Table 1 lists the Ginga Large Area Counter (LAC) observations of 4U 1630-47. The LAC comprised eight multi-wire collimated proportional counters, sensitive in the energy range 1.7-37 keV. Details of the MPC1 and MPC2 data compression modes used in these observations are given in Turner et al. (1989). Background subtraction of Ginga LAC data can be performed in a number of ways (Hayashida et al. 1989). One method, often used for sources near the galactic plane, utilizes a nearby off-source measurement obtained just before or after the relevant observation. However, suitable background exposures are not available for any of the observations listed in Table 1. Instead, we estimate the intrinsic background from high galactic latitude source-free observations obtained over three month intervals around the times of observation. Since 4U 1630-47 lies close to the galactic plane (, = , ), there is an additional background component due to diffuse X-ray emission (Warwick et al. 1985; Koyama 1989). The shape of this emission is consistent with bremsstrahlung with a narrow emission feature at 6.7 keV superposed (Koyama 1989).
Table 1. Ginga observations of 4U 1630-47
A bremsstrahlung model with variable temperature and normalization together with a narrow emission feature at 6.7 keV was first fit to each of the background subtracted spectra corresponding to the observations listed in Table 1. To avoid complications due to collimator reflection (see below) data below 6.5 keV were excluded from these fits. Similarly, to avoid problems with contamination by a known 22 keV line from the collimator (Turner et al. 1989), data within 2.5 keV of this energy were excluded. In the case of the 1988 April and 1988 October observations this gives satisfactory (at 95% confidence) fits with 's of 40 and 20, respectively, for 28 degrees of freedom (dof). In the case of the 1987 October observation, the fit is unacceptable with a of 54 for 28 dof. Towards the end of the 1984 outburst of 4U 1630-47, when the 1-50 keV luminosity had fallen by a factor 300 from its maximum, the spectrum could be represented by a power-law with a photon index of 1.2 and absorption, , of H atoms cm-2 (P86). Including such a component in the fits to the 1987 October, and 1988 April and October spectra gives 's of 25, 31, and 19 for 27 dof, respectively. These correspond to values of the F statistic of 35.6, 7.8 and 1.4 implying that this extra component is significant at 99% confidence in the first two observations, but not in 1988 October. The 3 confidence upper-limit 1-10 keV luminosity, L, for an assumed distance of 10 kpc is given in Table 1 for this observation.
This extra emission seen in the 1987 October and 1988 April observations may well originate from 4U 1630-47, but we cannot rule out the possibility of a faint uncataloged source within the full width half maximum field of view (FOV) of the Ginga LAC. In particular, an identical analysis to that described above performed on a supposedly source free observation within of 4U 1630-47 on 1988 March 25 gives a value of the F statistic of 13.5, implying the presence of a source at 99% confidence. Thus, we conclude that we have a probable detection of 4U 1630-47 in 1987 October, but that the detection in 1988 April may well be spurious.
During the 1989 March observation, when 4U 1630-47 is clearly in outburst, Ginga was being operated in an unusual manner. During the two day duration observation, the pointing direction was slowly changed to allow a number of sources on the galactic plane to be scanned. Consequently, the overall exposure to 4U 1630-47 is relatively short and spread over the entire interval. Most of the time, the source was offset in the LAC FOV. This allows scattering of low-energy X-rays off the collimator walls, giving rise to a soft excess (Turner et al. 1989). In order to limit this effect, intervals where the source is viewed with a collimator efficiency of 20% were excluded. From a total on-source time of 6678 s, this reduces the exposure to 1334 s, while the mean collimator transmission increases from 0.13 to 0.49. Although it is possible to directly correct the data to remove the soft excess (Williams & Kellett 1991), such a procedure requires a precise knowledge of the satellite pointing direction, which is not available. Therefore, in the spectral fitting discussed below, the presence of this soft excess is accounted for by adding an identical spectral component to the chosen model that is modulated by an exponential cutoff and a variable normalization (Stewart private com.). Thus, if the original spectral model is f, then the fitted model, , is given by:
where c is the normalization of the scattered component (typically 25% at 1 keV), is the cutoff energy (typically 3.5 keV) and is the folding energy (typically 0.08 keV). All three parameters are allowed to vary in the fitting process and 0.5% uncertainties were added quadratically to account for calibration uncertainties.
The changing and uncertain pointing direction, the lack of suitable background measurements, the contribution of the diffuse galactic emission and the short exposure time, all combine to limit the quality of the 1989 March Ginga spectrum. However, the large collecting area and low intrinsic background of the LAC provides sufficient statistics to justify detailed spectral fitting. Single component power-law, bremsstrahlung, blackbody, cutoff power-law (), and multicolor blackbody disk (Mitsuda et al. 1984) models were fit to the spectrum. Each of these models was modified by low-energy absorption using the coefficients of Morisson & McCammon (1983) and the effects of collimator scattering using Eq. (1). The Mitsuda et al. (1984) disk model was chosen to allow comparison with previous BHXT results (e.g. Tanaka & Lewin 1995).
