4. The origin of the broad emission feature at 6.7 keV

A possible source of the 6.7 keV line emission is the surface of the companion star which faces the pulsar. The considerable X-ray heating of the stellar surface might produce the high temperatures and ionization required for a broad 6.7 keV emission line. In fact Day and Stevens (1993) have suggested that ionization conditions at the companion's surface are so extreme that a radiatively driven stellar wind cannot exist and the wind is instead driven by X-ray heating. It is possible that if one of the pulsar beams is sweeping across the face of the companion star it could cause fluorescence, giving rise to a pulsing 6.7 keV line. However, since the line undergoes only a partial eclipse (Nagase et al. (1992)) at least some of the emission must come from a more extended source such as an ionized stellar wind. Another possibility is that the source of the 6.7 keV line is relatively compact and that the partial eclipse seen in the Ginga data and the broadening might be due to scattering in an extended medium. If a significant fraction of the photons from this line are scattered by electrons in the stellar wind, this might explain the partial eclipse seen in the Ginga data. However, the column density of the wind (Nagase et al. (1992)) is too low to produce significant Compton scattering ($\tau\sim \sigma_T N_H\sim 10^{-2}$
).

If the 6.7 keV line is produced by fluorescence in the stellar wind its observed equivalent width should be compatible with the wind column density. An estimate of the equivalent width from K fluorescence of iron in an optically thin slab is given by Kallman (1991) as $$EW = \varepsilon_{line}\,{{L(\varepsilon_{th})}\over{L(\varepsilon_{line})}} \,\sigma_{th}\,\omega_K \,x\,y\,N$$
where $\varepsilon_{line}^{}$ is the line energy, $L(\varepsilon_{th})$
is the ionizing specific luminosity at the threshold energy for K shell photoionization, $L(\varepsilon_{line})$
is the ionizing specific luminosity at the line energy, $\sigma_{th}$
is the cross section for K shell photoionization at the threshold energy, $\omega_{K}^{}$ is the fluorescence yield, x is the fraction of the iron in the ionization state of interest, y is the iron abundance, and N is the equivalent hydrogen column density. For a 6.7 keV line due to emission by Fe XXV, $\omega_{K}^{}$ $\sim$ 0.5 (Krolik and Kallman (1987)), $\varepsilon_{th}^{}$ = 8.8 keV (Lotz (1968)), and $\sigma_{th}^{}$ = 2 x 10^-20 cm² (Reilman and Manson (1978)). Taking y = 3 x 10^-5 (solar abundance) yields an equivalent width

EW = 1.5 keV N₂₄ x (86)

where N₂₄ is the equivalent hydrogen column density in units of 10²⁴ cm^-2. An equivalent width of $\sim$ 120 keV implies a column density of $\sim$ 8 x 10²² cm^-2, assuming that x = 1. From the size of the 6.7 keV line emission site and the ratio of the scattering continuum to the intrinsic emission Nagase et al. (1992) estimated a column density of stellar wind of a few times  10²² cm^-2. Thus the column density of the ionized wind may be sufficient to produce the observed equivalent width.

Bulk motions of the wind might explain the large width of the 6.7 keV line. Speeds of the order of 9000 km s^-1 would be required to produce a broadening $\Delta$E_FWHM $\sim$ 400 keV. Another possible broadening mechanism is the thermal motions of the wind. However an ion temperature of $\sim$ 10¹¹ K would be required to produce the observed width. If the broad 6.7 keV line originates near the magnetosphere where there are very high rotation velocities ( $\sim$ 5000 km/s) Doppler broadening could easily account for the observed width.

A more likely explanation is that the 6.7 keV line is a blend of several narrow resonance and fluorescence lines of Fe XXI-Fe XXVI with energies ranging from 6.5 to 6.9 keV. This interpretation is supported by ASCA observations of Cen X-3 during eclipse (Ebisawa et al. (1996)). The iron K emission was resolved into three lines at 6.4, 6.7, and 7.0 keV. While one would expect the spectrum of the source to be qualitatively different during eclipse because of the obscuration of the pulsar, it is not unlikely that these emission lines could be present outside of eclipse. Figure 3.4 shows the data with such a multiple line fit. A 2$\sigma$ narrow line at 7.5 keV is also included which may be an instrumental feature due to Ni  K_$\alpha$. The narrow lines have equivalent widths typically $\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$<$}}}50\ \rm eV$
which is consistent with fluorescence in the stellar wind. Although the fit is not statistically better than for the two line model, the most plausible explanation for the large physical width of the iron emission feature is that it is in fact a blend of narrow lines. The fact that using physically plausible line energies in the multiple line model does not make the fit worse also supports this. These data are consistent with a picture in which the H- and He-like lines originate in the photoionized stellar wind, which accounts for the gradual eclipse of the features observed with Ginga. The 6.4 keV line would be produced mainly by fluorescence of material irradiated by the X-ray beams. A clear prediction of this model is that the 6.4 keV line should be pulsed at the 4.8 s spin period. Day et al. (1993) were unable to determine which of the iron lines were pulsating. For the BBXRT data counting statistics only allow an upper limit of 70% on the amplitude of any pulsations.

Figure 8: Folded A0 data fit to four narrow lines with energies 6.4, 6.7, 6.93, and 7.53 keV. These energies correspond to emission by un-ionized iron, helium- and hydrogen-like iron, and un-ionized nickel.