- Email: [email protected]
BeppoSAX study of the spectral and temporal behavior of the black hole candidate XTE J1118+480
Nuclear Physics B (Proc. Suppl.) 132 (2004) 392–395 www.elsevierphysics.com
BeppoSAX study of the spectral and temporal behavior of the black hole candidate XTE J1118+480 L. Amatia and F. Fronteraab a
Istituto di Astrofisica Spaziale e Fisica Cosmica, CNR, Via P. Gobetti 101, I-40126 Bologna, Italy
b
Dipartimento di Fisica, Universit´ a degli studi di Ferrara, Via Paradiso 12, I-44100 Ferrara, Italy
The high galactic latitude black–hole candidate XTE J1118+480 was followed–up by BeppoSAX during its outburst and transition to quiescence with four TOOs from April to December 2000. In this paper, we summarize and briefly discuss the results of the spectral and temporal data analysis of these observations.
1. INTRODUCTION The compact object of the Low Mass X–ray Binary (LMXB) XTE J1118+480, a system discovered with the All-Sky Monitor (ASM) aboard the Rossi X-Ray Time Explorer (RXTE) satellite during a low state mini–outburst in X–rays in early 2000 [1], is a well-established Black Hole Candidate (BHC). The main characteristics of this system are: high galactic latitude (+48o ), mass of the compact object in the range from 6.0 to 7.7 solar masses [2], radio, near-infrared, optical and EUV counterparts, a surprisingly low X–ray–to–optical flux ratio of ∼ 5, against a typical value of ∼ 500 for LMXB/BHCs [3], a binary period P = 0.1703 days, inclination i = 81 ± 2 degrees, distance d = 1.9 ± 0.4 kpc [4– 6]. From multi-wavelength spectral investigations some relevant properties of the emission processes at the source site were derived. In the radio– submillimiter–near-infrared band, the spectrum shows a shape typical of hard/low spectral states of BHCs and believed to be due to synchrotron emission from self-absorbed jets [7]. Another relevant spectral property is the presence of a thermal component consistent with a blackbody at ∼ 40 eV. This component is apparent in the first observation of XTE J1118+480 with BeppoSAX [8] and it is also detected in the multi-wavelength spectrum of the source derived by McClintock et al. (2001) [9]. Markoff et al. (2001) [10] find the multi-wavelength spectrum from radio to X– 0920-5632/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.nuclphysbps.2004.04.069
rays also consistent with non thermal synchrotron emission from a jet-like matter outflow. Timing analysis of the source revealed, as in several other BHCs in their low/hard state, a low frequency QPO feature. The centroid frequency was observed to continuously drift from ∼ 0.07 Hz to ∼ 0.17 Hz from 2000 April 10 to June 11 [11]. With the BeppoSAX Narrow Field Instruments (NFIs) [12] we observed the source 4 times from April to December 2000. We present a summary and brief discussion of the results of the spectral and timing analysis for all the four BeppoSAX observations. 2. OBSERVATIONS AND DATA ANALYSIS The first three observations were performed in April, May and June 2000, at time intervals of about one month. During the second observation, the LECS was switched off. From 2000 June 30 to October 10, the source was outside the visibility window for BeppoSAX, thus the last BeppoSAX observation was performed in December 2000, long after the ASM outburst was over. Standard data reduction techniques were adopted for all the instruments. The LECS and MECS source spectra were extracted from a region with a radius of 8 around the centroid of the source image. As background spectra we used standard files obtained from the observation of blank fields. We used background spectra accumulated from
L. Amati, F. Frontera / Nuclear Physics B (Proc. Suppl.) 132 (2004) 392–395
393
Table 1 Most relevant best fit parameters of the bb+compps (slab, Z/Z free) and flux variation in the first three observations. Parameter TOO1 TOO2 TOO3 +0.18 20 −2 NH [10 cm ] 1.50−0.17 [1.28] 0.74+0.25 −0.18 +3 kTbb [eV] 35−4 [35] 52+7 −6 Lbb [1034 erg/s]a 5.8+2.2 [5.8] 1.3+2.0 −1.3 −1.2 kTe [keV] 63+20 96+37 90+12 −11 −29 −13 τ 0.64+0.20 0.75+0.29 0.80+0.11 −0.11 −0.24 −0.12 +0.08 +0.08 Ω/2π 0.16+0.11 0.22 0.24 −0.06 −0.12 −0.06 +0.11 +0.14 Z/Z 0.10−0.06 0.17−0.09 0.15+0.09 −0.07 F0.1−2 keV b 5.73 – 5.77 F2−10 keV b 6.95 6.62 5.89 19.4 18.2 15.7 F15−200 keV b 2 χ /dof 194/173 95/105 161/165 a Scaled to d = 1.5 kpc. b In units of 10−10 erg cm−2 s−1 . dark Earth data for the HPGSPC (note that the instrument’s collimator was kept on the source for the entire observation). The energy bands used for spectral fitting were limited to those where the response functions were best known, i.e, 0.12–4 keV, 1.7–10 keV, 7–29 keV, and 15– 200 keV, for the LECS, MECS, HPGSPC and PDS, respectively. A systematic error of 1% was added in quadrature to the statistical uncertainties of the spectral data, on the basis of the calibration results obtained with the Crab Nebula, that was observed on 10 April 2000, 4 days before the first observation of XTE J1118+480. We allowed for a free normalization of the instruments in multi-instrument fits with respect to MECS. The quoted errors for the spectral parameters correspond to 90% confidence for one parameter (∆χ2 = 2.71). We assumed a source distance of 1.5 kpc and a disk inclination of i = 70. The parameter values within square parentheses in the Tables are kept fixed in the fits. 3. RESULTS The source was detected in the first three observations. A large number of spectral models were tested: power–law (pl), cutoff power–law (cut-
offpl), which is equivalent to the pexriv model [13] in xspec with no Compton reflection, Comptonization model (comptt) of Titarchuk [14], thermal-Compton model ( compps) of Poutanen & Svensson [15], all either singly taken or with the addition of a blackbody (bb), all photoelectrically absorbed (wabs model in xspec). Either spherical (sp) or slab (sl) geometries of the Comptonizing electron cloud as well as solar or free metal abundances Z of the cold medium were tested. In Table 1 we report the results obtained with a bb plus compps model assuming a slab geometry of the Comptonizing electron cloud and the metallicity of the reflecting medium free to vary. These best fitting models, superimposed to the F(E) spectra and together with the residuals, are shown in Figure 1. The column density measured in the first BeppoSAX observation is consistent with the Galactic NHG obtained from radio measurements in the direction to the source (NH (1.28–1.44) × 1020 cm−2 ). However the NH derived from the June observation (TOO3) is lower by about a factor 2. Possibly a local absorption contributes to the absorption measured during the April observation (TOO1), if the radio measurements overestimate the Galactic NH . The source spectra show a soft excess which is
394
L. Amati, F. Frontera / Nuclear Physics B (Proc. Suppl.) 132 (2004) 392–395
Table 2 Main timing analysis results of the 0.01–50 Hz power spectral density TOO Energy band r.m.s. fQP O (keV) % (rms) (×10−2 Hz) 1 0.1–1.5 62±3 7.81+0.29 −0.25 1.6–10 42±2 8.11+0.16 −0.17 15-200 18±1 7.82+0.39 −0.29 2 1.6–10 38±4 12.08+0.12 −0.11 15–200 16±1 12.20+0.35 −0.41 3 0.1–1.5 52±4 7.29+0.41 −0.33 1.6–10 32±2 [7.29] 15–200 14±1 [7.29]
consistent with a bb emission with a temperature that increases with time from 35+3 −4 eV (TOO1) eV (TOO3). The high energy part of to 52+7 −6 the spectra is well fit by a Comptonization in a thermal plasma with stable Thomson optical depth of τ ∼ 1 and electron temperature which does not significantly change with time (mean value of Te = 84+13 −15 keV). The values of kTe and τ we find are very similar to those typical of more luminous black-hole binaries in the hard state (e.g., LX ∼ 0.02LE in Cyg X-1), whereas the X-ray luminosity in our measurement is only ∼ 10−3 LE . A Compton reflection component is apparent and stable, even if weak (mean value of Ω/2π = 0.21+0.05 −0.04 ). This component is only detected when the reflector metallicity is left free to vary in the fits, but the BeppoSAX data, in particular those of the third observation, definitely point to a reflector with a non-solar metallicity (mean value of Z/Z = 0.13+0.06 −0.04 ), as expected by an environment at high galactic latitude, as it is the case for XTE J1118+480. No Fe Kα line is detected (2σ upper limit of ∼ 2×10−4 phot cm−2 s−1 ). For each observation, a timing analysys of the light curves in the energy bands 0.1–1.5, 1.5–10 and 15–200 keV was performed by using fourier techniques. The results are reported in Table 2. The QPOs centroids were determined by fitting each average power density spectrum with a Lorentzian plus a smoothly broken power–law function (e.g. [17]). The centroid frequency of
