F. R. N. Schneider

Binary Interaction Dominates the Evolution of Massive Stars

Star Partners Stars more massive than eight times the mass of the Sun are rare and short-lived, yet they are fundamentally important because they produce all the heavy elements in the universe, such as iron, silicon, and calcium. Sana et al. (p. 444) examined the properties of a sample of ∼70 massive stars in six stellar clusters located nearby in our galaxy. Over half of the stars in the sample belong to a binary system and, during the course of their lifetimes, most of the stars in these binaries will interact with one another, either by merging or exchanging mass. Binary interaction may thus affect the evolution of the majority of massive stars. Analysis of a sample of massive stars in our Galaxy implies that most will interact with a nearby companion. The presence of a nearby companion alters the evolution of massive stars in binary systems, leading to phenomena such as stellar mergers, x-ray binaries, and gamma-ray bursts. Unambiguous constraints on the fraction of massive stars affected by binary interaction were lacking. We simultaneously measured all relevant binary characteristics in a sample of Galactic massive O stars and quantified the frequency and nature of binary interactions. More than 70% of all massive stars will exchange mass with a companion, leading to a binary merger in one-third of the cases. These numbers greatly exceed previous estimates and imply that binary interaction dominates the evolution of massive stars, with implications for populations of massive stars and their supernovae.

The incidence of stellar mergers and mass gainers among massive stars

Because the majority of massive stars are born as members of close binary systems, populations of massive main-sequence stars contain stellar mergers and products of binary mass transfer. We simulate populations of massive stars accounting for all major binary evolution effects based on the most recent binary parameter statistics and extensively evaluate the effect of model uncertainties. Assuming constant star formation, we find that 8^(+9)_(-4)% of a sample of early-type stars are the products of a merger resulting from a close binary system. In total we find that 30^(+10)_(-15)% of massive main-sequence stars are the products of binary interaction. We show that the commonly adopted approach to minimize the effects of binaries on an observed sample by excluding systems detected as binaries through radial velocity campaigns can be counterproductive. Systems with significant radial velocity variations are mostly pre-interaction systems. Excluding them substantially enhances the relative incidence of mergers and binary products in the non-radial velocity variable sample. This poses a challenge for testing single stellar evolutionary models. It also raises the question of whether certain peculiar classes of stars, such as magnetic O stars, are the result of binary interaction and it emphasizes the need to further study the effect of binarity on the diagnostics that are used to derive the fundamental properties (star-formation history, initial mass function, mass-to-light ratio) of stellar populations nearby and at high redshift.

AGES OF YOUNG STAR CLUSTERS, MASSIVE BLUE STRAGGLERS, AND THE UPPER MASS LIMIT OF STARS: ANALYZING AGE-DEPENDENT STELLAR MASS FUNCTIONS

Massive stars rapidly change their masses through strong stellar winds and mass transfer in binary systems. The latter aspect is important for populations of massive stars as more than 70% of all O stars are expected to interact with a binary companion during their lifetime. We show that such mass changes leave characteristic signatures in stellar mass functions of young star clusters that can be used to infer their ages and to identify products of binary evolution. We model the observed present-day mass functions of the young Galactic Arches and Quintuplet star clusters using our rapid binary evolution code. We find that the shaping of the mass function by stellar wind mass loss allows us to determine the cluster ages as 3.5 ± 0.7 Myr and 4.8 ± 1.1 Myr, respectively. Exploiting the effects of binary mass exchange on the cluster mass function, we find that the most massive stars in both clusters are rejuvenated products of binary mass transfer, i.e., the massive counterpart of classical blue straggler stars. This resolves the problem of an apparent age spread among the most luminous stars exceeding the expected duration of star formation in these clusters. We perform Monte Carlo simulations to probe stochastic sampling, which support the idea of the most massive stars being rejuvenated binary products. We find that the most massive star is expected to be a binary product after 1.0 ± 0.7 Myr in Arches and after 1.7 ± 1.0 Myr in Quintuplet. Today, the most massive 9 ± 3 stars in Arches and 8 ± 3 in Quintuplet are expected to be such objects. Our findings have strong implications for the stellar upper mass limit and solve the discrepancy between the claimed 150 M ☉ limit and observations of four stars with initial masses of 165-320 M ☉ in R136 and of supernova 2007bi, which is thought to be a pair-instability supernova from an initial 250 M ☉ star. Using the stellar population of R136, we revise the upper mass limit to values in the range 200-500 M ☉.