All the above models gave unacceptable fits to the spectrum. The cutoff power-law and multicolor blackbody disk models coming closest to being acceptable, with 's of 15. The multicolor disk blackbody model assumes that the gravitational energy released by the accreting material is locally dissipated into blackbody radiation, that the accretion flow is continuous throughout the disk and that the effects of electron scattering on the spectrum are negligible. There are only two parameters in the model; where is the innermost radius of the disk, the inclination angle of the disk, the source distance in units of 10 kpc, and the blackbody effective temperature at . If a power-law component of photon index, , and normalization, , is added to these two models, then significantly better fits are obtained with 's of . The results of fitting these combined models to the extracted spectrum are presented in Table 2. All spectral uncertainties are given at 68% confidence. Fig. 1 illustrates the best-fit cutoff power-law and power-law model. The best-fit parameters and 1-50 keV luminosity ( erg s-1, assuming a distance of 10 kpc) are similar to those obtained by P86 during the first two EXOSAT observations of 4U 1630-47. The best-fit value of is a factor 10 higher than the average of 30 km seen from other BHXT systems (Tanaka & Lewin 1995). This may suggest that the black hole in 4U 1630-47 is more massive than in these other systems, or it may reflect inadequacies in the spectral modeling. In particular, examination of the residuals in Fig. 1 suggests the presence of an absorption feature at 6 keV, perhaps resulting from a reflection component (e.g. Ebisawa 1991). Including a smeared edge in the spectral model does indeed produce significantly better fits, but is not justified given the uncertainties in background subtraction.
Table 2. Fit results to the 1989 March Ginga spectrum
The ASCA archive was searched for observations of the region of sky containing 4U 1630-47 and two short observations made on 1994 September 3 between 15:15 and 16:54 UTC and between 16:55 and 18:34 UTC were found. A bright absorbed source, at a position consistent with 4U 1630-47, is visible in both observations in the FOVs of both Gas Imaging Spectrometers (GIS; Tanaka et al. 1994) at offsets of between and . The source is outside the FOV of the SIS detectors in both observations, thus only GIS data were considered. Due to the alignment of the two GIS instruments and the different pointing in the two observations, only data from GIS3 for the first interval and GIS2 for the second interval were analyzed. In each case the source is located off-axis. In the GIS2 first interval and GIS3 second interval, the source is located in regions of high background counting rate and uncertain gain calibration at offset angles of . The standard data selection filters of an Earth elevation angle of , and a cutoff rigidity of 6 GeV c-1 were applied. This gives a total exposure of 1600 s for the first interval and 1880 s for the second. The spectra were extracted using events accumulated within radii centered on the source positions and the backgrounds estimated from identically sized regions located diametrically opposite in the FOVs. The mirror vignetting correction was applied giving a count rate of 13 s-1, during both observations. For spectral fitting, the response matrices "gis2v4_0" and "gis3v4_0" (1995 March 2) provided by the ASCA Guest Investigator Facility and corrected for vignetting were used. The spectra were rebinned to have at least 20 photons in each channel and simultaneously fit.
The same spectral models as applied to the Ginga data were used, except that a high-energy power-law component is not necessary due to the limited energy response of the GIS, which results in few detected counts above 8 keV. Both the cutoff power-law and the multicolor blackbody disk models gave satisfactory fits with similar values of of 1. The best-fit parameters are listed in Table 3 and are significantly different from those derived by Ginga. Fig. 2 shows the best-fit multicolor blackbody disk model fit to the GIS spectra. The best-fit parameters for both models give a 1-50 keV luminosity of (2.0-2.4) erg s-1. This is a factor 4 lower than during the second EXOSAT observation (P86), which occurred 40 days after the outburst start (P95). The spectral parameters given in P86 for the cutoff power-law model for the second EXOSAT observation are different from the best-fit values derived here using ASCA. However, if is fixed at a value of -2, consistent with the EXOSAT observations, the fit to the ASCA spectra still gives an acceptable of 499 for 511 dof. The best-fit temperature is now keV, comparable with that during the second EXOSAT observation. The 90% confidence upper-limit to a narrow emission line at 6.4 keV is 55 eV.
Table 3. Fit results to the ASCA GIS spectraa
4U 1630-47 was observed by the Einstein HRI (Giacconi et al. 1979) on 1979 February 22 between 21:17 and 23:04 UTC for an exposure of 4132 s and on 1980 February 18 between 17:49 and 18:26 UTC for an exposure of 2007 s. During the 1979 observation, which occurred 13 days before the Einstein SSS observation reported in P95, a source was detected at a location consistent with 4U 1630-47 with a count rate of s-1. (All quoted HRI count rates are corrected for instrument sampling deadtime losses, mirror vignetting, and for events that fall outside the extraction region). The high HRI count rate means that 4U 1630-47 was clearly in outburst during the observation, consistent with the ephemeris in P95 and the SSS detection 13 days later. The position is R.A. = , (J2000) which is from the center of the 90% confidence radius EXOSAT position for 4U 1630-47 in P86.
During the 1980 HRI observation, which occurred at an outburst ephemeris of 0.48 (i.e. approximately mid-way between expected outburst times), a faint source was detected at a position consistent with 4U 1630-47 with a count rate of s-1. Source counts were extracted from a circular region centered on the source with a radius of . The background rates were extracted from a concentric annular region with inner and outer radii of and , respectively. The extraction radii were chosen to optimize the signal to noise ratio. The total number of counts detected was 15 with an expected background of 1.38 counts. The probability that the source arises from a fluctuation in the background counting rate is , assuming Poisson statistics.
© European Southern Observatory (ESO) 1997
Online publication: July 3, 1998