Wind-wind collision in the η Carinae binary system: a shell-like event near periastron

The exact nature of η Carinae is still an open issue. Strict periodicity in the light curves at several wavelengths seem to point out to a binary system, but the observed radial velocities, measured from space with high spatial resolution are in conflict with the ground based observations used to calculate the binary orbit. Also, the observed 2-10 keV X-ray flux is much larger that what is expected from a single star, and favors the wind-wind collision hypothesis, characteristic of high mass binary systems. However, to explain the duration of the dip in the light curve by wind collisions, it is necessary to postulate a very large increase in the η Carinae mass loss rate. Finally, the optical and UV light curves are better explained by periodic shell-ejection events. In this paper we conciliate the two hypothesis. We still assume a binary system to explain the strong X-ray emission, but we also take into account that, near periastron and because of the highly eccentric orbit, the wind emerging from η Carinae accumulates behind the shock and can mimic a shell-like ejection event. For this process to be effective, at periastron the secondary star should be located between η Carinae and the observer, solving also the discrepancy between the orbital parameters derived from ground and space based observations. We show that, as the secondary moves in its orbit, the shell cools down and the number of available stellar ionizing photons is not enough to maintain the shell temperature at its equilibrium value of about 7500 K. The central part of the shell remains cold and under these conditions grain formation and growth can take place in timescales of hours. Using recently published HST data for the optical and UV continuum during the 2003 shell event, we estimated that a fraction of 0.01% of the interstellar dust to H ratio is enough to explain the observations. We also calculated the neutral gas column density intercepting the line of sight at each point of the orbit near periastron, and were able to reproduce the form and duration of the X-ray light curve without any change in the η Carinae mass loss rate. This same column density can explain the observed Hα light curve observed during the 2003 event.