9. Nova Muscae and 1E 1740.7-2942
Recent observations with SIGMA telescope have revealed annihilation features in the vicinity of keV in spectra of two Galactic black hole candidates, 1E 1740.7-2942 (hereafter the 1E source; Bouchet et al. 1991; Sunyaev et al. 1991; Churazov et al. 1993; Cordier et al. 1993), and Nova Muscae (Sunyaev et al. 1992; Goldwurm et al. 1992). During all periods of observation the hard X-ray emission, 35-300 keV, was found to be consistent with the same law. Observations of Nova Muscae after the X-ray flare (January 9, 1991) are well fitted by a power law of index 2.4-2.5 or by Sunyaev-Titarchuk (1980) model with keV and in the disc geometry, the spectrum of the 1E source is well described by Sunyaev-Titarchuk model with keV and . Meanwhile soft -ray emission of these sources seems to be highly variable.
During the last 13 hr of a 21 hr observation on January 20-21, 1991, a clear emission feature around 500 keV was found in the spectrum of Nova Muscae (Fig. 6), with a line flux of photons cm-2 s-1, and an intrinsic line width keV (Sunyaev et al. 1992; Goldwurm et al. 1992). Since the first 8 hr of the observation did not give a positive detection, the inferred rise time is equal to several hours. The next observation, held on February 1-2, did not show this feature restricting the lifetime to days.
The Galactic center region was intensively monitored by SIGMA telescope since 1990. Three times during these years a broad excess was observed in the 200-500 keV region (Fig. 6).
In an observation performed between 1990 October 13 and 14, a spectacular unexpected feature was found in the 1E emission spectrum, the corresponding flux was estimated at photons cm-2 s-1, with a line width of 180-240 keV (Bouchet et al. 1991; Sunyaev et al. 1991). The observations of this region performed two days before (on October 10-11), and a few hours after (October 14-15) did not exhibit any spectral feature beyond 200 keV. The total duration of this state is estimated between 18 and about 70 hr.
Seven October 1991 observations have shown an evident excess at high energies, while the source was in a low state (Churazov et al. 1993). The excess was observed during 19 days and was not so intensive as in October 1990, the average flux was photons cm-2 s-1 in the 300-600 keV region.
The 1992 September 19-20 observational session (Sep. 19.42-20.58) showed a feature beyond 200 keV (Cordier et al. 1993), which resembles that of 1990 October 13-14. The line flux was estimated as photons cm-2 s-1. The previous (Sep. 18.59-19.30) and the next (Sep. 22.57-23.14) sessions did not show any evidence for emission in excess of 200 keV, restricting the lifetime of the state between 27 and about 75 hr, while the rise time approaches probably several hours.
The spectral features observed by SIGMA are, commonly believed, related to electron-positron annihilation. Relatively small line widths imply that the temperature of the emitting region is quite low, keV for 1E and 4-5 keV for Nova Muscae. Since the hard X-ray spectra keV showed no changes, most probably that electron-positron pairs produced somewhere close to the central object were injected into surrounding space where they cool and annihilate. Radiation pressure of a near-Eddington source alone can accelerate -plasma up to the bulk Lorentz factor of (Kovner 1984), while Comptonization by the emergent radiation field (Levich & Syunyaev 1971) could provide a mechanism for cooling the pairs which further annihilate "in flight" (for a discussion see also Gilfanov et al. [1991, 1994]). If there is enough matter around a source, then particles slow down due to Coulomb energy losses and annihilate in the medium. We explore further this last possibility by checking whether the inferred parameters of the emitting region are consistent with those obtained by other ways. We assume single and short particle ejection on a timescale of hours. It seems reasonable: since the ejection would probably impact on the whole spectrum, longer spectral changes would be observable.
where is the initial Lorentz factor of the plasma stream, c is the light speed, and is the characteristic time scale of the annihilation line appearance. The Coulomb energy loss rate in a medium of is (see Fig. 1). Taking a reasonable value for the bulk Lorentz factor, (e.g., Kovner 1984), one can obtain estimations of the order of magnitude as cm-3 days for the 1E source, and cm-3 hr for Nova Muscae.
If the energetic particles were injected into the medium only once, then the annihilation feature lifetime is directly connected with annihilation rate as . It yields one more estimation of the number density in the emitting region
Annihilation rate is a weak function of (see Fig. 1) and we can take it equal to a constant . Total duration of the hard state is hr for the 1E source and days for Nova Muscae, that gives cm-3 and cm-3 days , correspondingly. The values obtained from Eqs. (45)-(46) restrict the electron number density in the volume where particles slow down and annihilate.
Therefore, to be consistent with the annihilation lifetime the annihilation rise time for the 1E source should be hr. This is supported by the 1992 September 19-20 observation when the annihilation rise time was restricted by a few hours.
