2009 Bromm et al., The formation of the first stars and galaxies (1119309), страница 2
Текст из файла (страница 2)
In numerical simulations, the final mass of a population III star is usually estimated from thedensity distribution and velocity field of the surrounding gas when thefirst protostellar fragment forms, but this may well be inaccurate even inthe absence of protostellar feedback. Whereas protostellar feedbackeffects are well studied in the context of the formation of contemporarystars24, they differ in several important respects in primordial stars25.First, primordial gas does not contain dust grains. As a result,radiative forces on the gas are much weaker.
Second, it is generallyassumed that magnetic fields are not important in primordial gasbecause, unless exotic mechanisms are invoked, the amplitudes ofmagnetic fields generated in the early Universe are so small that theynever become dynamically significant in primordial star-forminggas26.
Magnetic fields have at least two important effects in contemporary star formation: they reduce the angular momentum of the gasout of which stars form, and they drive powerful outflows that dispersea significant fraction of the parent cloud. It is likely that the pre-stellargas has more angular momentum in the primordial case, and this isborne out by cosmological simulations. Third, primordial stars area Cosmological halo300 pcd New-born protostar25 R .b Star-forming cloud5 pcc Fully molecular part10 AUFigure 1 | Projected gas distribution around a primordial protostar. Shownis the gas density (colour-coded so that red denotes highest density) of asingle object on different spatial scales.
a, The large-scale gas distributionaround the cosmological minihalo; b, a self-gravitating, star-forming cloud;c, the central part of the fully molecular core; and d, the final protostar.Reproduced by permission of the AAAS (from ref. 20).much hotter than contemporary stars of the same mass, resulting insignificantly greater ionizing luminosities27.State-of-the-art numerical simulations of the formation of the first(population III.1) stars represent a computational tour de force, inwhich the collapse is followed from cosmological (comoving megaparsec) scales down to protostellar (sub-astronomical-unit) scales,revealing the entire formation process of a protostar. However, furthergrowth of the protostar cannot be followed accurately without implementing additional radiative physics.
For now, inferring the subsequent evolution of the protostar requires approximate analyticcalculations. By generalizing a theory for contemporary massive-starformation28, it is possible to approximately reproduce the initial conditions found in the simulations and to then predict the growth of theaccretion disk around the star29. Several feedback effects determine thefinal mass of a first star25: photodissociation of H2 in the accreting gasreduces the cooling rate, but does not stop accretion.
Lyman-a radiation pressure can reverse the infall in the polar regions when theprotostar grows to 20–30 M8 , but cannot significantly reduce theaccretion rate. The expansion of the H II region produced by the largeflux of ionizing radiation can significantly reduce the accretion ratewhen the protostar reaches 50–100 M8 , but accretion can continue inthe equatorial plane. Finally, photoevaporation-driven mass loss fromthe disk30 stops the accretion and fixes the mass of the star (see Fig.
2).The final mass depends on the entropy and angular momentum of thepre-stellar gas; for reasonable conditions, the mass spans 60–300 M8 .A variety of physical processes can affect and possibly substantiallyalter the picture outlined above. Magnetic fields generated through themagneto-rotational instability may become important in the protostellar disk31, although their strength is uncertain, and may play animportant role in the accretion phase18. Cosmic rays and otherexternal ionization sources, if they existed in the early Universe, couldsignificantly affect the evolution of primordial gas32. A partiallyionized gas cools more efficiently because the abundant electronspromote H2 formation.
Such a gas cools to slightly lower temperaturesthan a neutral gas can, accentuating the fractionation of D into HD sothat cooling by HD molecules becomes important33–36.50©2009 Macmillan Publishers Limited. All rights reservedREVIEWSNATUREjVol 459j7 May 2009significantly impede the build-up of the high dark-matter densitiesrequired to power the stellar luminosity via dark-matter annihilation.Nevertheless, if neutralinos are detected in the appropriate massrange40, early star formation models may need to include the effectof dark-matter annihilation.10–2.feebFeedback from the first starsackfee10–3.ckbam (M. yr–1)NothWim*.mevap10–410100m* (M.)1,000Figure 2 | Feedback-limited accretion.
