3 The cyclic feeding of Sgr A*

A few million years ago, the GC underwent a burst of star formation (Tamblyn et al. 1996) which resulted in a number of clusters of massive stars (IRS 16, the Arches Cluster, and the Quintuplet Cluster). Remnants of this mini-starburst can still be seen in GC star-forming regions such as the "50 km s^-1 cloud'' (Serabyn et al. 1992; Mezger et al. 1989). The central early-type cluster may have formed more than a parsec away from Sgr A* and then, over a million years, spiraled inward via dynamical friction (Gerhard 2001). The disrupted remnant of the parent molecular cloud may be seen as the "ionized gas halo'' which fills the central 10 parsecs (Anantharamaiah et al. 1999).

Then, ${\sim}10^4$ yrs ago, a 13-20 $M_{\odot}$ star, located ${\sim} 5$ pc east of Sgr A*, exploded in a mixed-morphology type II supernova (SN), sweeping up the "ionized gas halo'' and producing what we now see as Sgr A East (Maeda et al. 2000). The eastern edge of the explosion has been confined by the "50 km s^-1 cloud'' while the western edge has interacted with Sgr A* and the central cluster. Indeed, a few thousand years ago, after overwhelming the winds from the mass-losing stars of the early-type cluster, the dense frontal shock of the explosion swept over Sgr A*, triggering a period of high accretion rate and X-ray luminosity for the blackhole (Baganoff et al. 2001).

This period of high accretion lasted until the ram pressure of the front shock near Sgr A* dropped beneath that of the central early-type cluster's winds ( ${\sim} 10^{(-7)-(-6)}$ dynes cm^-2). Based on SN models by Pittard et al. (2001), this "binging'' by Sgr A* lasted 10^2-3 yrs. For the last thousand years or so, after the passage of the dense shell, the IRS 16 and other early-type stars have been "purging'' the central parsecs of the hotter, less dense post-shock SN cavity material. Sgr A West is ${\sim} 1$ pc in radius, consistent with the volume these stars could have cleared in ${\sim}10^3$ yrs while the arms of the "mini-spiral'' are tendrils of infalling gas from this inherently unstable process. Thus, $t_{\rm w} \sim 10^3$ yrs and $\bar n \sim 10^2~{\rm cm}^{-3}$, consistent with the X-ray observations. The lower density also reduces the expected IR thermal bremsstrahlung emission from the central arcsecond to below that which is observed (Menten et al. 1997). Note that radiation pressure from the early-type cluster is not likely to influence the accumulation of gas in the central arcsecond. For radiation pressure to be significant, the luminosity of the cluster, estimated to be , would need to be larger than the Eddington luminosity for Sgr A*. This is almost certainly not the case (Latvakoski et al. 1999).