10408-01-22-02a     &50275.3572   &9008.1 &  2.561$\pm$0.005     
464: &10.45$\pm$0.09   &11.39$\pm$0.12  &	 -0.027$\pm$0.006  & -1.7$\pm$0.4\\
465:  10408-01-22-02b     &50275.4238   &9175.3 &  2.840$\pm$0.023     &9.31$\pm$0.10 
466:    &9.96$\pm$0.87   &	 -0.032$\pm$0.034  & -1.8$\pm$1.9\\
467:  10408-01-23-00a     &50278.4921   &9735.2 &  3.480$\pm$0.004     &9.87$\pm$0.08 
468:    &10.27$\pm$0.12  &	 -0.098$\pm$0.007  & -4.5$\pm$0.3\\
469:  10408-01-23-00b     &50278.5585   &9709.6 &  3.614$\pm$0.005     
470: &10.20$\pm$0.07   &9.90$\pm$0.11   &	 -0.103$\pm$0.006  & -4.6$\pm$0.3\\
471:  10408-01-23-00c     &50278.6266   &10457.7 &  4.209$\pm$0.009    &9.65$\pm$0.08 
472:    &9.12$\pm$0.11   &	 -0.122$\pm$0.006  & -4.6$\pm$0.2\\
473:  10408-01-24-00a     &50280.1702   & 8923.9 &  2.238$\pm$0.005    
474: &11.45$\pm$0.10   &12.13$\pm$0.11  &	 -0.001$\pm$0.006  &  -0.1$\pm$0.4\\
475:  10408-01-24-00b     &50280.2266   &8900.8 &  2.316$\pm$0.005     
476: &10.95$\pm$0.12   &12.00$\pm$0.16  &	 0.001$\pm$0.005  &  0.1$\pm$0.4\\
477:  10408-01-24-00c     &50280.2933   &8955.6 &  2.537$\pm$0.004     
478: &11.62$\pm$0.15   &11.34$\pm$0.15  &	 -0.036$\pm$0.006  & -2.3$\pm$0.4\\
479:  10408-01-24-00d     &50280.3599   &9011.1 &  2.598$\pm$0.008     
480: &10.68$\pm$0.15   &10.95$\pm$0.28  &	 -0.055$\pm$0.011  & -3.3$\pm$0.7\\
481:  10408-01-25-00a     &50283.4947   &8263.6 &  1.110$\pm$0.003     
482: &12.92$\pm$0.11   &11.67$\pm$0.20  &	 0.172$\pm$0.007  & 24.7$\pm$1.0\\
483:  10408-01-25-00b     &50283.5620   &8154.0 &  1.077$\pm$0.003     
484: &13.06$\pm$0.10   &11.24$\pm$0.21  &	 0.185$\pm$0.007  & 27.3$\pm$1.0\\
485:  10408-01-25-00c     &50283.6286   &8076.9 &  1.184$\pm$0.002     
486: &12.28$\pm$0.14   &12.06$\pm$0.16  &	 0.154$\pm$0.006  & 20.6$\pm$0.8\\
487:  10408-01-27-00a     &50290.5774   &7887.2 &  0.644$\pm$0.002     
488: &13.73$\pm$0.11   &9.74$\pm$0.28   &	 0.247$\pm$0.014  & 72.7$\pm$4.1\\
489:  10408-01-27-00b     &50290.6322   &7925.8 &  0.621$\pm$0.003     
490: &13.12$\pm$0.12   &9.66$\pm$0.22   &	 0.285$\pm$0.010  & 73.0$\pm$2.7\\
491:  10408-01-27-00c     &50290.6988   &7826.5 &  0.629$\pm$0.003     
492: &13.04$\pm$0.20   &10.28$\pm$0.19  &	 0.272$\pm$0.008  & 68.9$\pm$2.1\\
493:  10408-01-28-00a     &50298.5696   &7751.1 &  0.999$\pm$0.003     
494: &12.33$\pm$0.11   &11.36$\pm$0.18  &	 0.189$\pm$0.006  & 30.0$\pm$1.0\\
495:  10408-01-28-00b     &50298.6362   &7755.6 &  0.962$\pm$0.003     
496: &13.49$\pm$0.12   &11.21$\pm$0.21  &	 0.201$\pm$0.007  & 33.3$\pm$1.2\\
497:  10408-01-28-00c     &50298.7029   &7533.5 &  0.928$\pm$0.003     
498: &13.16$\pm$0.13   &11.55$\pm$0.19  &	 0.221$\pm$0.007  & 38.0$\pm$1.2\\