The evolution of rotating very massive stars with LMC composition

Context. With growing evidence for the existence of very massive stars at subsolar metallicity, there is an increased need for corresponding stellar evolution models. Aims. We present a dense model grid with a tailored input chemical composition appropriate for the Large Magellanic Cloud (LMC). Methods. We use a one-dimensional hydrodynamic stellar evolution code, which accounts for rotation, transport of angular momentum by magnetic fields, and stellar wind mass loss to compute our detailed models. We calculate stellar evolution models with initial masses from 70 to 500 M⊙ and with initial surface rotational velocities from 0 to 550 km s-1, covering the core-hydrogen burning phase of evolution. Results. We find our rapid rotators to be strongly influenced by rotationally induced mixing of helium, with quasi-chemically homogeneous evolution occurring for the fastest rotating models. Above 160 M⊙, homogeneous evolution is also established through mass loss, producing pure helium stars at core hydrogen exhaustion independent of the initial rotation rate. Surface nitrogen enrichment is also found for slower rotators, even for stars that lose only a small fraction of their initial mass. For models above ~150 M⊙ at zero age, and for models in the whole considered mass range later on, we find a considerable envelope inflation due to the proximity of these models to their Eddington limit. This leads to a maximum ZAMS surface temperature of ~56 000 K, at ~180 M⊙, and to an evolution of stars in the mass range 50 M⊙...100 M⊙ to the regime of luminous blue variables in the Hertzsprung-Russell diagram with high internal Eddington factors. Inflation also leads to decreasing surface temperatures during the chemically homogeneous evolution of stars above ~180 M⊙. Conclusions. The cool surface temperatures due to the envelope inflation in our models lead to an enhanced mass loss, which prevents stars at LMC metallicity from evolving into pair-instability supernovae. The corresponding spin-down will also prevent very massive LMC stars to produce long-duration gamma-ray bursts, which might, however, originate from lower masses.

The VLT-FLAMES Tarantula Survey. XI. A census of the hot luminous stars and their feedback in 30 Doradus

Context. The VLT-FLAMES Tarantula Survey has an extensive view of the copious number of massive stars in the 30 Doradus (30 Dor) star forming region of the Large Magellanic Cloud. These stars play a crucial role in our understanding of the stellar feedback in more distant, unresolved star forming regions. Aims. The first comprehensive census of hot luminous stars in 30 Dor is compiled within a 10 arcmin (150 pc) radius of its central cluster, R136. We investigate the stellar content and spectroscopic completeness of the early type stars. Estimates were made for both the integrated ionising luminosity and stellar wind luminosity. These values were used to re-assess the star formation rate (SFR) of the region and determine the ionising photon escape fraction. Methods. Stars were selected photometrically and combined with the latest spectral classifications. Spectral types were estimated for stars lacking spectroscopy and corrections were made for binary systems, where possible. Stellar calibrations were applied to obtain their physical parameters and wind properties. Their integrated properties were then compared to global observations from ultraviolet (UV) to far-infrared (FIR) imaging as well as the population synthesis code, Starburst99. Results. Our census identified 1145 candidate hot luminous stars within 150 pc of R136 of which >700 were considered to be genuine early type stars and contribute to feedback. We assess the survey to be spectroscopically complete to 85% in the outer regions (>5 pc) but only 35% complete in the region of the R136 cluster, giving a total of 500 hot luminous stars in the census which had spectroscopy. Only 31 were found to be Wolf-Rayet (W-R) or Of/WN stars, but their contribution to the integrated ionising luminosity and wind luminosity was ~40% and ~50%, respectively. Similarly, stars with Minit > 100 M⊙ (mostly H-rich WN stars) also showed high contributions to the global feedback, ~25% in both cases. Such massive stars are not accounted for by the current Starburst99 code, which was found to underestimate the integrated ionising luminosity of R136 by a factor ~2 and the wind luminosity by a factor ~9. The census inferred a SFR for 30 Dor of 0.073 ± 0.04 M⊙ yr-1. This was generally higher than that obtained from some popular SFR calibrations but still showed good consistency with the far-UV luminosity tracer as well as the combined Hα and mid-infrared tracer, but only after correcting for Hα extinction. The global ionising output was also found to exceed that measured from the associated gas and dust, suggesting that ~6+55-6 % of the ionising photons escape the region. Conclusions. When studying the most luminous star forming regions, it is essential to include their most massive stars if one is to determine a reliable energy budget. Photon leakage becomes more likely after including their large contributions to the ionising output. If 30 Dor is typical of other massive star forming regions, estimates of the SFR will be underpredicted if this escape fraction is not accounted for.

