Explaining the Sr and Ba Scatter in Extremely Metal-Poor Stars
Wako Aoki, Takuma Suda, Richard Boyd, Toshitaka Kajino, Michael Famiano
aa r X i v : . [ a s t r o - ph . S R ] F e b Explaining the Sr and Ba Scatter in Extremely Metal-Poor Stars
W. AokiNational Astronomical Observatory of Japan, 2-21-1 Mitaka, Tokyo, 181-8588, Japan [email protected]
T. SudaNational Astronomical Observatory of Japan, 2-21-1 Mitaka, Tokyo, 181-8588, Japan [email protected]
R.N. BoydNational Astronomical Observatory of Japan, 2-21-1 Mitaka, Tokyo, 181-8588, Japan [email protected]
T. Kajino National Astronomical Observatory of Japan, 2-21-1 Mitaka, Tokyo, 181-8588, Japan [email protected] andM.A. Famiano National Astronomical Observatory of Japan, 2-21-1 Mitaka, Tokyo, 181-8588, Japan [email protected]
Received ; accepted Dept. of Astronomy, Graduate School of Science; Univ. of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Dept. of Physics and Joint Institute for Nuclear Astrophysics, Western Michigan Uni-versity, 1903 W. Michigan Avenue, Kalamazoo, MI 49008-5252, USA 2 –
ABSTRACT
Compilations of abundances of Strontium and Barium in extremely metal-poor stars show that an apparent cutoff is observed for [Sr/Ba] at [Fe/H] < -3.6and large fluctuations for [Fe/H] > -3.6 with a clear upper bound depending onmetallicity. We study the factors that place upper limits on the logarithmic ratio[Sr/Ba]. A model is developed in which the collapses of type II supernovae arefound to reproduce many of the features seen in the data. This model is consistentwith galactic chemical evolution constraints of light-element enrichment in metal-poor stars. Effects of turbulence in an explosive site have also been simulated,and are found to be important in explaining the large scatter observed in the[Sr/Ba] data. Subject headings: stars: Population II — nuclear reactions, nucleosynthesis,abundances — black hole physics — Galaxy: evolution — equation of state
1. Introduction
Nuclei heavier than iron and nickel are mostly made either in the s-process (slowneutron-capture process) or r-process (rapid neutron-capture process) (B2FH 1957;Wallerstein et al. 1997), which is thought to produce half the nuclides heavier than Fe andnearly all actinides observed in most metal-poor stars. Although effects of the s-processcan occur in first generation stars (Goriely & Siess 2001; Suda et al. 2004; Nishimura et al.2009), the s-process appears not to be responsible for the enrichment of the interstellarmedium with the elements produced in the early life of the galaxy because it requires longertimescales to populate the ISM with its products. The r-process, however, produces itsnuclei earlier in the galactic history. Because massive stars produce most of the requiredconditions for the r-process, they have long been supposed to synthesize its nuclides.However, detailed calculations (Fischer et al. 2010; Hudepohl et al. 2010; Goriely et al.2011) have suggested that massive stars may not fulfill all of the r-process conditions. Wepresent data that sheds important new light on this question.Stars with masses greater than 8 M ⊙ are thought to become core-collapse supernovae.Those less than 25 M ⊙ will form neutron stars, emitting neutrinos and driving some oftheir newly synthesized nuclei into the interstellar medium. Those more massive thanabout 40 M ⊙ are thought to collapse directly to black holes, contributing nothing to theinterstellar medium. Stars with 25 ≤ M ≤ M ⊙ are thought to collapse to neutron starsthen, as some of the expelled material falls back onto the neutron star, to black holes (Fryer2003; Heger et al. 2003), producing “accretion-induced black holes” and, assuming somematerial can escape, would be expected to contribute some newly synthesized nuclides tothe interstellar medium.We report on some striking astronomical data from extremely metal-poor (EMP) stars,those having [Fe/H] < -2.5. Their metals are thought to have been produced in the first 4 –generation massive stars that were the progenitors of the ones currently observed. Becausethese stars have very low abundances of anything except primordial hydrogen and helium,their abundances of heavy nuclei are presumed to arise primarily from the r-process. Weconcentrate on elements made in the r-process and that are easy to detect in these stars:strontium and barium.Previous models (Qian & Wasserburg 2008; Wanajo et al. 2011; Travaglio et al. 2004)have described separate r-process sites as the cause of an enrichment of light neutron-capturer-process elements. We contrast our model with the “weak” r-process and the light elementprimary process (LEPP) by describing an enrichment in the lighter neutron-captureelements (Honda et al. 2006) that is a natural consequence of a single site but with differinglevels of enrichment of light and heavier neutron-capture elements. Our model producesthe r-process from an explosive event (e.g., a type II supernova), while the r-process ejectais prohibited in some fraction of those events from escaping due to a subsequent collapseof the proto-neutron star to a black hole. The collapse time in this event relative to ther-process ejection time will determine how much r-process material is ejected, as well as itscomposition.