the source QPOs is found higher during the second observation, consistently with the monitoring results found by [11], but we find that in the third observation (2000 Jume 26), unexpectedly it again decreases, going back to the April 12 value [11]. No statistically significant X–ray emission was detected in the long TOO observation of December 2000 (TOO4). The 2σ upper limit of the quiescent source flux in the 0.1–2, 2–10 and 15–200 keV energy bands is given by 1.2 × 10−13 erg cm−2 s−1 , 4.9 × 10−14 erg cm−2 s−1 and 1.9 × 10−11 erg cm−2 s−1 , respectively. Assuming a distance of 1.5 kpc, the corresponding 2σ upper limit of the source luminosity is 3.2 × 1031 erg s−1 , 1.3 × 1031 erg s−1 , 5.1 × 1033 erg s−1 , respectively. These upper limits are in agreement with the luminosity estimates derived for other black hole candidates with the Chandra satellite [18]. 4. DISCUSSION Our results are consistent with the assumption, discussed by Mirabel et al. [19], that XTE J1118+480 is the relic of an ancient massive star formed in the Galactic halo. On the basis of the fit results, a hot central disk appears the best scenario for the high energy photons. As far as the origin of the soft photons is concerned, the low amplitude of reflection and the iron line intensity would point to the assumption that they are bb
L. Amati, F. Frontera / Nuclear Physics B (Proc. Suppl.) 132 (2004) 392–395
395
radiation from the outer cold disk. However, taking into account the temporal properties, in particular the non-poissonian variance of the time variations of the soft X–ray flux, likely a relevant part of them have a different origin, probably synchrotron emission internal to the hot disk. A more detailed presentation and discussion of the results reported here can be found in [20]. REFERENCES
Figure 1. The BeppoSAX spectra (crosses) fitted by (solid curve) a bb plus compps model assuming a slab geometry of the Comptonizing electron cloud and the metallicity of the reflecting medium free to vary (see Table1). The compps best fit model is shown separately by dashed curves while the direct bb is shown by dotted lines. The dot– dashed curve gives the reflection component. The systematic presence of a reflection component is apparent, while the bb component is reduced in the third observation. The spectra are normalized to the level of the MECS (see text). The bottom side of each panel shows the fit residuals.
1. Remillard, R. A., et al. 2000, IAU Circ. 7389 2. Wagner, R. M., Foltz, C. B., Starrfield, S. G., & Hewett, P. 2000, IAU Circ. 7542 3. Garcia, M.R., et al. 2000, IAU Circ. 7392 4. McClintock, J., Garcia, M., Zhao, P., Caldwell, N., & Falco, E. 2000, IAU Circ. 7542 5. McClintock, J. E., Garcia, M. R., Caldwell, N., Falco, E. E., Garnavich, P. M., & Zhao, P. 2001, ApJ, 551, L147 6. Wagner, R. M., Foltz, C. B., Shahbaz, T., Casares, J., Charles, P., Starrfield, S. G., & Hewett, P. 2001, ApJ, 556, 42 7. Fender, R.P. 2001b, MNRAS, 322, 31 8. Frontera, F., et al. 2001, ApJ, 561, 1006 9. McClintock, J. E. et al. 2001, ApJ, 555, 477 10. Markoff, S., Falcke, H. & Fender 2001, A&A, 372, L25 11. Wood, K. et al. 2000, ApJ, 544, L45 12. Boella, G., et al. 1997, A&AS, 122, 327 13. Magdziarz, P. & Zdziarski, A. A. 1995, MNRAS, 273, 837 14. Titarchuk, L. G. 1994, ApJ, 434, 570 15. Poutanen, J., & Svensson, R. 1996, ApJ, 470, 249 16. Wardzi´ nski, G., & Zdziarski, A. A. 2000, MNRAS, 314, 183 17. Lazzati, D. 2002, MNRAS, 337, 1426 18. Garcia, M.R., et al. 2001, ApJ, 553, L47 19. Mirabel, I. F. et al. 2001, Nature, 413, 139 20. Frontera, F., et al. 2003, ApJ, 592, 1110
BeppoSAX study of the spectral and temporal behavior of the black hole candidate XTE J1118+480 L. Amatia and F. Fronteraab a
Istituto di Astrofisica Spaziale e Fisica Cosmica, CNR, Via P. Gobetti 101, I-40126 Bologna, Italy
b
Dipartimento di Fisica, Universit´ a degli studi di Ferrara, Via Paradiso 12, I-44100 Ferrara, Italy
The high galactic latitude black–hole candidate XTE J1118+480 was followed–up by BeppoSAX during its outburst and transition to quiescence with four TOOs from April to December 2000. In this paper, we summarize and briefly discuss the results of the spectral and temporal data analysis of these observations.