Bonnsai: a Bayesian tool for comparing stars with stellar evolution models

Powerful telescopes equipped with multi-fibre or integral field spectrographs combined with detailed models of stellar atmospheres and automated fitting techniques allow for the analysis of large number of stars. These datasets contain a wealth of information that require new analysis techniques to bridge the gap between observations and stellar evolution models. To that end, we develop Bonnsai (BONN Stellar Astrophysics Interface), a Bayesian statistical method, that is capable of comparing all available observables simultaneously to stellar models while taking observed uncertainties and prior knowledge such as initial mass functions and distributions of stellar rotational velocities into account. Bonnsai can be used to (1) determine probability distributions of fundamental stellar parameters such as initial masses and stellar ages from complex datasets, (2) predict stellar parameters that were not yet observationally determined and (3) test stellar models to further advance our understanding of stellar evolution. An important aspect of Bonnsai is that it singles out stars that cannot be reproduced by stellar models through 2 hypothesis tests and posterior predictive checks. Bonnsai can be used with any set of stellar models and currently supports massive main-sequence single star models of Milky Way and Large and Small Magellanic Cloud composition. We apply our new method to mock stars to demonstrate its functionality and capabilities. In a first application, we use Bonnsai to test the stellar models of Brott et al. (2011a) by comparing the stellar ages inferred for the primary and secondary stars of eclipsing Milky Way binaries of which the components range in mass between 4:5 and 28 M . Ages are determined from dynamical masses and radii that are known to better than 3%. We show that the stellar models must include rotation because stellar radii can be increased by several percent via centrifugal forces. We find that the average age di erence between the primary and secondary stars of the binaries is 0:9 2:3 Myr (95% CI), i.e. that the stellar models reproduce the Milky Way binaries well. The predicted e ective temperatures are in agreement for observed e ective temperatures for stars cooler than 25; 000 K. In hotter stars, i.e. stars earlier than B1‐2V and more massive than about 10 M , we find that the observed e ective temperatures are on average hotter by 1:1 0:3 kK (95% CI) than predicted by the stellar models. The hotter temperatures consequently result in bolometric luminosities that are larger by 0:06 0:02 dex (95% CI) than those predicted by the models.

The R136 star cluster dissected with Hubble Space Telescope/STIS. I. Far-ultraviolet spectroscopic census and the origin of He ii λ1640 in young star clusters

We introduce a Hubble Space Telescope (HST)/Space Telescope Imaging Spectrograph (STIS) stellar census of R136a, the central ionizing star cluster of 30 Doradus. We present low resolution far-ultraviolet STIS spectroscopy of R136 using 17 contiguous 52 arcsec x 0.2 arcsec slits which together provide complete coverage of the central 0.85 parsec (3.4 arcsec). We provide spectral types of 90 per cent of the 57 sources brighter than m(F555W) = 16.0 mag within a radius of 0.5 parsec of R136a1, plus 8 additional nearby sources including R136b (O4 If/WN8). We measure wind velocities for 52 early-type stars from C IV lambda lambda 1548-51, including 16 O2-3 stars. For the first time, we spectroscopically classify all Weigelt and Baier members of R136a, which comprise three WN5 stars (a1-a3), two O supergiants (a5-a6) and three early O dwarfs (a4, a7, a8). A complete Hertzsprung-Russell diagram for the most massive O stars in R136 is provided, from which we obtain a cluster age of 1.5(-0.7)(+0.3) Myr. In addition, we discuss the integrated ultraviolet spectrum of R136, and highlight the central role played by the most luminous stars in producing the prominent He II lambda 1640 emission line. This emission is totally dominated by very massive stars with initial masses above similar to 100M(circle dot). The presence of strong He II lambda 1640 emission in the integrated light of very young star clusters (e.g. A1 in NGC 3125) favours an initial mass function extending well beyond a conventional upper limit of 100M(circle dot). We include montages of ultraviolet spectroscopy for Large Magellanic Cloud O stars in the appendix. Future studies in this series will focus on optical STIS medium resolution observations.

