2003ApJ...597..513S

I propose a scenario to account for the fast polar outflow detected to be blown by η Carinae in 2000. The scenario also accounts for the lack of this flow in 1998 and 1999. The scenario is based on the binary nature of η Carinae. The collision of the winds blown by the two stars, in particular near periastron passages, slows down ∼5% of the wind blown by η Carinae, i.e., the massive primary star, such that it stays bound to the system. I assume that most of this mass is accreted back through the equator by η Carinae during apastron passage, when its orbital velocity is much lower. The mass accretion rate of this back-flowing material may become ∼25% of the wind's mass-loss rate near apastron passages. If the back-flowing matter has enough specific angular momentum, it can form an accretion disk and may lead to the formation of a polar collimated fast wind (CFW) on top of the stellar wind. The gaps and uncertainties in this scenario, which should be closed and refined in future theoretical works, as well as several predictions that can be tested with observations in the near future, are discussed.