The story of how this probably happened, which astrophysicists have been trying to figure out for decades, is rather interesting. Only with results announced at the end of July has the story begun to come into good focus.
The main reason is has taken so long to understand how the first stars formed is that it is quite impossible to see individual stars from the earliest era. Indeed, some of the earliest galaxies we can see (consisting of billions of stars), even with our best telescopes, are about 13 billion light-years away, as they looked about 700 million years after the big bang. This corresponds to a redshift of about 7.5. (See here.)
It follows that the first stars had to have formed some time before that, but as of now we have no way to observationally verify an approximate date. Since it takes time for a galaxy to form out of individual stars, the first star probably formed within the first 500 million years or so after the big bang.
The only way, currently, we can even guess when the first star formed is by starting from what we know – the laws of physics and information we have about the composition of the universe in that time period – in order to do computer calculations (simulations) of the process that should have led to formation of the first stars. Results from the best simulation yet performed have recently been announced.
Although we cannot (yet) directly observe conditions or objects existing within the time period in question, we can infer a variety of facts about them. Some of the direct data we have is based on observations of the cosmic microwave background (CMB). This is radiation that is now observed in the microwave part of the spectrum, although it was much more energetic when it originated approximately 380,000 years after the big bang. Additional data came from observations by the Spitzer Space Telescope, announced in 2005, involving diffuse infrared light that began as ultraviolet light emitted by the first stars. (See here.)
What we know of that period is encompassed in what is called the cold dark matter model (CDM) of the universe. Very good evidence from a variety of sources exists for the overall parameters of this model. The parameters include an overall density of matter (both ordinary and dark matter) that is – at the present time – 30% of the total energy density (with the balance being dark energy). Of that 30%, 26% is dark matter and the remaining 4% is ordinary baryonic matter.
These fractions vary with time, because the density of matter is always decreasing as the universe expands. However, since dark energy (in the form of a cosmological constant) is proportional to volume, the amount of dark energy is always increasing, while its density (per unit volume) remains constant. What this means is that in the early universe during the time we're concerned with, the energy density due to matter was a much larger percentage. However, the ratio of dark matter to baryonic matter remained constant, at 6.5 to 1.
In the big bang model, the earliest chemical elements formed, just a few minutes after the big bang, were hydrogen, helium, and a little bit of lithium. (See here.) By mass, about 75% of this matter was hydrogen, and most of the rest was helium. Since these elements are stable, these proportions did not change for hundreds of millions of years – until the first stars formed.
Another thing we know from the CMB is that there were slight variations from place to place in the average density of matter. Over time, the regions which were slightly more dense than average tended to contract under the force of gravity, and these regions continued to grow denser, relative to everything else.
Eventually there were distinct, though rather diffuse, clouds consisting of dark matter, hydrogen atoms, hydrogen molecules (H2), and a little helium. The rate of collapse at this point is very much driven by the dark matter, since there's 6 times as much of it as of ordinary matter. In these low-density clouds, the pressure due to kinetic energy of gas particles was low compared to the force of gravitation.
You may be wondering why star formation at this early time is such a mystery. After all, stars are forming all the time in the present day. The process is more complex than might at first be supposed, but we have reasonable, albeit incomplete, models of how it happens, and there isn't any great mystery. We can, for example, predict that unless a gas cloud is sufficiently massive, it won't collapse to form a star at all. That is, the gas cloud will never become hot enough and dense enough for thermonuclear reactions to start, so that there is a sustainable source of energy (other than gravitational) to enable the star to shine. Instead, what you get from a cloud that's too small is a brown dwarf, essentially just a ball of gas where there is equilibrium between gravitational force and gas pressure.
But what stellar models show is that even if you start with a sufficiently large cloud of gas, in order that it can collapse far enough to begin thermonuclear reactions it is necessary, paradoxically, that at some point along the way the cloud can dispose of some of its internal kinetic energy. Unless this happens, the cloud has too much internal energy, so its pressure is too high, and equilibrium is reached before the cloud is dense enough to go thermonuclear.
The models further show that the factor which allows energy to be radiated away at the right time is the presence of enough heavy elements. But the kicker is that there were no heavy elements in the early universe – only hydrogen and helium. All other elements up to iron in atomic weight were formed in the first stars from internal thermonuclear reactions. And these elements were only distributed into the interstellar medium when stars of the first generation that were sufficiently large exploded as supernovae, and in the process created all other, heavier chemical elements as well.