2. Observational Data
Data were taken from the SAGA Database (Suda et al. 2008). The sample includesdata from 186 papers and 4270 stars (1488 unique stars) dealing with observations ofmetal-poor halo stars published prior to 2012. The number of stars for which [Fe/H] < -2.5is 428, reducing to 361 stars if carbon-enhanced EMP stars are excluded. This samplecould suffer errors due to systematic offsets between different studies; a comparison ofresults for the same stars but from different studies shows that these can sometimes beas large as a factor of 3. We thus selected the results of Sr and Ba from observations in 5 –which both elements are reported in the same paper - 260 total stars. The abundances inthis database were all obtained assuming local thermodynamic equilibrium (LTE). Thescatter found in Sr/Ba as well as in Sr/Fe and Ba/Fe in that sample in the metallicityrange of -3.6 < [Fe/H] < -3.0 is observed to be a factor of 100 or more, much larger than theobservational errors.Figure 1 shows the astronomical data for [Sr/Fe], [Ba/Fe], and [Sr/Ba] from the dataset described above, showing a wide range of [Sr/Ba], [Sr/Fe], and [Ba/Fe] values at eachvalue of [Fe/H]. Our interpretation of these features will be given in subsequent sections;the curves in Figure 1 that overlay the data, reflect these efforts.We find three remarkable features in the [Sr/Ba] distribution.1. Almost all stars appear in the range of the functional [Sr/Ba] < − . − [Fe/H]. Thereare two exceptions in [Fe/H] < − .
0. One of them is BD+80 o
245 ([Fe/H]= − . . α -deficient” star (e.g., [Mg/Fe]= − . − .
85 and[Ba/Fe]= − .
89 which was studied by Ivans et al. (2003)). The other is HE0029-1839studied by Barklem et al. (2005), for which little information exists.2. There are only a few stars with [Sr/Ba] < -0.5. We present calculations that show thatthis range is not expected from the r-process or from tr-process, which we discuss inthe next section. The exceptions are as follows: • CS 30322–023 ([Fe/H]=-3.25, [Sr/Ba]=-1.10): a moderately carbon-enhancedstar ([C/Fe]=+0.6) with enhancement of nitrogen and Ba (Masseron et al. 2006;Aoki et al. 2007). • CS 29493–090 ([Fe/H]=-2.82, [Sr/Ba]=-1.41): a moderately carbon-enhancedstar ([C/Fe]=+0.74) with excess of Ba (Barklem et al. 2005). 6 – • CS 22946–011 ([Fe/H]=-2.59, [Sr/Ba]=-0.99) and CS 22941–005 ([Fe/H]=-2.43,[Sr/Ba]=-0.77): binaries with excesses of Ba ([Ba/Fe]=+1.26 and +0.34,respectively: Preston & Sneden (2000)). CS 22950–173 ([Fe/H]=-2.5, [Sr/Ba]=-0.72) which also belongs to be a binary system, though Ba is not enhanced([Ba/Fe]=-0.04: Preston & Sneden (2000)). The CH feature would not bedetected due to high temperature ( ∼ • HE0305-4520 ([Fe/H]=-2.91, [Sr/Ba]=-1.25): Ba is slightly enhanced([Ba/Fe]=+0.59), while an excess of carbon is not clear ([C/Fe]=+0.33:Barklem et al. (2005)).Hence, except for HE0305-4520, all the above stars are (candidate) carbon-enhancedstars, which can be affected by an s-process contribution through mass transfer inbinary systems. HE0305-4520 could also be slightly affected by s-process. In theobjects with low Sr abundances, Sr/Ba could be driven to very low values even by asmall contamination of Ba from the s-process.There is no object for which s-process contamination is not suspected in the region[Sr/Ba] < − . ∼ ∼ − .