499:  10408-01-29-00a     &50305.3736   &7951.5 &  1.673$\pm$0.004     
500: &12.51$\pm$0.13   &11.98$\pm$0.17  &	 0.085$\pm$0.007  &  8.1$\pm$0.6\\
501:  10408-01-29-00b     &50305.4398   &8070.8 &  1.858$\pm$0.004     
502: &12.15$\pm$0.11   &11.97$\pm$0.16  &	 0.049$\pm$0.005  &  4.2$\pm$0.5\\
503:  10408-01-29-00c     &50305.5064   &8078.7 &  1.962$\pm$0.003     
504: &12.16$\pm$0.11   &11.95$\pm$0.15  &	 0.036$\pm$0.005  &  3.0$\pm$0.4\\
505:  10408-01-30-00a     &50313.3135  &11560.3  &  4.551$\pm$0.009    &8.75$\pm$0.10 
506:    &8.95$\pm$0.08   &	 -0.140$\pm$0.006  & -4.9$\pm$0.2\\
507:  10408-01-30-00b     &50313.3801  &12979.6  &  5.030$\pm$0.008    
508: &10.85$\pm$0.08   &7.26$\pm$0.22   &	 -0.171$\pm$0.006  & -5.4$\pm$0.2\\
509:  10408-01-30-00c     &50313.4468  &14701.7  &  5.551$\pm$0.015    
510: &13.21$\pm$0.13   &4.72$\pm$0.18   &	 -0.188$\pm$0.008  & -5.4$\pm$0.2\\
511: \hline
512: &&&&&&&\\
513: \multicolumn{8}{c}{1997 September 29 -- October 24} \\
514: &&&&&&&\\ 	
515: \hline  	
516:  20402-01-48-00a     &50720.5910  &21914.4   &  7.56 $\pm$0.03   &18.09$\pm$0.15 
517:    &2.50 $\pm$0.06  &	 -0.260$\pm$0.011  & -5.5$\pm$0.2\\
518:  20402-01-48-00b     &50720.6572  &12618.6   &  4.68 $\pm$0.01   &11.83$\pm$0.06 
519:    &5.62 $\pm$0.09  &	 -0.147$\pm$0.008  & -5.0$\pm$0.3\\
520:  20402-01-48-00c     &50720.6783  &11805.7   & 4.54$\pm$0.04     &11.45$\pm$0.20 
521:    &7.14$\pm$0.29   &	 -0.179$\pm$0.022  & -6.3$\pm$0.8\\
522:  20402-01-49-00a     &50729.3275  & 8714.6   &  2.911$\pm$0.005  &10.18$\pm$0.07 
523:    &10.89$\pm$0.11  &	 -0.033$\pm$0.006  & -1.8$\pm$0.3\\
524:  20402-01-49-00b     &50729.3940  & 8369.3   &  2.621$\pm$0.006  &10.60$\pm$0.10 
525:    &11.05$\pm$0.16  &	 -0.015$\pm$0.007  &  -0.9$\pm$0.4\\
526:  20402-01-49-01      &50730.3949  & 8197.8   &  2.721$\pm$0.005  &10.76$\pm$0.07 
527:    &10.96$\pm$0.12  &	 -0.041$\pm$0.006  & -2.4$\pm$0.3\\
528:  20402-01-50-00      &50735.5474  & 6579.2   &  0.844$\pm$0.003  &13.51$\pm$0.18 
529:    &10.52$\pm$0.29  &	 0.239$\pm$0.010  & 45.0$\pm$2.0\\
530:  20402-01-50-01a     &50737.4047  & 6607.5   &  1.016$\pm$0.003  &13.76$\pm$0.11 
531:    &10.82$\pm$0.21  &	 0.201$\pm$0.009  & 31.5$\pm$1.3\\
532:  20402-01-50-01b     &50737.4766  & 6373.4   &  1.079$\pm$0.003  &13.60$\pm$0.14 
533:    &11.09$\pm$0.22  &	 0.189$\pm$0.010  & 27.8$\pm$1.4\\
534:  20402-01-51-00a     &50743.2947  & 6626.3   &  1.399$\pm$0.006  &13.92$\pm$0.17 
535:    &11.71$\pm$0.31  &	 0.130$\pm$0.012  & 14.7$\pm$1.3\\
536:  20402-01-51-00b     &50743.3377  & 6602.3   &  1.355$\pm$0.004  &13.73$\pm$0.12 
537:    &11.60$\pm$0.20  &	 0.135$\pm$0.007  & 15.9$\pm$0.8\\