But what hasn't been clear, until now, is whether stars could form at all without elements heavier than helium. Perhaps the most that could happen, unless individual clouds were extremely massive, is that contraction would stall, as it does in brown dwarfs. On the other hand, if a gas cloud is too massive, it might be unstable and explode before entering a star-like state that is stable for some significant length of time. In the present universe, the largest known stars have masses around 100 times the mass of our sun, and such stars live only a million years or so before going supernova.
Fortunately, the new simulations now show how stars could form from sufficiently large clouds, even in the absence of heavy elements.
The set of simulations reported on here starts with conditions as they were about 300 million years after the big bang (corresponding to a redshift of 14). One example starts with a gravitationally bound gas cloud of 500,000 solar masses (M⊙), mostly dark matter. This cloud had a temperature of 1000 K, hydrogen and helium atoms, and a small fraction of molecular hydrogen, which enabled efficient radiative cooling to begin with.
The simulation proceeded through a range of 20 orders of magnitude in density, covering about 100,000 years. In the process, the gas became mostly opaque to radiation, so radiative cooling ceased. This means that from then on, the process was "adiabatic", unable to dissipate internal kinetic energy, so that temperature rose quickly. At a certain point in the simulation, a flattened disk-like structure of .1 M⊙ formed. Because the disk was thin, radiation could escape in a perpendicular direction, allowing further cooling. The final outcome, after several other stages, was a .01 M⊙ protostar – defined as a pressure-supported, constant-density atomic gas core.
The temperature of this protostar was 10,000 K, far short of what is needed for thermonuclear reactions. And the protostar was not especially dense – about the same as ordinary water. At this point, however, the simulation exhibited strong shock waves in the hot gas. The simulation stopped here because of the complexity of the protostar. So there is definitely further work to be done. The simulation did not reach the point where thermonuclear reactions would start, but it's a big step anyway, roughly halfway to the final goal.
At the point where the simulation ended, gas was accreting from the surrounding cloud rapidly enough to allow growth to 10 M⊙ in just 1000 years. This could continue to 100 M⊙ or more, which is the expected size of the largest initial stars. However, growth might stop short of that figure, if radiation pressure from thermonuclear reactions rises too fast. On the other hand, if the star grows to much more than 100 M⊙, it could collapse into a black hole, taking the heavy elements with it. Only further simulations can clarify what might happen.
Several lines of evidence show that extremely massive (~100 M⊙) stars existed in the first generation. For instance, there were stars large enough and hot enough to emit photons with enough energy to ionize hydrogen atoms. We know that before stars existed, all hydrogen must have been in the form of an unionized gas – yet before a billion years after the big bang, most of the hydrogen was ionized again. In addition, studies of the CMB indicate a large contribution of light from very bright stars and galaxies in that early time.
There are several other important results from these simulations. One is that it is actually much easier to simulate in detail the formation of the earliest stars than of later generations. This is because in the present universe there are a number of complicating factors, such as relatively abundant heavy elements, strong magnetic fields, and significant turbulence, that raise large obstacles to simulation. Being able to simulate star formation under simpler conditions is an important step to making good simulations under present conditions.
Another valuable result of full simulation of the earliest stars is the ability to predict what galaxies composed of such stars will look like (in terms of color, size, and luminosity) when we are eventually able to detect them with the upcoming James Webb Space Telescope after its projected launch in 2013. Having the predictions available beforehand will help increase confidence in the validity of the whole model.
Protostar Formation in the Early Universe – research article published 8/1/08 in Science
The Cosmic Rosetta Stone – commentary on the research, published 8/1/08 in Science
New simulation accurately tracks seeds of first stars – 7/31/08 news article in Science News
Filling the Gap in Stellar History – 7/31/08 news article in ScienceNOW
Universe's first stars bulk up in new simulation – 7/31/08 New Scientist news article
The first stars – 7/31/08 press release
Additional news reports:
- New Simulation Shows How Seeds of First Stars Formed – 8/1/08 Scientific American news article
- Simulation points to lightweight first stars – 7/31/08 Physics World news article
- How The First Stars In The Universe Came Into Existence – 7/31/08 press release
- Big Bang Ripples Formed Universe's First Stars – 7/31/08 National Geographic news article
- How the First Stars Were Born – 7/31/08 Space.com news article
- Universe's first star born tiny, grew huge: study – 7/31/08 Reuters news article
- Researchers may have found cosmic Rosetta stone – 7/31/08 Associated Press news article
- Simulating the universe's first stars – 8/6/08 blog post
|N. Yoshida, K. Omukai, L. Hernquist (2008). Protostar Formation in the Early Universe Science, 321 (5889), 669-671 DOI: 10.1126/science.1160259|
Tags: star formation
Links to this post: