4. Conclusions

The physical width of the 6.7 keV iron line in Cen X-3 has been measured for the first time out of X-ray eclipse. The blending of several lines from highly ionized iron must be invoked to explain the width of this line if it originates in the extended stellar wind. If this is in fact the case, ASCA should be able to easily resolve these lines. The presence of iron emission at 6.4 keV and 6.7 keV simultaneously indicates that there are at least two distinct emission sites. Fluorescence in a localized region of relatively low ionization is most likely responsible for the 6.4 keV emission. The broad 6.7 keV feature is a blend of several narrow lines due to the scattering of radiation from the neutron star in an extended highly ionized stellar wind. One would then expect to observe pulsations in the 6.4 keV line. If the size of the source of the 6.7 keV emission feature is greater than $\sim$ 8 x 10¹¹ cm, as the Ginga observations indicate, any pulsations in the 6.7 keV line should be washed out by the light travel time across the source. Another objection to pulsations of the 7.6 keV line caused by illumination of the stellar wind by the beamed X-ray emission of the pulsar is that the pulse might be smeared out by the wind's finite recombination time. Nagase et al. (1992) estimated that the average electron density in the region where $\xi$ > 10³ was 5 x 10¹⁰$\le$n_e$\le$8 x 10¹¹ cm^-3. Using the fitted formulae of Arnaud and Raymond (1992) for the radiative recombination and dielectronic recombination coefficients ( $\alpha_{r}^{}$(T) and $\alpha_{di}^{}$(T), respectively), and assuming n_e = 10¹¹ cm^-3 and T = 10⁴ K, the recombination rate for Fe XXVI will be $$1/\tau_{\rm recomb}=(\alpha_r(T)+\alpha_{di}(T))n_e = 58.4\rm\ s^{-1}\,.$$
Thus Fe XXV that has been photoionized by the pulsar's beam will recombine on a timescale of $\tau_{\rm recomb}\approx0.02\rm\ s$
which is much shorter than the 4.8 s pulse period. Thus, the recombination time of Fe XXVI to Fe XXV should not significantly smear out any pulsation of the 6.7 keV line.

No significant variations with pulse phase in the intensity of either iron line were detected in the BBXRT data. The 90% confidence upper limit on the variation in line strength is 70%, which is consistent with with the detection reported by Day et al. (1993).