538:  20402-01-51-00c     &50743.4092  & 6584.0   &  1.399$\pm$0.005  &13.84$\pm$0.13 
539:    &11.38$\pm$0.20  &	 0.133$\pm$0.007  & 15.1$\pm$0.8\\
540:  20402-01-51-00d     &50743.4801  & 6640.4   &  1.442$\pm$0.003  &13.55$\pm$0.14 
541:    &11.97$\pm$0.20  &	 0.116$\pm$0.008  & 12.8$\pm$0.9\\
542:  20402-01-52-00a     &50746.5509  & 6657.3   &  1.413$\pm$0.005  &13.77$\pm$0.12 
543:    &11.27$\pm$0.22  &	 0.134$\pm$0.009  & 15.2$\pm$1.1\\
544:  20402-01-52-00b     &50746.6190  & 6667.5   &  1.482$\pm$0.005  &13.86$\pm$0.14 
545:    &11.89$\pm$0.24  &	 0.149$\pm$0.010  & 16.1$\pm$1.1\\
546:  20402-01-52-00c     &50746.6898  & 6767.9   &  1.602$\pm$0.004  &12.90$\pm$0.18 
547:    &11.86$\pm$0.27  &	 0.083$\pm$0.010  &  8.3$\pm$1.0\\
548: \tableline 
549: \enddata
550: \tablenotetext{a}{In the range 2-60 keV and 5 PCU.}
551: \tablenotetext{b}{Normalization according to Belloni \& Hasinger (1990).}
552: \end{deluxetable} 
553: %---------------------------------------------------------------------------- 
554:  
555: \clearpage 
556: 
557: \figcaption[fig1.ps]{
558: Variation of the $rms$ amplitude and frequency of the QPO with count rate.
559: \label{cs_qpo_rms}
560: }
561: 
562:  
563: \figcaption[fig2.eps]{
564: Power and phase lag spectra versus Fourier frequency for four QPO frequency
565: ranges.  The phase lag  spectra  are the average of all those  included  in the
566: frequency  range  shown.  The power  spectra  are  individual  examples of a
567: representative member of the group. Positive lags mean that hard photons lag
568: soft photons.
569: \label{pds_lag} 
570: } 
571:  
572: \figcaption[fig3.ps]{ 
573: Phase lag at the QPO frequency (top panel) and phase lag of the continuum
574: in the frequency range 0.1--5  Hz  (bottom panel) versus QPO  frequency. 
575: Different  symbols represent  different  observing runs as
576: follows:  1996 July 11  -- August 18 (circles), 1997 September 29
577: --October 24 (stars) 
578: \label{freq_lag} 
579: } 
580: 
581: \figcaption[fig4.ps]{
582: Dependence of the lag (at the QPO frequency) on energy.  Different symbols
583: correspond to three different values of the QPO frequency: squares for 
584: $\nu_{QPO}$=0.999 Hz, dots for $\nu_{QPO}$=2.316
585: Hz and stars for $\nu_{QPO}$=4.68 Hz. The lags were calculated for the
586: energy bands 5.4--6.9 keV, 7.2--9.4 kev and  9.8--13.1 keV with respect to
587: the soft band  2.0--5.0 keV.
588: \label{lag_ener}
589: }
590: 
591: \figcaption[fig5.ps]{ 
592: Phase lag and QPO frequency as functions of the soft color HR1=5--13
593: keV/2--5 keV. Negative lags are seen only when the source is soft.
594: \label{hr1_lag}
595: } 
596: 
597:  
598: 
599: \end{document} 
600:  
601: