M82, NGC 4038/9 (the Antennae) and NGC 3256 are all well-studied starbursts. M82 is a nearby (3Mpc), nearly edge-on dwarf interacting with M81. It is a modest starburst with infrared luminosity . H- and previous X-ray observations have revealed extended soft X-ray emission above and below the plane of the galaxy associated with the superwind (Cappi et al 1999, A&A 350, 777). The Antennae are a pair of interacting galaxies at a distance of 29Mpc, famous for their bright H- knots and super starclusters imaged with HST (Whitmore et al 1999 AJ 118, 1551). NGC 3256 is a classic ``train wreck'' merger induced starburst, with double nuclei and tidal tails. It has been classified as a super starburst on the basis of its high infrared luminosity, and it is the most X-ray luminous starburst known, with (Moran et al 1999, ApJ, 526, 649). M82 and the Antennae were observed with ACIS during cycle 1 as part of the GTO program; M82 has also been observed with the HRC as a calibration target. NGC 3256 was observed with ACIS as a GO observation (PI Ward).
One of the most extreme and controversial examples of a Super-Eddington source is a bright X-ray source that dominates the central region of M82. Pre-Chandra observations of M82 have shown that this source is variable and close to the center of the galaxy. It has been interpreted as a low luminosity AGN, a highly X-ray luminous supernova (the X-ray counterpart to the bright radio supernova remnant 41.95+57.5) and an accreting black hole with a mass in excess of 500 (e.g. Ptak & Griffiths 1999 ApJ 517, L85). Chandra resolves the central ``source'' in M82 into several discrete sources, as shown in Figure 1. Figure 1 shows two HRC images taken 3 months apart. Although Chandra resolves the central ``source'' observed by ROSAT and ASCA, one single source (CXOU J095550.2+694047) still dominates the flux.
It is very highly variable, increasing in flux a factor of 7 between Oct 1999 and Jan 2000. In January it had a luminosity of , and dominated the X-ray flux from the entire galaxy! This source is almost certainly responsible for the variability seen in ASCA and ROSAT observations (Ptak & Griffiths 1999). With the excellent positional accuracy of Chandra we can confidently place CXOU J095550.2+694047 9 arcsec away from the kinematic center of M82, and 4 arcsec away from the radio supernova remnant 41.95+57.5. The large amplitude of the variability suggests that the flux from CXOU J095550.2+694047 really is dominated by a single object, and the fact that it is displaced from the kinematic center of the galaxy indicates that it is not a low luminosity AGN. It is unlikely to be an X-ray supernova remnant, because of the stochastic nature of the variability and the fact that it is not spatially coincident with a radio supernova remnant. Therefore CXOU J095550.2+694047 really does seem to be a ``Super-Eddington'' source, and probably the best documented! See Kaaret et al 2001 (MNRAS 321 L29) and Matsumoto et al 2001 (ApJ 547 25) for even more detail.
How common are Super-Eddington sources? We (Ward et al 2001 in preparation) find thirty-two sources in M82, twelve of which are variable with timescales of a few months. At least four of these may be Super-Eddington. Fourteen sources in the Antennae have been detected with (Fabbiano, Zezas & Murray 2001, ApJ submitted), and another fourteen sources in NGC 3256! The count in NGC 3256 includes the nuclei of both galaxies, which may of course be low luminosity AGN). NGC 3256 is much further away than M82, so the probability of source confusion increases. Nevertheless, these sources are excellent Super-Eddington candidates. The X-ray luminosity function of these starburst galaxies is in stark contrast to that observed in the galaxy or M31 where very few super luminous sources are observed (e.g. Fabbiano, Zezas & Murray 2001). While it is clear that a thorough analysis is necessary to determine whether these differences are statistically significant, it suggests that the brightest non-nuclear X-ray sources may be associated preferentially with starburst galaxies.
What are Super-Eddington sources? The Chandra (and previous ROSAT and ASCA) observations of the canonical Super-Eddington source CXOU J095550.2+694047 strongly suggest some kind of accretion process. If this source is radiating isotropically at the Eddington luminosity, the inferred mass is . There are well documented binary systems in our own galaxy with isotropic luminosities 2-3 times greater than the Eddington luminosity (Orosz et al 1998, ApJ 499, 375). These sources are usually beamed, and the luminous phase associated with an X-ray flare lasts from a day to a month. This explanation cannot be ruled out with the current data. It is therefore possible that CXOU J095550.2+694047 is a black hole of with flaring X-ray jets pointed straight at us! However, CXOU J095550.2+694047 has an (isotropic) luminosity an order of magnitude larger than galactic sources. So if the mass is really as low as the beaming must be much tighter.
There must be a population of un-beamed sources if CXOU J095550.2+694047 and his friends are microquasars, and we must explain why there are such a large number of sources are pointed toward us! One possibility (suggested by King, Davies, Ward, Fabbiano and Elvis 2001, preprint) is that super-Eddington sources are high-mass X-ray binaries, similar to galactic microquasars. The ``ultraluminous'' phase occurs in almost every system, but is short-lived. The short-lived-but-common scenario explains why these sources are preferentially associated with starburst galaxies, which are undergoing short but violent episodes of star formation. There are potentially interesting consequences if CXOU J095550.2+694047 (and other luminous sources like it) really are ``Mid-Mass'' black holes. First of all, as pointed out by Ptak & Griffiths in their paper in 1999, we may be watching the formation of a supermassive black hole. Or at least that the processes by which mid-mass black holes form may tell us how supermassive black holes form, even if not every mid-mass black hole graduates to become supermassive. Secondly, there may be an as yet unknown population of Mid-Mass black holes in galaxies which have gone through a starburst phase (including potentially our own.) Chandra has shown us that the ultraluminous source in M82 really is a single source, and almost certainly an accretion-powered black hole. By extension, most of the other Super-Eddington sources are probably black hole candidates as well. Whether they are High Mass X-ray Binaries going through a microquasar phase, or Mid-Mass black holes (or both!) remains to be determined, and makes these enigmatic objects excellent targets for future studies with Chandra.
- Andrea Prestwich, Phil Kaaret, Andreas Zezas, Pepi Fabbiano, Paulina Lira and Martin Ward