1. INTRODUCTION The compact object of the Low Mass X–ray Binary (LMXB) XTE J1118+480, a system discovered with the All-Sky Monitor (ASM) aboard the Rossi X-Ray Time Explorer (RXTE) satellite during a low state mini–outburst in X–rays in early 2000 [1], is a well-established Black Hole Candidate (BHC). The main characteristics of this system are: high galactic latitude (+48o ), mass of the compact object in the range from 6.0 to 7.7 solar masses [2], radio, near-infrared, optical and EUV counterparts, a surprisingly low X–ray–to–optical flux ratio of ∼ 5, against a typical value of ∼ 500 for LMXB/BHCs [3], a binary period P = 0.1703 days, inclination i = 81 ± 2 degrees, distance d = 1.9 ± 0.4 kpc [4– 6]. From multi-wavelength spectral investigations some relevant properties of the emission processes at the source site were derived. In the radio– submillimiter–near-infrared band, the spectrum shows a shape typical of hard/low spectral states of BHCs and believed to be due to synchrotron emission from self-absorbed jets [7]. Another relevant spectral property is the presence of a thermal component consistent with a blackbody at ∼ 40 eV. This component is apparent in the first observation of XTE J1118+480 with BeppoSAX [8] and it is also detected in the multi-wavelength spectrum of the source derived by McClintock et al. (2001) [9]. Markoff et al. (2001) [10] find the multi-wavelength spectrum from radio to X– 0920-5632/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.nuclphysbps.2004.04.069
rays also consistent with non thermal synchrotron emission from a jet-like matter outflow. Timing analysis of the source revealed, as in several other BHCs in their low/hard state, a low frequency QPO feature. The centroid frequency was observed to continuously drift from ∼ 0.07 Hz to ∼ 0.17 Hz from 2000 April 10 to June 11 [11]. With the BeppoSAX Narrow Field Instruments (NFIs) [12] we observed the source 4 times from April to December 2000. We present a summary and brief discussion of the results of the spectral and timing analysis for all the four BeppoSAX observations. 2. OBSERVATIONS AND DATA ANALYSIS The first three observations were performed in April, May and June 2000, at time intervals of about one month. During the second observation, the LECS was switched off. From 2000 June 30 to October 10, the source was outside the visibility window for BeppoSAX, thus the last BeppoSAX observation was performed in December 2000, long after the ASM outburst was over. Standard data reduction techniques were adopted for all the instruments. The LECS and MECS source spectra were extracted from a region with a radius of 8 around the centroid of the source image. As background spectra we used standard files obtained from the observation of blank fields. We used background spectra accumulated from
L. Amati, F. Frontera / Nuclear Physics B (Proc. Suppl.) 132 (2004) 392–395
393
Table 1 Most relevant best fit parameters of the bb+compps (slab, Z/Z free) and flux variation in the first three observations. Parameter TOO1 TOO2 TOO3 +0.18 20 −2 NH [10 cm ] 1.50−0.17 [1.28] 0.74+0.25 −0.18 +3 kTbb [eV] 35−4 [35] 52+7 −6 Lbb [1034 erg/s]a 5.8+2.2 [5.8] 1.3+2.0 −1.3 −1.2 kTe [keV] 63+20 96+37 90+12 −11 −29 −13 τ 0.64+0.20 0.75+0.29 0.80+0.11 −0.11 −0.24 −0.12 +0.08 +0.08 Ω/2π 0.16+0.11 0.22 0.24 −0.06 −0.12 −0.06 +0.11 +0.14 Z/Z 0.10−0.06 0.17−0.09 0.15+0.09 −0.07 F0.1−2 keV b 5.73 – 5.77 F2−10 keV b 6.95 6.62 5.89 19.4 18.2 15.7 F15−200 keV b 2 χ /dof 194/173 95/105 161/165 a Scaled to d = 1.5 kpc. b In units of 10−10 erg cm−2 s−1 . dark Earth data for the HPGSPC (note that the instrument’s collimator was kept on the source for the entire observation). The energy bands used for spectral fitting were limited to those where the response functions were best known, i.e, 0.12–4 keV, 1.7–10 