The VLT-FLAMES Tarantula Survey. XVII. Physical and wind properties of massive stars at the top of the main sequence

The evolution and fate of very massive stars (VMS) is tightly connected to their mass-loss properties. Their initial and final masses differ significantly as a result of mass loss. VMS have strong stellar winds and extremely high ionising fluxes, which are thought to be critical sources of both mechanical and radiative feedback in giant Hii regions. However, how VMS mass-loss properties change during stellar evolution is poorly understood. In the framework of the VLT-Flames Tarantula Survey (VFTS), we explore the mass-loss transition region from optically thin O to denser WNh star winds, thereby testing theoretical predictions. To this purpose we select 62 O, Of, Of/WN, and WNh stars, an unprecedented sample of stars with the highest masses and luminosities known. We perform a spectral analysis of optical VFTS as well as near-infrared VLT/SINFONI data using the non-LTE radiative transfer code CMFGEN to obtain stellar and wind parameters. For the first time, we observationally resolve the transition between optically thin O and optically thick WNh star winds. Our results suggest the existence of a kink between both mass-loss regimes, in agreement with recent MC simulations. For the optically thick regime, we confirm the steep dependence on the Eddington factor from previous theoretical and observational studies. The transition occurs on the MS near a luminosity of 10^6.1Lsun, or a mass of 80...90Msun. Above this limit, we find that - even when accounting for moderate wind clumping (with f = 0.1) - wind mass-loss rates are enhanced with respect to standard prescriptions currently adopted in stellar evolution calculations. We also show that this results in substantial helium surface enrichment. Based on our spectroscopic analyses, we are able to provide the most accurate ionising fluxes for VMS known to date, confirming the pivotal role of VMS in ionising and shaping their environments.

B fields in OB stars (BOB): Low-resolution FORS2 spectropolarimetry of the first sample of 50 massive stars

Within the context of the collaboration “B fields in OB stars” (BOB), we used the FORS2 low-resolution spectropolarimeter to search for a magnetic field in 50 massive stars, including two reference magnetic massive stars. Because of the many controversies of magnetic field detections obtained with the FORS instruments, we derived the magnetic field values with two completely independent reduction and analysis pipelines. We compare and discuss the results obtained from the two pipelines. We obtained a general good agreement, indicating that most of the discrepancies on magnetic field detections reported in the literature are caused by the interpretation of the significance of the results (i.e., 3–4σ detections considered as genuine, or not), instead of by significant differences in the derived magnetic field values. By combining our results with past FORS1 measurements of HD 46328, we improve the estimate of the stellar rotation period, obtaining P = 2.17950 ± 0.00009 days. For HD 125823, our FORS2 measurements do not fit the available magnetic field model, based on magnetic field values obtained 30 years ago. We repeatedly detect a magnetic field for the O9.7V star HD 54879, the HD 164492C massive binary, and the He-rich star CPD −57 3509. We obtain a magnetic field detection rate of 6 ± 4%, while by considering only the apparently slow rotators we derive a detection rate of 8 ± 5%, both comparable with what was previously reported by other similar surveys. We are left with the intriguing result that, although the large majority of magnetic massive stars is rotating slowly, our detection rate is not a strong function of the stellar rotational velocity.

Confronting uncertainties in stellar physics: calibrating convective overshooting with eclipsing binaries

As part of a larger program aimed at better quantifying the uncertainties in stellar computations, we attempt to calibrate the extent of convective overshooting in low to intermediate mass stars by means of eclipsing binary systems. We model 12 such systems, with component masses between 1.3 and 6.2 solar masses, using the detailed binary stellar evolution code STARS, producing grids of models in both metallicity and overshooting parameter. From these, we determine the best fit parameters for each of our systems. For three systems, none of our models produce a satisfactory fit. For the remaining systems, no single value for the convective overshooting parameter fits all the systems, but most of our systems can be well described with an overshooting parameter between 0.09 and 0.15, corresponding to an extension of the mixed region above the core of about 0.1-0.3 pressure scale heights. Of the nine systems where we are able to obtain a good fit, seven can be reasonably well fit with a single parameter of 0.15. We find no evidence for a trend of the extent of overshooting with either mass or metallicity, though the data set is of limited size. We repeat our calculations with a second evolution code, MESA, and we find general agreement between the two codes. For the extension of the mixed region above the convective core required by the MESA models is about 0.15-0.4 pressure scale heights. For the system EI Cep, we find that MESA gives an overshooting region that is larger than the STARS one by about 0.1 pressure scale heights for the primary, while for the secondary the difference is only 0.05 pressure scale heights.