New cosmological simulations performed on a supercomputer have provided
astrophysicists with the best indication to date of how the first star in
the universe formed. The simulations, detailed in a paper in the November
16 issue of Science, suggest that the first star resulted from the
gravitational collapse of a cloud of hydrogen and helium some 100 times
more massive than the sun.

“Our modeling suggests that the first star may have condensed from a
protogalactic cloud into a self-gravitating, fully molecular mass of
hydrogen and helium at least 100 times that of our sun,” says Michael L.
Norman, a professor of physics at the University of California, San Diego
and a senior fellow at UCSD’s San Diego Supercomputer Center. “We think
that such objects, one per protogalactic mass, may have become the first
stars to shine in the universe.”

In addition to Norman, who is also an astrophysicist at UCSD’s Center for
Astrophysics and Space Sciences, the other authors of the paper include
Tom Abel of the Harvard-Smithsonian Center for Astrophysics, now working
at England’s Cambridge Institute of Astronomy, and Greg L. Bryan of the
Massachusetts Institute of Technology.

One important consequence of star formation are the chemical elements
produced in stars. Elements heavier than lithium, which astronomers
call metals, that occur naturally throughout universe, are the result
of nucleosynthesis, the process by which stars forge heavier elements
from lighter ones within the nuclear fires of their cores.

Astrophysicists believe that it has taken many generations of stars,
each processing the debris left by earlier ones and then distributing
them through massive star explosions, or supernovae, to produce the
elemental abundances revealed by spectroscopic observations of the
stars, gas and dust found in the universe. Elements heavier than
hydrogen and helium are found even in stars so far away that the epoch
of their formation corresponds to a time when the universe was only
about 10 percent of its current age.

“Thus the first heavy elements had to be not only synthesized, but
also released and distributed through the intergalactic medium, within
the first billion or so years after the Big Bang,” says Norman. Only
supernovae of sufficiently short-lived massive stars can provide such
an enrichment mechanism.

While astronomers agree that the first generation of cosmic structures
formed massive stars, there has been no general agreement on the nature
of the first large-scale structures. Globular clusters, super-massive
black holes and Jupiter-size bodies all have been proposed. Yet the
evolution of the large-scale structure of the universe depends very
much on the details of the very first structures to form.

Norman and his former students have worked for a number of years to
increase the predictive ability of their model of the early universe,
“basically waiting for the size of the supercomputers to catch up to
the spatial dynamic range that we need,” he explains.

The group’s latest calculations extend its previous calculations by
some five orders of magnitude. “We follow length scales from a few
kiloparsecs down to 100 solar radii,” Abel says, “while calculating
gravity, hydrodynamics, primordial gas chemistry, and radiative
processes accurately.”

“The formation of the first star takes place in a simpler environment
than any other: the gas is just hydrogen and helium and the initial
conditions can be precisely specified,” Norman says.

What was the first star like? “The picture we get from our simulations
suggests that all metal-free stars are very massive and form in
isolation,” Norman says. While their supernovae occurred so long ago
that we have seen no remnants, measurements of stars in some galactic
halos suggest that these stars were enriched by a single population
of massive stars, which would support the picture derived from the
calculations. With a few more orders of magnitude in dynamic range,
it will be possible to close in on the actual mass and fate of the
very first stars.

The results reported in Science were carried out on a 16-processor SGI
Origin 2000 supercomputer at the National Center for Supercomputing
Applications (NCSA) in Illinois; current calculations are running on
the 1100-processor IBM Blue Horizon machine at UCSD’s San Diego
Supercomputer Center.

Norman is eager to extend the calculations to a new, multi-center,
networked TeraGrid of computers at NCSA, SDSC, Caltech and Argonne
National Laboratory. The $53-million configuration, financed by the
National Science Foundation, will have a peak speed of about 14
trillion calculations per second. It will allow Norman and his group
to extend the dynamic range of its calculations on both the high end
(to start with a larger portion of the universe) and the low (to
follow the condensing protostellar cloud to the point of ignition).

Norman compares the range of the computer code written by the team of
astrophysicists to perform their new calculations to the dynamic range
of a great orchestra. “Bruno Walter was famous for being able to
deliver, with equal clarity, the tinkling of a triangle or the roaring
of a cannon of kettledrums and brass,” he says. “We hope to be thought
of as the ‘Bruno Walters’ of astrophysics.”

The Very First Protostellar Object A slice through the protostellar
object determined by Enzo in temperature (above) and density (below)
gradients. Credit: Tom Abel