Publication embargo: Until 19:00 April 14.

RESEARCHERS RESOLVE INTERMEDIATE MASS BLACK HOLE MYSTERY

New research, funded by the Royal Netherlands Academy of Sciences, the Institute of Advanced Physical and Chemical Research, NASA and the University of Tokyo, solved the mystery of how a black hole, with the mass more than several hundreds times larger than that of our Sun, could be formed in the nearby starburst galaxy, M82.

Recent observations of the Chandra X-ray observatory (Matsumoto et al., 2001 ApJ 547, L25) indicate the presence of a unusually bright source in the star cluster MGG11 in the starburst galaxy M82. The properties of the X-ray source are best explained by a black hole with a mass of about a thousand times the mass of the Sun, placing it intermediate between the relatively small (stellar mass) black holes in the Milky way Galaxy and the supermassive black holes found in the nuclei of galaxies. For comparison, stellar-mass black holes are only a few times more massive than the Sun, whereas the black hole in the center of the Milky-way Galaxy is more than a few million times more massive than the Sun.

An international team of researchers, using the world's fastest computer, the GRAPE-6 system in Japan, were engaged in a series of simulations of star clusters that resembled MGG11. They used the GRAPE-6 to perform simulations with two independently developed computer programs (Starlab and NBODY4 developed by Sverre Aarseth in Cambridge), both of which give the same qualitative result. The simulations ware initiated by high resolution observations of the star cluster MGG11 by McCrady et al (2003, ApJ 596, 240) using the Hubble Space Telescope and Keck, and by Harashima et al (2001) using the giant Subaru telescope.

The GRAPE's detailed, star-by-star simulations represent the state of the art in cluster modeling. For the first time using the GRAPE, researchers perform simulations of the evolution of young and dense star clusters with up to 600000 stars; they calculate the orbital trajectory and the evolution of each star individually. Using this unique tool, the team found they could reproduce the observed characteristics of the star cluster MGG11. As a bonus, however, the star cluster produces a black hole with a mass between 800 and 3000 times the mass of the Sun. The black hole is produced within 4 million years which is in an early phase in the evolution of the star cluster. During this phase the stellar density in the center becomes so high that physical collisions between the stars become frequent. If the stellar densities exceed a million times the density in the neighborhood of the Sun, collision start to dominate the further evolution of the star cluster.

In this over-dense cluster center, stars experience repeated collisions with each other, resulting in a collision runaway in which a single stars grows to enormous mass. After the central fuel of this star is exhausted, it collapses to a black hole of about 1000 times the mass of the Sun.

New results of these detailed computer simulations, published in Nature show that the star cluster in which the X-ray source resides has characteristics such that a black hole of 800-3000 times the mass of the Sun can form within a very short time. The calculations therewith provide compelling evidence for the process which produces intermediate mass black holes and at the same time provide an explanation for the bright X-ray source observed in the cluster.

The GRAPE team's members are Simon Portegies Zwart, from the University of Amsterdam in the Netherlands; Holger Baumgardt, from RIKEN in Tokyo; Piet Hut, of the Institute for Advanced Study in Princeton, N.J.; Jun Makino from Tokyo University; Steve McMillan, from Drexel University in Philadelphia.

The GRAPE group's results appear in the April 15, 2004, issue of Nature.

Relevant internet addresses:

$\bullet$ http://www.astrogrape.org

$\bullet$ http://www.manybody.org

$\bullet$ http://www.manybody.org/manybody/starlab.html

Contact information:

$\bullet$ Dr. S. Portegies Zwart, University of Amsterdam, Astronomical Institute 'Anton Pannekoek' and Section Computational Science, the Netherlands, tel. +31 (0)20 5257510, email: spz@science.uva.nl

$\bullet$ Dr. H. Baumgardt, RIKEN Tokyo, Japan, tel. +81 (0)48 4679417, email: holger@postman.riken.go.jp

$\bullet$ Prof. P. Hut, Institute for Advanced Study, NJ, USA, tel. +1 609 7348075, email: piet@ias.edu

$\bullet$ Prof. J. Makino, Tokyo University, Tokyo, Japan, email: makino@astron.s.u-tokyo.ac.jp

$\bullet$ Prof. S. McMillan, Drexel Univeristy, Philadelphia, USA, tel. +1 215 8952723, email: steve@kepler.physics.drexel.edu

Figure 1: The spiral galaxy M81 (left) interacting with the starburst galaxy M82 imaged by Robert Gendler. The yellow rectangle indicates the field of view of the Subaru image.

Figure 2: (NHK) Color image taken with the Subaru telescope from the National Astronomical Observatory of Japan (16 Jan 1999). The position of the star cluster MGG11 (image to the right) is at the crosspoint of the two lines.

Figure 3:High resolution Keck image of the star cluster MGG11 by McCrady et al. (2003, ApJ 596, 240).

Figure 4:Artist's rendering, by Dr. Hitoshi Muller (Musashino Art University, Japan), of the compact star cluster MGG11 in the starburst galaxy M82, based on simulations performed by Portegies Zwart et al. Stellar collisions are commonplace in the dense heart of this congested stellar system. The inset shows the result of a hydrodynamical calculation of one such collision, between stars of 53 and 18 times the mass of the Sun, performed by Dr. James Lombardi (Vassar College, USA).

Figure 5:Chandra X-ray image (in R.A. and Dec.) of the relevant part of the starburst galaxy M82 with the observed star clusters indicated. The color image is from the 28 October 1999 X-ray observation by Matsumoto et al. (2001), with about arcsecond positional accuracy. The brightest X-ray source (M82 X-1) is near the center of the image. The star clusters, from table 3 of McCrady et al. (2003), are indicated by circles. The positions of the two star clusters MGG-9 and MGG-11 are indicated with squares. The magnified infrared images of these star clusters from the McCrady et al observations are presented in the upper right (MGG-11) and lower left (MGG-9) corners. A recently discovered 54.4mHz quasi-periodic oscillator is not shown because of its low (7 arcsecond) positional accuracy, but its position is consistent with the X-ray source in MGG-11. A millimeter source roughly centered around the two clusters is not shown either, because it is a large shell-like structure with a diameter of 14''x 9''

Figure 6: The growth in mass of the collision runaway star with time. The choice of initial concentration is labeled by the central potential, where W12(9) implies $W_0 = 12(9)$. For both choices, the top curves give the {\tt NBODY4} results, and the bottom curves the {\tt Starlab} results. The runaway masses in {\tt NBODY4} are larger, since that code adopts larger stellar radii, as discussed in the {\em Supplementary material}.
The star symbols indicate the moment when the runaway experiences a supernova, typically around 3\,Myr. The open and filled stars indicate simulations performed with {\tt NBODY4} and {\tt Starlab}, respectively.
The solid and dashed curves show $M_r$ for a Salpeter IMF with a lower limit of 1\,{\msun} and $c \simeq 2.1$ ($W_0 = 9$) and $c \simeq 2.7$ ($W_0=12$). The dash-dotted curves are for two models with $W_0=9$ with an upper limit to the IMF of 50\,{\msun}, instead of the standard 100\,{\msun} used in the other calculations; we terminated these runs at the moment the runaway star experiences a supernova. The dash-3-dotted curve shows the result for $W_0=12$ with a Salpeter IMF and with 10\% primordial binaries. Finally, the dotted curve shows results for $W_0=9$ and a Kroupa IMF with a minimum mass of 0.1\,\msun, in a simulation with 585,000 stars.
The observed age range of MGG-11 and MGG-9 is indicated by the horizontal bar near the bottom of the figure.

Figure 7: The area of parameter space for which runaway collision can occur, and where the process is prevented. Conditions for runaway merging identified through our simulations in the \{$t_{\rm df}$, $c$\} plane, where $t_{\rm df}$ is the dynamical friction time scale for 100\,{\msun} stars, and $c$ the cluster concentration parameter. The horizontal error bars for the initial conditions for MGG-9 and MGG-11 reflect errors in the observed cluster mass and projected half light radius. The best fit for the concentration parameter of the Antennae star cluster [W99]1 is indicated by a vertical bar near $t_{\rm df} \simeq 17$\,Myr, the horizontal error bar reflect the uncertainty in the measured cluster mass and radius. Solid circles indicate simulations resulting in runaway merging, while open circles correspond to simulations in which no runaway merging occurred. The evolution of the mass of the runaway for the two leftmost filled circles are presented in Figure \ref{fig:Mbh}, for a variety of initial conditions. Star clusters in the upper left corner enclosed by the dashed line are expected to host an IMBH formed by the runaway growth of a single star. In the other parts of the diagram (lower concentration and/or larger dynamical friction time) an IMBH cannot form by this process.

Figure 8: The 64 Tflops (64 trillion calculations per second) GRAPE-6 at the University of Tokyo. GRAPE, short for GRAvity PipE, is a family of special-purpose hardware specially designed and built by a group of astrophysicists at the University of Tokyo to perform large simulations of star clusters. Like a graphics accelerator speeding up graphics calculations on a workstation, without changing the software running on that workstation, the GRAPE acts as a Newtonian force accelerator, in the form of an attached piece of hardware. In a large-scale gravitational N-body calculation, where N is the number of particles, almost all instructions of the corresponding computer program are thus performed on a standard workstation, while only the gravitational force calculations, in innermost loop, are replaced by a function call to the special-purpose hardware.