aa r X i v : . [ a s t r o - ph ] J a n Conference Summary
Brian W. O’Shea ∗ , Christopher F. McKee † , Alexander Heger ∗∗ and Tom Abel ‡ ∗ Theoretical Astrophysics Group, Los Alamos National Laboratory, Los Alamos, NM 87545; [email protected] † Departments of Physics and Astronomy, University of California, Berkeley, CA 94720;[email protected] ∗∗ Theoretical Astrophysics Group, Los Alamos National Laboratory, Los Alamos, NM 87545; [email protected] ‡ Kavli Institute for Particle Astrophysics and Cosmology, Stanford University, Menlo Park, CA 94025;[email protected])
Abstract.
The understanding of the formation, life, and death of Population III stars, as well as the impact that these objectshad on later generations of structure formation, is one of the foremost issues in modern cosmological research and has beenan active area of research during the past several years. We summarize the results presented at “First Stars III,” a conferencesponsored by Los Alamos National Laboratory, the Kavli Institute for Particle Astrophysics and Cosmology, and the JointInstitute for Nuclear Astrophysics. This conference, the third in a series, took place in July 2007 at the La Fonda Hotel inSanta Fe, New Mexico, U.S.A.
Keywords:
Cosmology; Population III stars; Conference summary
PACS:
1. INTRODUCTION
In July 2007, more than 130 international researchersmet at the La Fonda Hotel in Santa Fe, New Mex-ico, U.S.A. to discuss the formation, life, and death ofzero-metallicity (Population III) and very low metallicitystars, as well as the impact that these objects had on latergenerations of stars and on structure formation. This fieldhas made significant theoretical and observational ad-vances since First Stars I, which was a MPA/ESO/MPEJoint conference held in 1999 in Garching b. München,Germany, and First Stars II, which was held at The Penn-sylvania State University in State College, Pennsylva-nia, U.S.A., in 2003. Though major advances have beenmade, the understanding of Population III and very lowmetallicity stars is still in its infancy, and many importantquestions remain unanswered.Donald Rumsfeld is known for many things, but hiscontribution to the philosophy of science is less wellrecognized. In an oft-cited speech he delivered on Feb12, 2002, he said:There are known knowns. These are thingswe know that we know. There are known un-knowns. That is to say, there are things that weknow we don’t know. But there are also un-known unknowns. There are things we don’tknow we don’t know.He applied this to intelligence data, but one can describeour task as scientists is to uncover “unknown unknowns"by pure thought or by serendipitous observation or exper-iment, thereby converting them to “known unknowns;" and to then use systematic observation, experiment andtheory to convert these to "known knowns."The study of the first stars is a new field. There arealmost no “known knowns," except for the backgroundcosmology and an increasing set of data on abundancesin very metal poor stars. There are many “known un-knowns:" What is the nature of dark matter? What isthe strength of the magnetic field? What are the abun-dance and size distribution of dust? What is the rate ofmixing of metals into primordial gas? Until recently, theeffect of dark matter annihilation on the first stars wasan “unknown unknown," but this has become a “knownunknown" through the work of Freese and her collabora-tors.We conjecture that the number of “unknownunknowns"—i.e., the discovery potential—in a fieldis proportional to the ratio of the number of “knownunknowns" to “known knowns:" UU (cid:181) KU / KK . (1)If so, the study of the First Stars is an excellent field foryoung people—as reflected in the youth of the audienceand the organizers (although not in the summarizer)!In the following sections, we shall summarize the re-sults presented at this conference, and raise some signifi-cant issues that have yet to be explored. In the interests ofconserving space, we discuss only the results presentedduring First Stars III, and direct interested readers to in-dividual contributions for more in-depth information andcitations to refereed papers. In addition, we have made aconcerted effort to refer only to contributions in this pro-eedings, and do so by the last name of one of the authors(typically the first) of each contribution.
2. A PROPOSED NAMINGCONVENTION FOR PRIMORDIALAND METAL-POOR STARS
A significant point of confusion in the literature on Pop-ulation III and low metallicity stars arises from the abun-dance of (often contradictory and/or confusing) namingconventions. Given that essentially all researchers in-volved in the field were present at First Stars III, a discus-sion of this subject took place and the following namingconvention has been proposed:
Population III:
This is a blanket term that describesall stars of primordial composition (i.e., gas whose com-position was determined during Big Bang Nucleosynthe-sis and as a result is composed almost entirely of hydro-gen and helium), regardless of how, when, or where theyformed. It has been found that this term is too broad, andthus the need for the existence of “
Population III.1 ” and“
Population III.2 ” stars, as described below.
Population III.1:
These are the true “first generation”stars of primordial composition, whose properties havebeen determined entirely by cosmological parametersand the process of cosmological structure formation, andhave not been significantly affected by previous star for-mation. Examples of Population III.1 star formation areshown in contributions by Norman, Turk, and Yoshida.