The size of the emitting region could be estimated from a simple relation if we assume the upper limit for the positron number density . It gives cm day cm for 1E and cm days for Nova Muscae 2, which are well inside of the upper limits cm hr and cm, correspondingly. From the above consideration follows that emitting regions in both sources are optically thin and do not affect the Comptonized spectra at keV nor the annihilation line form. Experimental data and the estimated parameters are summarized in Table 1.
Table 1. Observational data and parameters of the emitting region.
The column density of the medium where injected particles slow down and annihilate should exceed the value , which follows from previous estimations for and , viz. cm-2 day cm-2 for 1E, where we took into account Eq. (47), and cm-2 days for Nova Muscae. The total column density of the gas cloud measured along the line of sight, where the 1E source embedded, is high enough cm-2 (Bally & Leventhal 1991; Mirabel et al. 1991). Note that recent ASCA measurements of the column density to this source give a best fit value cm-2 (Sheth et al. 1996). For Nova Muscae the corresponding value is cm-2 (Greiner et al. 1991), less or marginally close to the obtained lower limit. If, on contrary, one suggests it yields a condition cm-2 days , which considerably exceeds the measured value.
These estimations put us on to an idea that the 500 keV emission observed from Nova Muscae was coming from -plasma jet () rather than from particles injected in a gas cloud 3 (), therefore, particles have to annihilate "in flight" producing a relatively narrow line blue- or red-shifted dependently on the jet orientation. If so, then our estimation of the electron number density from annihilation time scale is related to the average electron/positron number density in the jet, its total volume is of cm3 days . The reported 6%-7% redshift of the line centroid (Goldwurm et al. 1992; Sunyaev et al. 1992) supports probably the annihilation-in-jet hypothesis, although authors noted that statistical significance of this shift is not very high. The large size of the emitting region and a small width of the line, both except the gravitational origin of the redshift, since in this case the annihilation region have to be quite close to the central object where typical flow velocities should result in a much broader line (Gilfanov et al. 1991). The Compton scattering of the anisotropic emergent radiation could provide effective mechanism for blowing away and acceleration of -pair plasma (e.g., Kovner 1984; Misra & Melia 1993) cooling it at the same time. Since the maximal energy during the X-ray flare of Nova Muscae released near keV (Greiner et al. 1991), the average kinetic energy per particle should be nearly the same (which is consistent with the small line width).
The case of the 1E source is not definitively clear, because our estimations give in the emitting region. Two flares, October 1990 and September 1992, have shown very similar time scales, spectra and photon fluxes, which are consistent with single injection of energetic particles into the thermal (hydrogen) plasma. Meanwhile, the redshift of the line % reported by authors (Bouchet et al. 1991; Sunyaev et al. 1991; Cordier et al. 1993) implies that positrons probably annihilate in a plasma stream moving away from the observer. The estimation of the size of the emitting region ruled out its gravitational nature, since it is too large in comparison with gravitational radius of a stellar mass black hole. A natural explanation of this controversial picture is that the propagating plasma stream captures matter from the source environment and annihilation occurs in a moving plasma volume. In this case the estimation of the electron number density is related to the average electron number density in the jet, cm3 day gives its total volume, and the jet length has to be of the order of cm day .
While a part of the -pair probably annihilate in a thermal plasma near the 1E source producing the broad line, the remainder could escape into a molecular cloud, which was found to be associated with the 1E source (Bally & Leventhal 1991; Mirabel et al. 1991). The time scale for slowing down 4 due to the scattering could be obtained from Eq. (45). Taking cm-3 for the average number density of the molecular cloud near 1E (Bally & Leventhal 1991; Mirabel et al. 1991) one gets year, the same as that obtained by Ramaty et al. (1992). The size of the turbulent region in the cloud caused by propagation of a dense jet should be of the same order. It agrees well with the length 2-4 ly (15-30 arcsec at the 8.5 kpc distance) of a double-sided radio jet from the 1E source found recently with the VLA (Mirabel et al. 1992).
If the lines from the 1E source (Fig. 6) were produced by continuous injection of energetic particles, then the observations of the narrow 511 keV line emission from the Galactic center allows to put an upper limit on the particle escape rate into the interstellar medium. Recent reanalysis of HEAO 3 data has shown that under suggestion of a single point source at the Galactic center narrow line intensities are photons cm-2 s-1 for the fall of 1979 and photons cm-2 s-1 for the spring of 1980 (Mahoney et al. 1994). Taking yr for the positron lifetime in cm-3 dense cold molecular cloud (Ramaty et al. 1992), and suggesting one hard state of days long per period , one can obtain an escape rate , where we took photons cm-2 s-1 (see Table 1). This is consistent with the upper limits of 1990 October 13-14 spectrum and the two most energetic points in 1992 September 19-20 spectrum. The dashed line in 1990 October 13-14 spectrum (Fig. 6) shows the annihilation line shape for Gaussian-like injection, , of energetic particles into the thermal plasma of keV for . The longest hard state ( days) with the average flux of photons cm-2 s-1 observed in October 1991 places the upper limit at almost the same level .
© European Southern Observatory (ESO) 1997
Online publication: May 5, 1998