keV, 7–29 keV, and 15– 200 keV, for the LECS, MECS, HPGSPC and PDS, respectively. A systematic error of 1% was added in quadrature to the statistical uncertainties of the spectral data, on the basis of the calibration results obtained with the Crab Nebula, that was observed on 10 April 2000, 4 days before the first observation of XTE J1118+480. We allowed for a free normalization of the instruments in multi-instrument fits with respect to MECS. The quoted errors for the spectral parameters correspond to 90% confidence for one parameter (∆χ2 = 2.71). We assumed a source distance of 1.5 kpc and a disk inclination of i = 70. The parameter values within square parentheses in the Tables are kept fixed in the fits. 3. RESULTS The source was detected in the first three observations. A large number of spectral models were tested: power–law (pl), cutoff power–law (cut-
offpl), which is equivalent to the pexriv model [13] in xspec with no Compton reflection, Comptonization model (comptt) of Titarchuk [14], thermal-Compton model ( compps) of Poutanen & Svensson [15], all either singly taken or with the addition of a blackbody (bb), all photoelectrically absorbed (wabs model in xspec). Either spherical (sp) or slab (sl) geometries of the Comptonizing electron cloud as well as solar or free metal abundances Z of the cold medium were tested. In Table 1 we report the results obtained with a bb plus compps model assuming a slab geometry of the Comptonizing electron cloud and the metallicity of the reflecting medium free to vary. These best fitting models, superimposed to the F(E) spectra and together with the residuals, are shown in Figure 1. The column density measured in the first BeppoSAX observation is consistent with the Galactic NHG obtained from radio measurements in the direction to the source (NH (1.28–1.44) × 1020 cm−2 ). However the NH derived from the June observation (TOO3) is lower by about a factor 2. Possibly a local absorption contributes to the absorption measured during the April observation (TOO1), if the radio measurements overestimate the Galactic NH . The source spectra show a soft excess which is
394
L. Amati, F. Frontera / Nuclear Physics B (Proc. Suppl.) 132 (2004) 392–395
Table 2 Main timing analysis results of the 0.01–50 Hz power spectral density TOO Energy band r.m.s. fQP O (keV) % (rms) (×10−2 Hz) 1 0.1–1.5 62±3 7.81+0.29 −0.25 1.6–10 42±2 8.11+0.16 −0.17 15-200 18±1 7.82+0.39 −0.29 2 1.6–10 38±4 12.08+0.12 −0.11 15–200 16±1 12.20+0.35 −0.41 3 0.1–1.5 52±4 7.29+0.41 −0.33 1.6–10 32±2 [7.29] 15–200 14±1 [7.29]
consistent with a bb emission with a temperature that increases with time from 35+3 −4 eV (TOO1) eV (TOO3). The high energy part of to 52+7 −6 the spectra is well fit by a Comptonization in a thermal plasma with stable Thomson optical depth of τ ∼ 1 and electron temperature which does not significantly change with time (mean value of Te = 84+13 −15 keV). The values of kTe and τ we find are very similar to those typical of more luminous black-hole binaries in the hard state (e.g., LX ∼ 0.02LE in Cyg X-1), whereas the X-ray luminosity in our measurement is only ∼ 10−3 LE . A Compton reflection component is apparent and stable, even if weak (mean value of Ω/2π = 0.21+0.05 −0.04 ). This component is only detected when the reflector metallicity is left free to vary in the fits, but the BeppoSAX data, in particular those of the third observation, definitely point to a reflector with a non-solar metallicity (mean value of Z/Z = 0.13+0.06 −0.04 ), as expected by an environment at high galactic latitude, as it is the case for XTE J1118+480. No Fe Kα line is detected (2σ upper limit of ∼ 2×10−4 phot cm−2 s−1 ). For each observation, a timing analysys of the light curves in the energy bands 0.1–1.5, 1.5–10 and 15–200 keV was performed by using fourier techniques. The results are reported in Table 2. The QPOs centroids were determined by fitting each average power density spectrum with a Lorentzian plus a smoothly broken power–law function (e.g. [17]). The centroid frequency of
the source QPOs is found higher during the second observation, consistently with the monitoring results found by [11], but we find that in the third observation (2000 Jume 26), unexpectedly it again decreases, going back to the April 12 value [11]. No statistically significant X–ray emission was detected in the long TOO observation of December 2000 (TOO4). The 2σ upper limit of the quiescent source flux in the 0.1–2, 2–10 and 15–200 keV energy bands is given by 1.2 × 10−13 erg cm−2 s−1 , 4.9 × 10−14 erg cm−2 s−1 and 1.9 × 10−11 erg cm−2 s−1 , respectively. Assuming a distance of 1.5 kpc, the corresponding 2σ upper limit of the source luminosity is 3.2 × 1031 erg s−1 , 1.3 × 1031 erg s−1 , 5.1 × 1033 erg s−1 , respectively. These upper limits are in agreement with the luminosity estimates derived for other black hole candidates with the Chandra satellite [18]. 4. DISCUSSION Our results are consistent with the assumption, discussed by Mirabel et al. [19], that XTE J1118+480 is the relic of an ancient massive star formed in the Galactic halo. On the basis of the fit results, a hot central disk appears the best scenario for the high energy photons. As far as the origin of the soft photons is concerned, the low amplitude of reflection and the iron line intensity would point to the assumption that they are bb
L. Amati, F. Frontera / Nuclear Physics B (Proc. Suppl.) 132 (2004) 392–395
395
radiation from the outer cold disk. However, taking into account the temporal properties, in particular the non-poissonian variance of the time variations of the soft X–ray flux, likely a relevant part of them have a different origin, probably synchrotron emission internal to the hot disk. A more detailed presentation and discussion of the results reported here can be found in [20]. REFERENCES
Figure 1. The BeppoSAX spectra (crosses) fitted by (solid curve) a bb plus compps model assuming a slab geometry of the Comptonizing electron cloud and the metallicity of the reflecting medium free to vary (see Table1). The compps best fit model is shown separately by dashed curves while the direct bb is shown by dotted lines. The dot– dashed curve gives the reflection component. The systematic presence of a reflection component is apparent, while the bb component is reduced in the third observation. The spectra are normalized to the level of the MECS (see text). The bottom side of each panel shows the fit residuals.
1. Remillard, R. A., et al. 2000, IAU Circ. 7389 2. Wagner, R. M., Foltz, C. B., Starrfield, S. G., & Hewett, P. 2000, IAU Circ. 7542 3. Garcia, M.R., et al. 2000, IAU Circ. 7392 4. McClintock, J., Garcia, M., Zhao, P., Caldwell, N., & Falco, E. 2000, IAU Circ. 7542 5. McClintock, J. E., Garcia, M. R., Caldwell, N., Falco, E. E., Garnavich, P. M., & Zhao, P. 2001, ApJ, 551, L147 6. Wagner, R. M., Foltz, C. B., Shahbaz, T., Casares, J., Charles, P., Starrfield, S. G., & Hewett, P. 2001, ApJ, 556, 42 7. Fender, R.P. 2001b, MNRAS, 322, 31 8. Frontera, F., et al. 2001, ApJ, 561, 1006 9. McClintock, J. E. et al. 2001, ApJ, 555, 477 10. Markoff, S., Falcke, H. & Fender 2001, A&A, 372, L25 11. Wood, K. et al. 2000, ApJ, 544, L45 12. Boella, G., et al. 1997, A&AS, 122, 327 13. Magdziarz, P. & Zdziarski, A. A. 1995, MNRAS, 273, 837 14. Titarchuk, L. G. 1994, ApJ, 434, 570 15. Poutanen, J., & Svensson, R. 1996, ApJ, 470, 249 16. Wardzi´ nski, G., & Zdziarski, A. A. 2000, MNRAS, 314, 183 17. Lazzati, D. 2002, MNRAS, 337, 1426 18. Garcia, M.R., et al. 2001, ApJ, 553, L47 19. Mirabel, I. F. et al. 2001, Nature, 413, 139 20. Frontera, F., et al. 2003, ApJ, 592, 1110