Документ взят из кэша поисковой машины. Адрес оригинального документа : http://hea.iki.rssi.ru/integral06/papers/F33_Hjalmarsdotter04-03May07_17:14:48_bandw.pdf Дата изменения: Thu May 3 17:14:48 2007 Дата индексирования: Mon Oct 1 23:33:35 2012 Кодировка: Поисковые слова: m 101

IDENTIFYING A NEW CLASS OF VARIABILITY IN GRS 1915+105
D.C. Hannikainen(1), L. Hja lma rsdotter(1), J. Rodriguez(2), O. Vilhu(1), A.A . Zdziarski(3), and T.Belloni(4
(1) )

Observatory, PO Box 14, FI-00014 University of Helsinki, Finland, E-ma il:diana@astro.h elsinki.fi (2) CEA Sacla y, UMR 7158/AIM, DS M/D APNIA/Sap, F91191 Gif sur Yvette, France (3) Nico laus Copern icus Astronomical Center, Bartycka 18, 0 0-716, Warsza wa, Poland (4) INAF ­ Osserva torio Astronomico di Brera, Via E. Bianch i 46, I-23807, Merate, I taly

ABSTRAC T We report on th e analysis of 100 ks INTEGRAL observations of the Galactic black hole bin ary GRS 1915+105. We focus on INTEG RAL Revolution 48 (MJD 52704­52705) when the source was found to exhibit a new type of variab ility. The var iability pattern, which we name , is char acter ized by a pulsing behavior consisting of a main pu lse and a shorter, softer, and smaller amp litude precursor pulse, on a timescale of 5 min in the JEM-X 3­35 keV lightcurve. We present simultaneous RXTE data. W e sep arated the INTEG RAL data into two subsets cover ing th e maxima and the min ima of the pulses and fitted the r esulting two broadband spectr a w ith a hybrid th ermalnonthermal Comp tonization mod el.

2.

OBSERVA TIONS

*

GRS 1915+105 was monitored with INTEG RAL as part of the AO-1 open time program. Th is monitoring continued in AO-2 and AO-3. Here we presen t the results of the very first satellite revo lution (48) of our monitoring program. Th e observation lasted for 100 ksec. W e also had a simultaneous campaign with RXTE to complement our INTEGRAL observations. RXTE observed GRS 1915+105 for about one third of the INTEG RAL time. (See [7] for details of the observations and data reduction.) Th e JEM-X lightcurve from th e whole observation is shown in Fig. 1 and superposed is th e duration of the RXTE observation.

1.

GRS 1915+105

GRS 1915+105 has been ex tensively observed at all wavelengths sin ce its d iscov ery. It was orig inally detected as a hard X-ray source with the WA TCH allsky monitor on the G RANAT satellite [1]. Appar ent superluminal ejections h ave been observed from G RS 1915+105 several times with the V LA and w ith MERLIN [2, 3]. Both times the tru e ejection velocity was calcu lated to be >0.9 c. The mass-donating star is a K-M g ian t [4], while th e mass of the black hole h as been calcu lated to be 14.0±4.0 M [5]. These ar e in a binary orbit of 33.5 days. The Rossi X-Ray Timing Explorer (RXTE) has observed GRS 1915+105 since its launch and has shown the source to b e high ly var iable on all timescales from milliseconds to months ( see, e.g., [6]). Refer ence [6] categorized th e variability into twelve distinct classes wh ich th ey labelled w ith Gr eek letters and they iden tif ied three distin ct X-ray states: two softer states (A and B) and a harder state (C) . H ere we will d iscuss simultaneous INTEG RAL and RXTE observations of G RS 1915+105 and we will show how we def ined a n ew class of variability. * The observations presen ted here are on behalf of a much larger collabor ation, wh ich the au thors th ank.

Fig. 1 The JEM-X 3­35 keV lightcurve for the whole of Rev. 48. The time bin is 8 s. As can be seen, the source varied between ~0.25­2 Cr ab. Shown on the plot is our simultaneous RXTE cover age. 3. A NEW TY PE OF VA RIA BILITY?

Throughout most of th e JEM-X lightcurve we see a novel class of variability, char acter ized by 5-minute pulses, not observ ed before [7]. The entire JEM-X lightcurve is dominated by these 5-min pulses, as eviden ced by the 3 mHz quasi-periodic oscillation

_____________________________________________________________________ Proc. 6th IN TEGRAL Workshop `The Obscured Universe', Moscow, 3-7 July 2006 (ESA SP-622)


which r esulted from a Fourier tran sform of the whole JEM-X lightcurve [7]. Although this kind of pulsed variab ility resembles the -heartb eat and oscillations [6], the oscillations presented th ere are more uniform and occur on shorter timescales. 3.1 RXTE and "Pulses" During some of the RXTE are clear ly detected also by segmen t in particu lar, w e s represen tativ e of the var iab of the JEM-X lightcurve. pointings, the 5-min pulses RXTE/PCA (Fig. 2). In one ee nin e consecutive pulses, ility pattern in the major ity Fig. 3 Th e mean of the n ine RXTE/PCA pulses. The lines ar e as follows: top (2­5 keV; middle 5­13 keV; bottom 13­40 keV .

The RXTE/PCA lightcurve w as d ivided into three energy bands (2­5, 5­13, 13­40 keV) and smoothed with a boxcar aver age of 5 time bins. Fig . 3 shows the mean of the n ine RXTE/PCA pu lses. The pulses consist of the main pulse (with the r ising phase being shorter and harder than the declining phase) and a precursor pulse, which is shorter , softer , and of smaller amp litude than the main pulse. W e can see that th e rising phase of the ( main) pulse is h arder than the declining phase.

Fig. 2 The RXTE/PCA 2.5­15 keV lightcurve (time bin 1 s). Each segmen t repr esen ts on e RXTE orbit. To further investigate the ind icated limit cycle, we constructed a color-co lor diagr am from the RXTE/PCA data, shown in Fig. 4. The X-ray colors were def ined as HR1=B/A and H R2=C(A+B), wher e A, B, and C are counts in the 2­5 keV, 5­13 keV, and 13­40 keV, respectively . In Fig. 4, the dots repr esen t the whole RXTE observation (includ ing the non-pulsed part), while the solid line shows the cy cle traced by the pulses. The main aim of the figure is to show the small range of variability of th e 5-min pu lses. The pulse cycle goes clo ckwise as ind icated by the arrows. The maxima of the pulses are at th e upper-right of the pulse cycle curve, while th e minima are at the low er left. Superposed in Fig . 4 are model curves from a study by [8], calcu lated as exp lained below. Although a d irect

Fig. 4 Color-co lor diagram showing the RXTE/PCA from this study. The dots r epresen t the whole observation, while th e solid line superposed is from the pulses. Th e curved dotted lines corr espond to the colors expected from a Comp tonizing spherical corona with a given rad ial optical d epth 50 (i.e. optical d epth for the assumed electron temp eratur e Tin). The solid lines show the colors expected for the mod el for given Tin and varying T50. comparison of the parameters is not possible due to the RXTE/PCA gain evolution through the years, r esulting in a change of th e energy channel correspondence, we simp ly want to illustrate qualitatively the position in the color-co lor diagram of th is pr esen t observ ation. In order to have an idea of what physical parameters the observed colors correspond to, we simu lated the colors predicted by a Comp tonization mod el. We assu med the Compton izing source to be a sphere surrounded by a cooler d isk with the inner rad ius equal to th at of the sphere [9]. Seed photons coming from the d isk are characterized by an inner disk temperature Tin. The electrons in the cloud were assumed to be thermal w ith temp erature kT=50 k eV. The Thomson optical d epth of


the cloud along the r adius is 50, with subscr ipt 50 mean ing th at this corresponds to 50 keV electrons. For a d ifferen t T, similar Compton ization spectr a are produced by a cloud of optical dep th =50в50 keV/kT, because th e slop e of the Compton ization spectrum, , is a function of the Kompaneets par ameter, y=4kT/mec2, and =(9 /4)y-2/9 [10]. For the given Tin and 50, we compute th e Comptonization spectrum using th e method of [11]. Then we ob tain th e colors expected from the model spectrum using the responses of RXTE/PCA and plot th em on the co lor-color diagram. W e see th at the pulses d iscussed in this p aper correspond to optical dep th 0.7 and inn er disk temp erature of Tin~1.3­14 keV. Th is temperature differs slightly from that obtain ed using a blackbody + powerlaw model, because the Co mptonization spectrum is no t a power law clo se to the seed photon energies. 4. INTEGRA L AND THE HY BRID M ODEL

the characteristic size, r, wh ich is usu ally expr essed in dimensionless form as the compactness p arameter l=LT/(rmec3), wh ere T is the Thomson cross section and me is the electron mass. Th e compactn ess corresponding to th e electron acceleration at a powerlaw r ate with an index, inj and to a d irect heating (i.e. in addition to Coulomb energy ex change with non-thermal e± and Comp ton heating) of the thermal e± are deno ted as lnth and lth respectively, and lh=lnth+lth. Follow ing [15], we assume here a constan t ls=100, compatib le with the high lu minosity of GRS 1915+105. For example, for half of the Eddington luminosity , LE, and a spherical geometry, the size of the plasma corresponds to r ~ 100GM/c2. We note, however, that th e dep endence of the f it on ls is rather weak, as this parameter is importan t only for e± pair production and Coulomb scatter ing, w ith the former not constrained by our data and the latter only important at ls1 or so . We include Comp ton ref lection parametrized by an effectiv e so lid angle subtended by th e r eflector as seen from the hot plasma, , and an Fe K fluorescen t line from an accr etion disk assumed to extend down to 6GM/c2, th e radius of the last stab le orb it for par ticles around a Schwarzschild black hole. W e define the ionizing par ameter as ion=4Fion/n, where Fion is the ionizing flux and n the ref lector d ensity. As the data poorly constrain ion, but clear ly require the ref lector to be ionized , we fr eeze it to a value of 100 erg cm s-1, in the middle of the confidence in tervals for both fits. W e further assume the elemen tal abundances of [16], an absorbing colu mn of NH1 .8в1022 cm-2, which is an estimate of the Galactic column d ensity in the dir ection of the source by [17], and an inclin ation of 66° [3]. 4.1 Results We will present the results only briefly here. The two spectra ar e shown in Fig. 5 togeth er with the best-fit models. Fig. 6 shows the spectral components of the fits of th e two sp ectra. Figs. 5 and 6 show that the main differ ence b etw een the pulse maxima and pulse minima is in the observed flux, which is h igher by a factor of about 2 during maxima. Both sp ectra show that GRS 1915+105 during Revo lution 48 was in a soft state (follow ing [6]), with spectra similar to those of State A in [18] and in between th e softest sp ectrum of the var iab ility class C/ and that in th e B/ class, as observ ed by RXTE and CGRO . The fits imp ly that during the pulse maxima as well as minima the spectru m below abou t 6 keV is dominated by unscattered d isk emission, w ith

In this section, we will in troduce the hybrid mod el used to fit the INTEG RA L sp ectra. In order to conduct a detailed analysis of th e sp ectrum, we used co-added spectra, ex tracted from the h igh and low flux p arts in the INTEG RAL lightcurve ­ d ata above 70 counts s-1 (~0.5 Crab) contribu ted to the "high " par t, while data below 70 counts s-1 contribu ted to the "low" par t. Th is provided us with two broadband spectra (JEM-X + ISGRI + SPI) in the energy range 3­200 keV, for the pulse max ima and minima separ ately . Th e two sp ectra were f itted using XSPEC V11.3 [12]. We in terpret the spectra in terms of Comp tonization of soft X-ray seed photons, assu med her e to b e a d isk blackbody with a maximum temp erature, Tin. We use the eqpair Compton ization model by Coppi [13, 14]. In the eqpair mod el, th e electron distr ibution can be set to be purely thermal Maxwellian or a hybrid consisting of a Maxw ellian component and a higher energy component. The introduction of the latter component is to tak e into account the situation wh ere the presence of an acceleration process w ill giv e r ise to a high-energy tail in addition to a normal Maxwellian distr ibution . The electron distr ibution, including the temp erature (Te) is calculated self-consistently from the assumed form of the acceleration (if present) and from the luminosities corresponding to the plasma heating rate, Lh, and to the seed photons irradiating the cloud, Ls. The total p lasma optical dep th, , includes contributions from electrons formed by ion ization of the atoms in th e plasma, i (which is a free parameter) and contribu tion from e± pairs, -i (which is calcu lated self- consisten tly by the model) . The importance of pairs and the relative importan ce of Coulomb scattering d epend on the ratio of th e luminosity , l, to