Change in mass (m_ ) versusprotostellar mass (m*) for a number of key processes. The protostellar_ ) is shown in the cases of ‘no feedback’ (black dotted line)accretion rate (mand ‘with feedback’ (blue dashed line). Even as an H II region is built up,accretion continues through an accretion disk, which is eventually destroyed_ evap ; bluevia photoevaporation. Also shown is the corresponding rate (msolid line). The intersection of the blue dashed and solid curves determinesthe final population III mass. Reproduced by permission of the AAS (fromref. 25).More significant modifications to the standard model result if theproperties of the dark matter are different from those assumed above(see Fig. 3). A key assumption in the standard model is that the darkmatter interacts with the baryons only via gravity.
However, darkmatter can indirectly affect the dynamics of a pre-stellar gas. A popularcandidate for CDM is the neutralino, for which the self-annihilationcross-section is large. Neutralino dark matter is thus expected to pairannihilate in very dense regions, producing high-energy particles suchas pions and electron–positron pairs and high-energy photons. Theseannihilation products may effectively heat collapsing primordial gasclouds when the density is sufficiently high, thereby arresting thecollapse37.
Calculation of the structure of stars with dark-matterannihilation suggest that they can undergo a phase of evolution inwhich they have temperatures of 4,000–10,000 K, well below those forconventional population III stars38,39. The magnitude of this effectdepends sensitively on details such as the dark-matter concentrationand the final products of neutralino annihilation. Furthermore, calculations to date have assumed spherical symmetry, whereas it ispossible that the angular momentum of both the baryons (which leadsto the formation of an accretion disk29) and of the dark matter couldaSome of the feedback processes described above that affect the formation of individual stars also influence primordial star formation onlarge scales.
The enormous fluxes of ionizing radiation and H2dissociating Lyman–Werner radiation emitted by massive population III stars27,41 dramatically influence their surroundings, heatingand ionizing the gas within a few kiloparsecs of the progenitor anddestroying the H2 within a somewhat larger region17,33,42–44.Moreover, the Lyman–Werner radiation emitted by the first starscould propagate across cosmological distances, allowing the buildup of a pervasive Lyman–Werner background radiation field45,46. Theeffect of radiation from the first stars on their local surroundings hasimportant implications for the numbers and types of population IIIstars that form. The photoheating of gas in the minihaloes hostingpopulation III.1 stars drives strong outflows, lowering the density ofthe gas in the minihaloes and delaying subsequent star formation byup to 100 Myr (ref.
47). Furthermore, neighbouring minihaloes maybe photoevaporated, delaying star formation in such systems aswell48–50. The photodissociation of molecules by Lyman–Wernerphotons emitted from local star-forming regions will, in general,act to delay star formation by destroying the main coolants that allowthe gas to collapse and form stars51.The photoionization of primordial gas can, however, also stimulatestar formation by fostering the production of abundant moleculeswithin the relic H II regions surrounding the remnants of populationIII.1 stars44,47,52,53 (see Fig.
4). It is still debated whether this radiativefeedback is positive or negative in terms of its overall impact on thecosmic star formation rate54. However, some robust conclusions haveemerged from the recent simulations. First, the Lyman–Werner feedback is much less ‘suicidal’ than was originally thought55. It is nowbelieved that star formation in neighbouring minihaloes is not completely suppressed, but merely delayed. Second, the ionizing radiationfrom the first stars is initially very disruptive because it substantiallydecreases the density in the host minihalo. This effect leads to thesubstantial gap between the formation of the first and second generations of stars.
In each region of space, the drama of ‘first light’ thusoccurred in two clearly separated stages.Most of the work on the evolution of population III stars and on thesupernovae they produce has been based on the assumption that thestars are not rotating56. For initial stellar masses in the range 25M8 =M = 140M8 and M > 260M8 , population III stars end their lives bycollapsing into black holes with relatively little ejection of heavybFigure 3 | Dark-matter properties and early starformation.
Projected gas distribution in CDM(a) and warm dark matter (WDM; b) simulationsat z 5 20. If the power in the primordial densityspectrum is reduced on small scales, the first starswill form much later than in the standard CDMbased scenario. If the dark matter is warm, havinga substantial velocity dispersion, densityperturbations on small length scales aresmoothed. The hierarchy of structure formationis then truncated at a corresponding mass scale,and the first cosmological objects could be moremassive than 106M8 .
For the case of lightWDM98, gas collapses into filaments, whichmight then fragment into multiple stellar cores.The abundance of star-forming haloes issignificantly reduced in this model. Reproducedby permission of the AAS (from ref. 99).51©2009 Macmillan Publishers Limited. All rights reservedREVIEWSNATUREjVol 459j7 May 2009realistic three-dimensional simulations that start from cosmologicalinitial conditions, and that resolve the detailed physics of the supernova blast wave expansion63,64.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