6, which will be addressed below, isfound. There are only four stars (CS 22172–002, CD − ◦ < − . < . > .
05 than with [Sr/Ba] < .
05 in the region − . < [Fe/H] < − .
0. If the same distribution is assumed for − . < [Fe/H] < − . > > < − . < − .
6, five stars are carbon-enhanced, and four have very low [Sr/Fe] or low upper limits, suggesting that Sr/Baratios are also quite low. There is no signature to have high Sr/Ba for other twoobjects although the upper limits for Ba abundances are still weak.We note that no new stars with both Sr and Ba measurements for [Fe/H] < − .
3. Galactic Chemical Evolution Interpretation and Turbulence
The most daunting challenges for any attempt to explain the data distribution for theEMP stars are reproduction of the sharp cutoff in [Sr/Ba] at [Fe/H]=-3.6, the explanationof the upper and lower limits as a function of metallicity in that distribution, and the widedispersion in [Sr/Ba].A reduction in the Sr and Ba production in a SN collapse scenario was studiedpreviously (Boyd et al. 2012), using the formalism and results from Woosley et al. (1994).There, the r-process occurs in upper-most 15 shells that appear just at the surface of thenascent neutron star. The mass, thermodynamic trajectories, and initial composition of eachshell are described in Woosley et al. (1994). Those shells were examined in the framework ofthe ejection time, the black-hole collapse time, and the resultant nucleosynthesis, assumingthat the ejection is halted by the collapse to a black hole. We assume that any collapse timeoccurring prior to the shell ejection time in this model will prevent that shell from beingejected. At present, we extend that model to study the relationship between the stellar 8 –metallicity, the progenitor mass, the collapse time, and elemental yields. This model iscoupled to a galactic chemical evolution (GCE) code (Timmes, Woosley, & Weaver 1995).The net result is that a more massive progenitor will generally collapse earlier, resulting inejecta enriched in the lighter neutron-capture elements (A . Y e in the laterstages of the r-process, which would result in a reduced synthesis of the r-process nucleiwith masses in excess of A ∼ ⊙ < M <
30 M ⊙ , the yields are roughtly equalto those of Model 1Max in Cescutti et al. (2006) but slightly less than those of model 2Maxin the same reference and less than those of Raiteri et al. (1999) by about a factor of four(for 10-11 M ⊙ SNIIs). For M <
10 M ⊙ yields are estimated from Figure 3 of Cescutti et al.(2006). The Sr yields (Figure 2) were chosen to reproduce the [Sr/Fe] and [Sr/Ba] GCE 9 –results in Figure 1. The only significant difference between the Ba yields in this paper andthose of Cescutti et al. (2006) corresponds to 30 M ⊙ stars, which are roughly 30 timeshigher than those of model 1Max. This produces Ba early in the GCE model, though thecontribution is insignificant at late times.A power law IMF was used with a Salpeter exponent of -1.31. The black hole collapsetime was computed based on progenitor mass and metallicity and the nuclear equation ofstate (EOS) (O’Connor & Ott 2010). This time was then assumed to be the truncationtime in a dynamic tr-process. For later truncation times, mass shells closer to theproto-neutron star core, blown off later in the explosion model, are assumed to be ejectedeither by achieving a velocity in excess of the escape velocity or perhaps by being ejectedin a jet. Later collapse times correspond to larger production yields of heavier r-processelements.In the r-process, most of the Sr is produced in the early-stage ejectt in mass shellsejected within the first four seconds post-bounce, while most of the Ba is produced inthe later-stage ejecta of a SNII in mass shells ejected from between four and 18 secondspost-bounce. Thus, the Sr yields in a tr-process are fairly constant with collapse time andfairly close to their complete r-process values except for very short collapse times for whichno r-process material is ejected. As only fairly late collapse times will allow for a significantamount of Ba ejection in the tr-process, [Sr/Ba] approaches zero in the later-stage ejecta.Results from the spherical collapse code GR1D(O’Connor & Ott 2010) were used tostudy the relationship between the progenitor star’s mass and metallicity and the Sr andBa yields in massive stars. It is noted that these yields are minimum yields, since in thenon-rotating spherical models employed here, the collapse times are minimum collapsetimes. Longer collapse times may result from effects such as asymmetric explosions,rotations, and neutrino heating induced by these effects. However, since Sr is produced in 10 –shells that are ejected very early on, increasing the collapse time will have little effect onthe Sr production, while significantly increasing the Ba production.Ejection of supernova materials into the interstellar medium resulting from turbulencein the stellar interior may also affect the observed abundances. Although it is difficult tosimulate such effects in a one-dimensional model, we have performed a simplified analysisthat at least suggests the sort of effects that might be observed, and probably to constrainthe magnitude of the predictions. This analysis took each of the 15 individual mass shellsthat were used in our analysis and that could be ejected from the surface of the neutronstar. We then assumed that turbulent ejection might allow a single individual shell to beejected from the star, while the others collapsed onto the proto-neutron star surface. Asnoted above, the outer shell would be rich in Sr and would have very little Ba, while theinner shells would have much less Sr and would be much richer in Ba. Ejection of only thetop or bottom shells should provide limits on the ability of turbulence to explain the rangeof [Sr/Fe], [Ba/Fe], and [Sr/Ba] observed and might also explain the dispersion in the datathat mixing might produce. The [Sr/Ba] results from sites ejecting only these shells areshown in the GCE model in Figure 1d.
4. Results
Assuming a tr-process model resulting in a reduced Sr and Ba production, thesingle-site [Sr/Ba] values as a function of metallicity [Fe/H] are shown in Figure 1 with acollapse model utilizing the LS220 EOS (Lattimer & Swesty 2004), which is one of themost widely used EOSs in numerical simulations of supernova explosions. Several otherEOSs have been examined, and it has been found that a softer EOS results in progressivelylarger [Sr/Ba] values at lower metallicity (corresponding to earlier times in the galacticevolution produced by stars with M ≥
20 M ⊙ ) (Famiano et al. 2012). 11 –In Figure 1, two GCE results are presented. One is the metallicity relationship from amodel based on the input yields from Figure 2 showing [Sr/Fe], [Ba/Fe], and [Sr/Ba] ratios.It can be seen that those fall within the extremes of the observational data points. Theother result is a metallicity relationship corresponding to a GCE model in which all starswith M ≥ ⊙ ultimately collapse to black holes.In this model, an earlier collapse means that shells closer to the neutron star surface(which produce more Ba) will not be ejected, and the overall Ba ejection (summed over allshells) is reduced. Since Sr is mostly produced in outer shells, it is not reduced as much.The net result is that the calculated lines in the figure will correspond to the minimum inthe [Sr/Fe] and [Ba/Fe] values and a maximum in the [Sr/Ba] values. Collapse times rangefrom about 0.25s for 40 M ⊙ stars with a soft EOS to over 3.5s for 20 M ⊙ stars with a stiffEOS.The results of these calculations can be compared to the astronomical data of [Sr/Ba]discussed here. These observations show a sharp maximum in the [Sr/Ba] value for a givenmetallicity. This could be explained by the tr-process limiting the Sr or Ba production asa function of stellar age or mass. We also note a similarly sharp lower limit in the [Ba/Fe]values at [Ba/Fe] <
5. Conclusions
Our detailed inspection of Sr and Ba abundances in Galactic metal-poor stars indicatesa clear cutoff in the distribution of [Sr/Ba] at [Fe/H] ∼ − . < − . − [Fe/H]) as well as a lower bound at [Sr/Ba] ∼ − . > REFERENCES
Aoki, W. et al. 2007, ApJ 655, 492Barklem, P.S. et al. 2005, A&A 439, 129Boyd, R.N., Famiano, M.A., Meyer, B.S., Motizuki, Y., Kajino, T., & Roederer, I.U. 2012,ApJ 744, L14Burbidge, E.M., Burbidge, G.R., Fowler, W.A. & Hoyle, R. 1957, Rev. Mod. Phys. 29, 547Cescutti, G., Francois, P., Matteucci, F., Cayrel, R., & Spite, M. 2006, A&A, 448, 557Famiano, M.A., Boyd, R.N., Kajino, T., Aoki, W., & Suda, T. 2012, ApJ, (in preparation)Famiano, M.A., Liu, T., Lynch, W.G., Rogers, A.M., Tsang, M.B., Wallace, M.S., Charity,R.J., Komarov, S., Sarantites, D.G., & Sobotka, L.G. 2006, Phys. Rev. Lett., 97,052701Fischer, T., Whitehouse, S.C., Mezzacappa, A., Thielemann, F.-K., & Liebendorfer, M.2010, A&A 517, A80Fryer, C.L. 2003, Class. Quantum Grav. 20, S73Goriely, S. & Siess, A. 2001, A&A 378, L25Goriely, S., Chamel, N., Janka, H.