Fig. 5 Deconvolved spectr a of G RS 1915+105 during Revo lution 48 from the pulse max ima (top) and pulse minima (bo ttom) . Th e cy an, magen ta and blue data points are from JEM-X, ISGRI and SPI r espectively. The dash ed and so lid curves repr esen t th e best-fit models to th e observed and unabsorbed sp ectra respectively . Compton ized emission domin ating only at higher energies, and in cluding a signif icant contr ibution from non-thermal electrons. The f igures show scattering by thermal and nonthermal electrons sep arately. W e see that ther e is a significan t soften ing of the Comptonized part of the spectrum during pulse minima. This is reflected in the ratio, lh/ls, being signif icantly smaller. Th e ratio lh/ls, or equivalen tly, Lh/Ls, is between the power supplied to the electrons in the Co mptonizing plasma and th at in the soft d isk blackbody photons irrad iating the plasma. Our fits do not r equire an additional component in soft photons not irradiating th e hot plasma and thus Ls corresponds to th e lu minosity in th e en tire disk blackbody emission. The value of Ls decreases by a factor of ~2 during minima. G iven the corresponding decrease of Lh/Ls by a factor of ~2, we find that Lh decreased by 4. Conversely there is an increase in the relative power supplied to the electrons in th e plasma during pulse max ima by a factor ~2. The r elative nonthermal power is ~0.26 for the pulse maxima and ~0.21 for the pulse min ima. It should be noted that the spectrum from the pulse maxima contains both th e end of the rising phase and the beg inning of the declining phase of each pu lse, and thus it is an aver age of the harder rising and softer declining ph ases. REFER ENC ES 1. Castro-Tirado, A.J., Brand t, S. & Lund, N., 1992, IAUC, 5590

Fig. 6 Spectr al componen ts of th e f its to th e pulse maxima (upper panel) and pulse minima (low er panel). The dotted and long-dash ed curves show the unscatter ed blackbody, and the Co mpton scattering from thermal and non-thermal electrons, including a component from e± pair ann ihilation, important around 511 keV. The short-dashed curv e shows the co mponent from Comp ton ref lection, including the Fe K lin e. 2. Mir abel, I .F. & RodrМgu ez, L.F., 1998, Na ture, 39 673 3. Fender , R.P. et al., 1999, MNRAS, 304, 865 4. Greiner, J. et al., 2001, A&A, 373, L37 5. Har laf tis, E. & Gr ein er, J., 2004, A&A, 414, L13 6. Belloni, T. et al., 2000, A&A, 355, 271 7. Hannik ain en, D .C. et al., 2005, A&A, 435, 995 8. Vilhu, O . & N evalainen, J., 1998, ApJ, 508, L85 9. Poutanen, J. et al., 1997, MNRAS, 292, L21 10. Beloborodov, A .M., 1999, ASP Conf. S er., 16 295 11. Poutan en, J. & Sv ensson, R., 1996, ApJ, 470, 249 12. Arnaud, K ., 1996, ASP Conf. Ser., 101, 17 13. Coppi, P.S., 1992, MNRAS, 258, 657 14. Coppi, P.S., 1998, ASP Conf. S er., 161, 375 15. Zdziarsk i, A.A ., et al., 2001, ApJ, 554, L45 16. Anders, E. & Eb ihara, M., 1982, Geoch im. Cosmoch im. Acta, 46, 2363 17. Dickey , J.M. & Lockman, F.J., 1990, ARA&A, 2 215 18. Sobolewska, M.A . & Zycki, P.T., 2003, A&A, 40 553 2,

1,

8, 0,