Population III.2:
These are “second generation”stars, which still have primordial composition. Their for-mation, however, has been significantly affected by pre-vious generations of star formation through the injec-tion of kinetic energy, photodissociating or ionizing ra-diation, by cosmic rays, or by other as-yet-unsuggestedprocesses, which may change the mass range of thesestars. Examples of Population III.2 star formation are dis-cussed in contributions by Ahn, Bromm, Bryan, Johnson,McGreer, Sato, Norman, Umemura and Whalen.
Population II.5:
This is the suggested term for a pos-sible class of stars with non-zero metal content where theamount of metals in the gas that the stars are formed fromwould not be sufficient to affect the cooling properties ofthe gas and thus the formation of the stars, but wouldplay a non-negligible role in the star’s main-sequenceevolution. An example of this class of object (as dis-cussed by Meynet) is a rapidly rotating, massive star witha metallicity of ≃ − Z ⊙ , which would have formed ina manner identical to a star of primordial composition,but would experience enhanced mass loss relative to aPop III star of otherwise identical properties due to thesmall amount of metals in the star. Population II:
These are stars whose metal content exceeds the “critical metallicity” discussed by Bromm,Glover, Omukai, Shull, and B. Smith – namely, themetallicity where the enhanced cooling properties ofmetal-enriched gas affects the star formation process,possibly resulting in a different IMF.
3. FORMATION AND IMF OF STARS ATZERO AND LOW METALLICITY
The ultimate goal of the theoretical and numerical workbeing performed by many investigators is to gain a funda-mental understanding of the Initial Mass Function (IMF)of Population III (Pop III) stars. This is also true of re-search relating to star formation in the present-day uni-verse – Population III star formation, however, is consid-ered to be a more tractable problem, given the relativesimplicity of the physics involved.In the absence of observations of Population III stars,we must rely on theory and simulations. As discussed byNorman, both theory and simulations predict that the firststars were massive, with an average mass that is muchgreater than stars in the Milky Way. This is consistentwith the absence of observed zero-metallicity stars at thepresent day. Simulations show that not all Population IIIstars form in exactly the same fashion, however, whichsuggests a broad spread of masses. In addition, there is noevidence for fragmentation in extremely high-resolutionsimulations of Population III star formation, even whentheory suggests that there should be (though note thatUmemura and Suwa displayed preliminary calculationsthat indicate fragmentation in an asymmetric runawaycollapse of primordial gas).Turk and Yoshida presented simulations of Popula-tion III.1 star formation using different methods. Yoshidaused the Gadget-2 smoothed-particle hydrodynamicscode, while Turk used the ENZO adaptive mesh refine-ment code. Both codes have been improved in the pastfew years with the addition of extended-precision arith-metic, particle splitting techniques (for SPH), updatedchemistry, approximations for cooling that take into ac-count the non-negligible optical depths at high densi-ties, and modifications to the ideal gas law equation ofstate at extremely high density. These improvements al-low both codes to simulate large ( ∼ Mpc) volumes ofthe universe while at the same time following the col-lapse of gas to protostellar densities, with a current max-imum density of n H ≃ cm − ! These fundamentallydifferent methods agree quite well to sub-parsec scales– there is no evidence for fragmentation in the collaps-ing primordial cloud cores, and the inferred accretionrates of gas flowing onto the evolving protostellar coreare extremely high, peaking at ˙ M ≃ − − − M ⊙ / yr.At smaller scales, some differences are apparent in theiralculations. It is unclear, however, whether this is dueto numerical issues or simply due to the use of differentcosmological realizations.Glover discussed issues relating to uncertainties in the3-body molecular hydrogen formation reaction rates. Atthe temperatures relevant to Population III star formation( T < − −
400 M ⊙ , withthe main-sequence stellar mass in the fiducial case being ≃
160 M ⊙ .Some aspects of Population III.2 star formation werealso discussed. Bromm (Greif et al.), Bryan, Yoshida,and McGreer presented results showing the formation ofprimordial stars in regions of pre-ionized gas, using twodifferent methods. All agree that significant amounts ofHD form in these regions, and result in a rapid coolingof gas down to the temperature of the CMB at the red-shift of formation. This lower temperature directly trans-lates into a reduced accretion rate onto the protostellarcore, suggesting that the resulting Population III.2 starswill be less massive than their Pop III.1 counterparts. Bromm (Greif et al.) also showed that the presence ofa cosmic ray background can cause the formation of sig-nificant amounts of HD in high-density primordial gas,again lowering gas temperature to that of the CMB andreducing the accretion rate. This is not to say that allPopulation III.2 star-forming regions are characterizedby reduced accretion rates: Norman showed results fromsimulations of Pop III stars forming in the presence of amolecular hydrogen photodissociating (Lyman-Werner)background, which results in higher overall halo temper-atures and higher inferred accretion rates onto the proto-stellar core.The transition between Population III and Popula-tion II star formation was discussed at length duringthe conference. Bromm (Greif et al.) suggested the ex-istence of a “critical metallicity,” or Z c rit , where metalline cooling dominates over molecular hydrogen cool-ing at low temperature (and hence presumably wherethe stellar IMF will transition from being top-heavy toa Salpeter-type function with a significantly lower meanmass), at 10 − < Z crit / Z ⊙ < − . Glover and Shullboth suggest that Z c rit may depend strongly on environ-ment. Omukai showed that dust is an extremely effec-tive coolant compared to gas-phase metals, and that thepresence of a small amount of dust can radically lowerthe critical metallicity. The degree with which this takesplace depends strongly on the abundance, type and sizeof the dust grains, and is highly uncertain. B. Smith usedhighly-resolved adaptive mesh refinement calculations toexamine the fragmentation of metal-enriched gas at smallscales and observed that there is not a clear relationshipbetween metallicity and fragmentation scale – rather, thefragment scale is determined by the density of the gaswhen it reaches the temperature of the CMB. In this way,higher-metallicity gas may reach the CMB temperatureat lower densities than lower-metallicity gas, resultingin larger overall clumps. Jappsen performed SPH sim-ulations of metal-enriched gas collapsing in an ionizedhalo. She showed that, for densities n H < cm − , theevolution of density and temperature are not changed bymetallicity for Z < . ⊙ , because H is the dominantcoolant rather than metal fine structure lines. In addi-tion, she does not find evidence in her calculations forthe “critical metallicity” threshold proposed by Bromm(Greif et al.) Clark, however, used high-density SPH sim-ulations to show that fragmentation can easily occur inthe high-density, dust-dominated regime at metallicitiesat or below Z = − Z ⊙ . . SEARCHES FOR POPULATION IIIAND VERY LOW METALLICITYSTARS AND OBSERVEDABUNDANCE PATTERNS4.1. The search for Population III stars To date, no stars of primordial composition have beendirectly detected, either in our galaxy or in the distantuniverse. This may be because these stars are very mas-sive and thus short-lived, or because surveys are lookingin the wrong places. If these objects still exist, how canthey be detected? And, have we already indirectly de-tected the traces of Population III stars?If, as Tan suggests, Population III.1 stars have massesof on the order of 100 −
400 M ⊙ , some of these starsmay explode as pair instability supernovae (PISN), as de-scribed by Woosley. These objects would have unusualnucleosynthetic patterns, which have not yet been ob-served in abundance ratio measurements of galactic halostars. This lack of evidence sets strong upper limits on onthe number of primordial stars in this mass range.N. Smith presented observations of SN2006gy, a TypeIIn supernova in NGC1260 (a S0/Sa galaxy located ap-proximately 73 Mpc away). This was the brightest super-nova known at the time of the conference, and the lightcurve and inferred expansion velocity are inconsistentwith other Type II supernovae and cannot be explainedby the interaction of a standard Type II supernova with acircumstellar medium. The observed light curve is con-sistent with results presented by Kasen showing theo-retical predictions of pair production supernova modellight curves and spectra, lending hope that Population IIIPISN exist. In principle, a pair instability supernovae at z ∼
20 would be bright enough at peak luminosity to beobservable with JWST, though the time dilation of thelight curve could make detection extremely difficult. Thiswould be much less of a problem if, as suggested bySchneider, a non-negligible fraction of stars forming atz ∼ ⊙ . In this scenario the supernova-like ejecta of subsequent pulses run into each other atlarge distance and convert their kinetic energy at almost100 % efficiency into radiation, making for a very brightdisplay. Generally, the mass range for pulsational pair in-stability in non-rotation primordial composition (Pop III)stars is about 100 −
140 M ⊙ .Examination of cosmic infrared background (CIB)fluctuations may yield indirect detections of Population III stars. Kashlinsky presented results from observationsusing the Spitzer Space Telescope, arguing that source-subtracted IRAC images contain significant CIB fluctu-ations that are in excess of the near-infrared backgroundexpected from resolved galaxy populations. These fluc-tuations appear to come from clustered sources that donot correlate with Hubble ACS source catalog maps ofthe same field. He claimed that this implies that the CIBfluctuations come from dim populations at high ( z > −
10 sources/arcsec (within the confusion limit ofpresent-day instruments, but resolvable by JWST), andclaimed that this source population is high-redshift pri-mordial stars. Thompson, on the other hand, claimed thatthe purported near-infrared background excess is due toimproper modeling and subtraction of zodiacal light, andthat the cosmic infrared background fluctuations detectedby Kashlinsky are mainly due to low-redshift (0 . < z < .
5) galaxies and are inconsistent with galaxies at z > − Metal-poor galactic halo stars are of great interest tothe community that studies structure formation in theearly universe for several reasons. Extremely metal-poorstars are possibly fossil records of the heavy elementabundances produced in a single Population III super-nova. The shape of the low-metallicity tail of the metal-licity distribution function (MDF) has the potential toprobe epochs of star formation in the early galaxy, andthe change of the MDF with distance from the center ofthe galaxy may provide useful clues as to the assemblyhistory of the Milky Way. In addition, the discovery ofobjects with enhancements of combinations of various a -, s-, and r-process elements can probe rare events in theformation history of our galaxy. Ferrara and Salvadori,however, argue that almost no true “second generation”stars (objects that have been enriched by a single Popu-lation III supernova) should be observable at z = < − < −
3. Overhe course of the project lifetime, SEGUE is projectedto find roughly 20,000 stars with [Fe/H] < −
2, and ap-proximately 2000 with [Fe/H] < −
3. The 0Z Project (assummarized by Cohen) describes the discovery of morethan 1500 candidate extremely metal poor stars from theHamburg/ESO Survey, with some stars having extremelypeculiar abundance ratios. The ESO LAMOST Surveyand Southern Sky Survey (described by Christlieb), willbegin production in the near future, will cover signifi-cantly larger areas of the sky than SEGUE (to a similarmagnitude) and will, as a result, find thousands of starswith metallicities of [Fe/H] < − < − − < [Fe/H] < − Much can be inferred from the observed abundancepatterns of extremely metal-poor galactic halo stars,though the concerns discussed in the previous sectionmust be kept in mind when interpreting these results.Frebel presented an update on the abundance of HE1327-2326 using spectra from the VLT, confirming itsextremely low iron abundance and carbon enhancement[FeI/H] = − . − .
5, depending on assumptions, and[FeII/H] < − .
4, and [C/Fe]
LTE = .
28 using a 3D modelatmosphere correction). This work was elaborated uponby Korn, who showed that HE 1327-2326 is a subgiant and not a main-sequence star, and that atomic diffusionmay have significantly altered the surface abundances ofthe star (though not enough to fully explain the puzzlingnon-detection of lithium). Frebel also presented work us-ing observations of carbon and oxygen-enhanced metalpoor stars to examine necessary conditions for forminglow-mass stars in the early universe through cooling viafine-structure lines, and showed evidence for a “transi-tion metallicity” where the sum of the carbon and oxy-gen abundance is approximately 10 − . the sum of thesolar abundances for these elements, which supports theBromm idea of a “critical metallicity.”Johnson argued that carbon-enhanced metal poor(CEMP) stars are created by the pollution of a low-massstar by a companion asymptotic giant branch (AGB)star. She suggested that the observed [C/N] ratios im-ply that a large number of primordial stars with massesof 2 − ⊙ were created. This result was supportedby Schuler, who presented observations of fluorine in aCEMP halo star. AGB stars are believed to be prodigiousproducers of both carbon and fluorine, suggesting thatAGB stars may be the source of the observed abundancepatterns in at least some CEMP stars. Both of these ob-servational results are supplemented by Pols, who pre-sented simulations of binary stellar evolution showingthat the observed CEMP abundance ratios can be ex-plained by binary pollution from AGB stars, and thatthermohaline mixing is important in the metal-poor com-panion star, and by Husti, who presents similar results.This latter effect is examined in more detail by Bisterzo,who presented numerical experiments detailing the ef-fects of thermohaline mixing on nucleosynthetic yieldsresulting from neutron-capture nucleosynthesis. Thesecalculations show that the yields for CEMP stars canbe well-matched when thermohaline mixing is included.Lucatello showed results from the HERES survey sug-gesting that there are different classes of CEMP stars,based on the presence or absence of r- and s-process el-ements. These stars might be enriched by different typesof stars, or via different mechanisms, and the monitoringof stellar radial velocities may give some clues to theirdifferent origins. Rossi showed that estimates of carbonabundances in CEMPs may be affected by calibrationproblems, and presented a set of refined estimates of car-bon abundances for a large sample of CEMPs.de Mink presented results from binary evolution at lowmetallicity. She showed that binaries at low metallicityexperience mass transfer in a very different way thansolar-metallicity stars do, which may have implicationsfor the metallicity dependence of the formation rate ofvarious objects through binary evolution channels. Shealso showed that low-metallicity binaries can experiencemuch higher rates of accretion before reaching a givensize compared to solar-metallicity stars, suggesting thatfewer low-metallicity binaries come into contact (andhus merge) during rapid mass transfer.Masseron presented results from the high-resolutionchemical analysis of a large sample of CEMP stars, andshowed that these stars naturally split into two groups:stars enriched only in s-process elements, and those en-riched by both r- and s-process elements. The formergroup is well explained by AGB mass transfer, andMasseron suggests that the latter type of star can beexplained by a primordial companion with a mass of8 −
10 M ⊙ . Tumlinson argued that the two observed hy-per metal poor stars were most likely formed in mass-transfer binaries with a top-heavy IMF, and that the highfrequency of CEMP stars at [Fe/H] < −
2, and gradi-ents with metallicity and location, suggest that the CMBsets the typical mass of early Pop II stars independent ofmetallicity. Komiya also suggested that the IMF of ex-tremely metal-poor stars is top heavy, and that this is con-sistent with the observed metallicity distribution functionin metal-poor halo stars.Sobeck presented results from a study of copper abun-dances in 50 metal-poor halo stars using the VLT UVESspectrograph. There appears to be a deficit of copperin these stars, implying the reduced extent of weak s-process in massive stars at low [Fe/H] or the delayed pro-duction from Type Ia supernovae. Boesgaard discussedberyllium abundances in a sample of 51 metal-poor halodwarf stars that have been examined at high resolutionand signal-to-noise using the Keck HIRES and SubaruHDS spectrographs. There is no evidence for a beryllium“plateau” – rather, [Be/Fe] remains constant with chang-ing [Fe/H] – and also no evidence for a difference in Bevs. [O/Fe] in stars that are found in the halo versus thoseobserved in the thick galactic disk, contrary to previouslower-resolution work.Cowan presented results from HST and ground-basedstudies of galactic halo stars, and argued that at the timethese stars formed the galaxy was chemically unmixedand inhomogeneous in r-process elements, but not in a -elements, and suggests different environments for thesynthesis of these elements. There is evidence for in-creasing contribution of the s-process with metallicity(and thus galactic age). Krugler presented results fromautoMOOG analysis of over 6500 stars from SDSS-Iand SEGUE that have estimated metallicities of [Fe/H] < − < T eff < Li and Li –with Li being formed during the epoch of Big BangNucleosynthesis (BBN), but not Li. There is an appar-ent discrepancy between the measured amounts of Liand the predicted amounts based on recent determina-tions of the baryon-to-photon ratio, with the observedamount of Li being too low. Asplund pointed out thatthere are actually two cosmological lithium problems –the observed abundance of Li is inconsistent with pre-dictions from standard BBN, and Li is apparently incon-sistent with galactic cosmic ray production. He suggestsa range of possible mechanisms to cause these inconsis-tencies, but cautions that they state inconsistencies arebased on extremely challenging observations and analy-sis, and should be taken with a grain of salt. Perez pre-sented observations of two metal-poor stars suggestingisotopic abundance ratios of Li/ Li ≃ . − .
05. Inone case this is similar to results found in the literaturefor the same star, and in the other case her derived valueis significantly higher. Perez suggested that the derivedabundance ratio is extremely sensitive to parameters usedin the analysis. Sbordone also found this sensitivity, andshowed that effective temperature scales for extremelylow metallicity stars are still poorly calibrated, leading toinaccurate determination of lithium isotopic abundanceratios.
5. STELLAR EVOLUTION,EXPLOSIONS ANDNUCLEOSYNTHESIS AT ZERO ANDVERY LOW METALLICITIES5.1. Stellar Evolution and Explosions atVery Low Metallicities
In his review talk, Woosley made several points. Heclaims that the known abundances in low metallicity starscan be fit by “ordinary” supernovae in the 10 −
100 M ⊙ range, and that there is no need for hypernovae or pairinstability supernovae. He argues that the favored Pop IIIstellar masses based on nucleosynthetic results are 10 −
20 M ⊙ , with explosion energies of roughly 10 ergs, andlittle mixing. He also suggested that metal-deficient starswill produce many more black holes than their metal-rich counterparts, with masses as high as 40 M ⊙ . Fi-nally, he argued that the pulsational pair instability cangive a wide range of light curves, from faintest to thebrightest observed supernovae. Yoon presented some re-sults based on simulations of massive stars at very lowetallicity, including rotation and binary interaction. Heclaims that the effects of rotation are particular impor-tant for the evolution of massive stars at low metallic-ity, and that a large fraction of these stars may produceGRBs or hypernovae with unique nucleosynthetic fea-tures. Additionally, he suggested that a significant frac-tion of massive stars may be Wolf-Rayet stars at verylow metallicity, and that stars of this type in close bina-ries may also produce GRBs. Meynet also spoke aboutthe evolution of Pop III and very metal poor stars, andstated that the effects of rotation on Pop III stars is sig-nificant, but less extreme than in very metal poor stars.He also suggested that a tiny amount of metals (on theorder of Z = − Z ⊙ ) may make a big difference, andthat rotation is a key parameter for very metal poor stars.Ekstrom argued that under certain conditions it may bepossible for very massive primordial stars to avoid pair-instability supernovae with the help of two effects of ro-tation: anisotropic winds and magnetic fields. This willhappen even with the assumption of reasonable initialequatorial velocities. Chiappini presented further resultsregarding the impact of stellar rotation on chemical en-richment at low metallicity, showing that fast stellar ro-tation at low metallicities is the only thing that can ex-plain the observed abundances in metal-poor halo stars inthe absence of AGB binary mass transfer. Lau presentedsimulation results from explosions of intermediate-mass,zero-metallicity stars, and showed that the stellar en-velopes of these stars will be enriched by nitrogen, andthat there is no s-process enrichment, owing to the lackof a third dredge-up. Gil-Pons presented results fromsimulations that examine the effects of overshooting inthe evolution of intermediate-mass stars that agrees wellwith Lao’s result. Brott discussed the efficiency of rota-tional mixing in massive stars, and finds that while inter-nal magnetic fields are necessary to understand angularmomentum transport (and thus rotational behavior), thecorresponding chemical mixing must be neglected in or-der to reproduce observations. She also showed that forlow metallicity stars, detailed initial abundances are ofprimary importance, since solar-scaled abundances mayresult in significant calibration errors. Tsuruta presentedresults from simulations of the evolution of very mas-sive (500 M ⊙ and 1000 M ⊙ ) Population III stars, andfinds that though these stars experience the pair insta-bility, they eventually undergo core collapse and createblack holes that, in the more massive star, may be up to500 M ⊙ .Pols presented theoretical results pertaining to carbon-enhanced metal-poor (CEMP) stars, arguing that thesestars are in binary systems and have been polluted bya former AGB companion, and also suggesting that itis important to study nucleosynthesis together with bi-nary evolution. Suda discussed the stellar evolution oflow- and intermediate-mass extremely metal poor stars, and suggested that these stars may be responsible forCEMP stars. Ludwig discussed hydrodynamical modelatmospheres of metal-poor stars, and showed that metal-poor stellar atmospheres are prone to exhibiting substan-tial deviations from radiative equilibrium. He also sug-gested that large abundance corrections may have to bemade to take into account assumptions made in 1D atmo-sphere calculations. In addition, the three-dimensionaleffects on the effective atmospheric temperature fromBalmer lines show a complex pattern, further compli-cating analysis. van Marle showed results from simula-tions of continuum-driven winds from primordial stars,and argued that continuum driving can produce strongmass loss from stars, even without metals. Krticka pre-sented the possibility that hot Population III stars mayexperience significant mass loss if their atmospheres areenriched by CNO elements through metal line-drivenwinds, with a range of effects depending on the metal-licity of the atmosphere. Muijres presented results fromsimulations exploring the effect of clumping on predic-tions of mass-loss rate of early-type stars, and suggeststhat the difference between theoretically expected andempirically derived mass-loss rates may be due to inho-mogeneities. She predicts that clumping leads to a highermass-loss rate, with only modest clumping factors re-quired to match observed values of mass-loss rates inmassive stars. Onifer showed attempts to calculate themass-loss rate of a Population III Wolf-Rayet star using amodified version of the CAK approximation. He showedthat even a star with zero initial metallicity will experi-ence significant mass loss due to radiation pressure ondredged-up nucleosynthetic products.Church presented multidimensional simulations ofprimordial supernovae, which investigate the effects ofRayleigh-Taylor-induced mixing and asymmetries in theexplosion on the final composition of the escaped gas.She finds that for spherically-symmetric explosions,mixing has little effect on the shells interior to oxygen,and that some asymmetry is needed in the explosion inorder for elements interior to oxygen to escape from thestar. Nozawa showed the evolution of dust grains thatformed in Population III supernovae, and argued thatdust grain size and composition strongly affects theirevolution. He also demonstrated that small grains arepreferentially destroyed in primordial supernova rem-nants, with the maximum destruction mass varying as afunction of the ambient medium properties. Qian presented a review of nucleosynthesis of metal-free and metal-poor stars. In his talk, he made severaloints; he suggested that the origin of Li in low metal-licity stars is still a puzzle; that the presence of nitro-gen in low metallicity stars indicates rotation; and thatthe presence of lead in metal-poor binary members in-dicates that the s-process occurs at low metallicity. Healso argued that the observed abundances in metal-poorgalactic halo stars suggests standard supernovae ratherthan pair-instability supernovae, but that there are con-tributions from both hypernovae and faint supernovae.Nomoto discussed nucleosynthesis in massive Popula-tion III stars, focusing on hypernovae and jet-inducedexplosions. He showed that explosions with large en-ergy deposition rates will create gamma ray bursts withhypernovae and that their yields can explain the abun-dances of normal extremely metal-poor stars, and thatexplosions with small energy deposition rates will be ob-served as GRBs without a bright supernova, and can beresponsible for the formation of CEMP and hyper metalpoor stars. Kratz (Farouqi et al.) presented results explor-ing nucleosynthesis modes in the high-entropy-wind sce-nario of Type II supernovae, and showed that a super-position of several entropy components can reproducethe overall solar system isotopic r-process residuals, aswell as the more recent observations of elemental abun-dances of metal-poor, r-process-rich halo stars. Pignataripresented results of simulations examining the weak s-process at low metallicity, and showed that the s-processefficiency changes significantly as a function of both stel-lar mass and metallicity.Woodward and Herwig (combined proceedings) dis-cussed nucleosynthesis and mixing in the first genera-tion of low- and intermediate-mass stars, with a focuson studying entrainment at convective boundaries. Theypoint out that one feature of stars at zero and low metal-licity is that convective-reactive mixing events are com-mon, which is not true in stars of higher metallicity. Theyshow that 1D models have difficulty simulating theseevolutionary phases and the mixing that comes fromthem, and show 2D and 3D simulations of this process.Cristallo presented work on the evolution and nucleosyn-thesis of low-mass metal-poor AGB models with carbon-and nitrogen-enhanced opacities (to address observedcarbon and nitrogen-rich metal poor stars), and showsthat the new opacities can cause significant changes inthe chemical and physical evolution of the stars, andmay cause non-negligible changes in the amount of car-bon, nitrogen, and s-process elements created. Camp-bell discussed the structural and nucleosynthetic evolu-tion of metal-poor and metal-free low and intermediate-mass stars from ZAMS to the end of the AGB phase, andshowed that many of these stars experience violent evo-lutionary episodes that are not seen at higher metallici-ties. These episodes may be coupled with strong mixing,causing surface pollution. Surman presented simulationsresults of nucleosynthesis in outflows from Kerr black hole accretion disks. She suggests that these accretiondisks may be important contributors to the nuclear abun-dances in the oldest stars, particularly for rare species orthose not uniformly observed.
6. EARLY FEEDBACK PROCESSES:INFLUENCE ON STRUCTUREFORMATION, REIONIZATION, ANDIGM ABUNDANCE PATTERNS
The first generations of stars are potentially prodigioussources of mechanical, chemical, and radiative feedback,and may contribute significantly to the enrichment andreionization of the intergalactic medium. Reed showedthat high-redshift halos are strongly biased, particularlyat small scales – this effect differs from low-redshiftbias, and is a function of mass. The clustering observedwill significantly affect the statistical properties of feed-back at high redshifts. Ciardi argued that feedback fromPopulation III stars is generally not as efficient as onemight expect from energetics estimates, and that objectswith masses of M > − M ⊙ are not greatly affectedby feedback. Wise presented AMR simulations model-ing the formation of a significant number of Pop III starsin a single volume, and showed that dynamical feedbackfrom HII regions and supernovae can expel most of thegas in first-generation halos and lead to low baryon frac-tions in subsequent star-forming halos. In addition, heshowed that metals are well-mixed within dwarf galax-ies, and that ejecta from Pop III supernovae will providea maximum metallicity of ∼ − Z ⊙ . Greif presented anSPH calculation of a Pop III pair-instability supernovaexpanding into a HII region formed by the progenitorstar, and found similar results – the supernova completelydisrupts the host halo and expands out to a significantfraction of the size of the HII region, ultimately polluting ≃ . × M ⊙ of gas with metals. Nagakura presented1D, spherically-symmetric simulations of the evolutionof supernova remnants in the early universe, paying par-ticular attention to the thermal and chemical evolution ofthe expanding dense shell of gas. He showed that at highredshift, regardless of metallicity, the minimum temper-ature of dense gas in the supernova remnant is limitedby the CMB temperature, suggesting that fragmentationof the shell depends more critically on the density ofthe ambient medium and the supernova energy than onthe metallicity of the gas. Nozawa discussed dust evo-lution in Pop III supernova remnants, and showed thatthe transport of dust within the evolving SNR dependsstrongly on the size and composition of the dust grains,and that small dust grains are preferentially destroyed inthe remnant evolution. Additionally, dust can be stronglysegregated from metal-rich gas, possibly explaining thebundance patterns of iron, magnesium and silicon in thelowest metallicity stars (assuming they were formed inthe shells of Pop III supernova remnants).Whalen presented 2D radiation hydrodynamical sim-ulations of the photoevaporation of minihalos by neigh-boring Population III stars, and demonstrated that, whenappropriate physics and coordinate geometries are used,feedback from the I-fronts of neighboring halos willlargely be positive or neutral, and that the impinging radi-ation drives primordial chemistry that is key to the hydro-dynamics of the halo, making multifrequency radiationtransport necessary. Umemura and Sato confirmed thisresult with 3D SPH simulations including multigroupradiation transport, and additionally suggested that theshielding due to H created in the ionization front mayallow the formation of multiple stars in a single halo.In contrast, Ahn presented work using a 1D Lagrangianradiation transport code that suggests that the result ofan impinging I-front will be roughly neutral. Bryan andMcGreer showed that HD cooling can play an importantrole in low-mass non-pre-ionized halos, reducing accre-tion rates onto the protostellar cloud core. Johnson (Greifet al.) showed that primordial stars forming in relic HIIregions would likely be protected from radiative feed-back from neighboring halos due to the large amounts ofmolecular hydrogen that form even at low densities dueto the high residual electron fraction.Schneider presented a set of cosmological simulationsof 5 −
10 Mpc boxes that include a metallicity-dependentstar formation and feedback algorithm algorithm. Thesesimulations indicate that, due to inefficient metal enrich-ment, Population III star formation may continue up to z ∼ . z ∼
20 by Population III stars and was90% complete by z =
7, and that quasars dominate only reionization at z <
6. Additionally, the model predictsthat more than 80 % of ionizing photons responsible forreionization were produced at z ≥ M < M ⊙ , implying that the bulk of the reionizationsources at high z have not yet been observed. This modelis consistent with the high-redshift QSO absorption linemeasurements presented by Fan and with several othersets of observations, including the observed number ofLyman-alpha emitters and damped Lyman-alpha sys-tems. Trac and Shin presented results from large vol-ume, high-resolution dark matter-only simulations thathave been post-processed to model the effects of radia-tive feedback from Population III and metal-enrichedstars, and found results that are also in good agreementwith Fan’s observations of the Gunn-Peterson troughin high-redshift quasar absorption lines, and are alsoconsistent with the most recent WMAP electron opti-cal depth measurements. Iocco presented results showingthat Pop III stars will produce a diffuse high-energy neu-trino background – the strength of this background, how-ever, will fall below detectable thresholds for all currentand planned neutrino telescopes even with highly opti-mistic assumptions, ruling out the neutrino backgroundas a useful diagnostic tool for metal-free stars. This wasconfirmed by Suwa, who also estimated gravitationalwave emission from the collapse of isolated PopulationIII stars, and suggested that the gravitational wave back-ground from these objects could be detectable by futuregravitational wave interferometers such as BBO or DE-CIGO.
7. COMPACT OBJECTS AT HIGHREDSHIFT7.1. Gamma-Ray Bursts and Quasars
Rockefeller showed three-dimensional simulations ofthe collapse of a massive Population III star, demonstrat-ing that it is possible to create a rapidly-accreting disk ofgas around the central black hole, which can then ejectsignificant amounts of the stellar envelope. Li used thesesimulations as initial conditions for a three-dimensionalMHD calculation of the collapsing stellar core, show-ing that the star’s magnetic energy can drive an outgo-ing shock, which leads to an explosion. In addition, themagnetic energy is primarily in “bubbles” of gas that aredominated by magnetic energy, and tend to evolve in avery inhomogeneous (and collimated) fashion. This hasprofound implications for Population III stars as gammaray burst progenitors, and also for nucleosynthesis.High-redshift gamma ray bursts have proven to be ex-tremely useful tools for probing the ISM and IGM. Chendiscussed using spectroscopy of gamma-ray burst after-lows to perform direct, detailed studies of the ISM inthe star-forming regions of distant galaxies, which is im-possible with quasars. A large sample of GRB sightlines,coupled with moderate resolution afterglow spectra andimaging of the host galaxies, will be key tools for under-standing the properties of the ISM in high-redshift galax-ies. Penprase showed more results related to GRB after-glows, including highly detailed estimates of the proper-ties of star-forming regions in damped Lyman-alpha sys-tems, and demonstrated the promise of GRBs as probesof star-forming regions in high-redshift galaxies. Yone-toku showed a strong correlation between the spectralpeak energy of prompt GRBs and the peak luminosity,and used this to estimate the possible redshifts for hun-dreds of BATSE GRBs of unknown redshift. This anal-ysis predicts that a large number of massive stars wereformed in the early universe, assuming Pop III stars formGRBs.Quasar absorption line studies are complementarytools to the examination of GRB afterglows, in that theyare very useful probes of the IGM at high redshift. Atpresent, measuring the Gunn-Peterson trough in QSO ab-sorption line spectra is one of the primary methods ofconstraining the end of the epoch of reionization. Thiswas shown by Fan, who demonstrated that the evolutionof the ionization state of the IGM experienced a veryrapid change at z ∼
6, increasing by an order of mag-nitude in D z ≃ .
5. This implies that z ∼ z ∼ z ≥ Can Population III stars be supermassive black holeprogenitors? There are many examples of z ∼ M BH ≥ M ⊙ sitting at the centers of massive galaxies. Trentiargued that these are extremely rare objects (one per ≃ . at z = z ∼ −
50) black hole seeds can be QSO pro-genitors. Alvarez, on the other hand, argued that high-redshift black holes form in halos where the progenitorstar has expelled the majority of the surrounding gas, ini-tially suppressing accretion. After gas collects in the haloagain, there will be strong self-regulation of accretion byradiative feedback. As a result, black holes formed from Pop III stars will not accrete rapidly enough to be SMBHprogenitors (this also has implications for the X-ray ion-ization scenario discussed by Ricotti)Begelman suggested that it may be possible to formextremely massive “quasistars” by the direct collapse ofgas in cosmological halos, without a stellar precursor.The required precursor object could be generated by theformation of halos with T vir > K, or possibly in theaftermath of a large halo merger. These objects wouldbe radiation-dominated, and would feature the formationof a central black hole, which then becomes a sourceof energy that would create a convective envelope. Thisobject could cool via thermal neutrinos, allowing it toaccrete at the Eddington limit of the envelope, rather thanthe black hole itself. This “quasistar” would have a loweffective surface temperature ( T eff ∼ years, and could potentiallybe observed by JWST.Colgate argued that supermassive black holes formnaturally as a result of cosmological structure forma-tion – high-entropy gas creates ∼ M ⊙ stars, whichcollapse due to relativistic instabilities and become thecentral point of a massive accretion disk, which canthen channel gas into the black hole at super-Eddingtonrates, using a combination of self-gravity instabilities andRossby vortices to efficiently transport angular momen-tum outward.
8. SUMMARY
The properties of the first generations of stars, as wellas their effects on later epochs of cosmological struc-ture formation, have long been a topic of great interest tothe astrophysical community. Progress in understandingthese objects has been significant and rapidly accelerat-ing in the past few years, and many insights have beenobtained by theory and by large-scale numerical compu-tations. Progress has also been made observationally, ina wide variety of settings, and many significant questionshave been raised. The impressive array of new facilitiesand surveys that are currently running or are planned forthe next few years, together with increasingly powerfulsimulations, should yield crucial new insights into theearliest epochs of star formation in the universe.