-T., & Pearson, J.M. 2011, A&A 531, A78Heger, A., Fryer, C.L., Woosley, S.E., Langer, N. & Hartmann, D.H. 2003, ApJ 591, 288Honda, S., Aoki, W., Ishimaru, Y, Wanajo, S., & Ryan, S.G. 2006, ApJ, 643, 1180Hudepohl, L., Muller, B., Janka, H.-T., Marek, A., & Raffelt, G.G. 2010, Phys. Rev. Lett104, 251101 15 –Ivans, I.I., Sneden, C.S., James, C.R., Preston, G.W., Fulbright, J.P., H¨oflich, P.A., Carney,B.W., & Wheeler, J.G. 2003, ApJ, 592, 906Johnson, J.A., Herwig, F., Beers, T.C., & Christlieb, N. 2007, ApJ, 658, 1203Lattimer, J.M. & Swesty, F.D. 1991, Nuc. Phys. A, 553, 131Masseron, T., van Eck, S., Famaey, B., Goriely, S., Plez, B., Siess, L., Beers, T.C., Primas,F., & Jorissen, A. 2006, A&A, 455, 1059Nishimura, T., Aikawa, M., Suda, T. & Fujimoto, M.Y. 2009, PASJ 61, 909O’Connor, E. & Ott, C.D.2010, Class. Quant. Grav. 27, 114103Preston, G.W. & Sneden, C.S. 2000, AJ, 120, 1014Preston, G.W., Sneden, C.S., Thompson, I.B., Shectman, S.A., Burley, G.S. 2000, AJ, 120,1014Qian, Y.-Z & Wasserburg, G.J. 2008, ApJ 687, 272Raiteri, C.M., Villata, M., Gallino, R., Busso, M., & Cravanzola, A. 1999, ApJ, 518, L91Steiner, A.W., Lattimer, J.M., Brown, E.F. 2010, ApJ, 722, 33Suda, T., Aikawa, M., Machida, M.N., & Fujimoto, M.Y. 2004, ApJ 611, 476Suda, T., Katsuta, Y., Yamada, S, Suwa, T., Ishizuka, C., Komiya, Y., Sorai, K., Aikawa,M., & Fujimoto, Y. 2008, PASJ 60, 1159Timmes, F.X., Woosley, S.E., & Weaver, T.A. 1995, ApJ 98, 617Travaglio, C., Gallion, R., Arnone, E., Cowan, J., Jordan, F., & Sneden, C. 2004, ApJ, 601,684 16 –Wallerstein, G. et al. 1997, Rev. Mod. Phys. 69, 995Wanajo, S., Janka, H.-T., & Muller, B. 2011,ApJ, 765,L15Woosley, S.E., Wilson, J.R., Mathews, G.J., Hoffman, R.D., & Meyer, B.S. 1994, ApJ, 433,229Yong, D., Norris, J. E., Bessell, M. S., et al. 2012, ApJ, in press, arXiv:1208.3003This manuscript was prepared with the AAS L A TEX macros v5.2. 17 – [Fe/H]-4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 [ S r / F e ] -2.5-2-1.5-1-0.500.51 GCE no tr-processGCE LS220 (a) [Fe/H]-4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 [ B a / F e ] -2.5-2-1.5-1-0.500.511.5 (b) [Fe/H]-4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 [ S r / B a ] -2-1.5-1-0.500.511.52 r-II StarsOutliersUpper/Lower Limits[Fe/H]<-3.5 (c) [Fe/H]-4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 [ S r / B a ] -2-1.5-1-0.500.511.52 Shell 0Shell 2Shell 4Shell 6Shell 8Shell 10Shell 14 (d)
Fig. 1.— Results for (a) [Sr/Fe], (b) [Ba/Fe], and (c) [Sr/Ba] assuming a tr-process. Thered dashed line corresponds to a GCE model with no production in a tr-process, and theblue dashed line corresponds to a GCE model with a primary production from a tr-processfor all stars with M ≥ ⊙ . Plot (d) shows GCE results for single-site tr-process productionfor turbulent ejection of specific shells assuming only those shells are ejected. 18 – -7 -6 -5