EDP Sciences logo
Subscriber Authentication Point
Search
Menu
Home All issues Volume 530 (June 2011) A&A, 530 (2011) L14 Full HTML
Issue
A&A
Volume 530, June 2011
Article Number L14
Number of page(s) 5
Section Letters
DOI https://doi.org/10.1051/0004-6361/201117043
Published online 25 May 2011

© ESO, 2011

1. Introduction

Recent claims for the existence of very massive stars (VMS) of up to 300 M⊙ at the centre of young clusters like R136 (Crowther et al. 2010) seem to link the formation of such objects to environments in the centres of massive clusters with  ~104−105   M⊙. This finding thus appears to support massive star-formation models that invoke competitive accretion and possibly even merging in dense clusters (e.g. Bonnell et al. 2004). Such scenarios emerged after it became clear that the disk-accretion scenario – commonly applied to low- and intermediate star formation – had problems explaining the formation of stars with masses above 10 M⊙, as radiation pressure on dust grains would halt and reverse the accretion flow onto the central object (e.g. Yorke & Kruegel 1977). However, recent multi-dimensional hydrodynamical monolithic collapse simulations indicate that massive stars may form via disk accretion after all (e.g. Kuiper et al. 2010). This illustrates that the discussion on clustered vs. isolated massive star formation is still completely open (see de Wit et al. 2005; Parker & Goodwin 2007; Lamb et al. 2010; Bressert et al., in prep.).

thumbnail Fig. 1

Combined YJKs image from the VISTA Magellanic Clouds survey (Cioni et al. 2011). The arrows correspond to 5 pc in the Northern and Eastern directions. The projected distance of VFTS 682 to the cluster R136 is 29 pc (for an assumed LMC distance of 50 kpc).

In this Letter, we present evidence that VFTS 682 (RA 05h38m55.51s Dec  − 69°04′26.72′′) in the Large Magellanic Cloud (LMC), found in isolation from any nearby massive, visible cluster, is one of the most massive stars known. The object suffers from high dust obscuration and is located in, or in the line-of-sight towards, an active star-forming region. On the basis of a mid-infrared (mid-IR) excess, Gruendl & Chu (2009) classified the star as a probable young stellar object (YSO). The VLT-FLAMES Tarantula Survey (VFTS) (Evans et al. 2011) identified it as a new Wolf-Rayet (WR) star, at a projected distance approximately 30 pc northeast of R136. Its spectrum was classified as WN5h and is similar to those of the very luminous stars in the core of R136, with strong H, He ii, and N iv emission lines. In this Letter, we present a photometric and spectroscopic analysis of VFTS 682 to investigate its origin. The star is relatively faint in the optical, which we argue is the result of significant reddening (with Av    ≈    4.5), implying a high intrinsic luminosity and a mass of order 150 M⊙. The sheer presence of such a massive star outside R136 (and apparently isolated from any notable cluster) poses the question of whether it was ejected from R136 or if it was formed in isolation instead.

2. Spectroscopic analysis of VFTS 682

The analysis is based on spectroscopic observations (λ4000 − 7000) from MEDUSA mode of VLT-FLAMES (Evans et al. 2011). We first compared observations taken at different epochs (Evans et al. 2011) as to identify potential shifts in the radial velocity (RV) as a result of binarity. To this purpose, we used the N iv λ4060 emission line. With detection probabilities of 96% for a period P < 10 d, 76% for 10 d  < P < 1 yr, and 28% for 1 yr  < P < 5 yr, we can basically exclude that VFTS 682 is a short period binary. We found the N v λ4944 line to be particularly useful for RV determinations, as this line remained stable for different wind velocity laws. With particular emphasis on the centroid of the line we obtained 300 ± 10 km s-1. If we give more weight to the red wing, we obtain 315 ± 15 km s-1, with a similar number for the He ii λ4686 line.

For the spectral analysis we use the non-LTE model atmosphere code cmfgen (Hillier & Miller 1998) which has been developed to provide accurate abundances, stellar parameters, and ionising fluxes for stars dominated by wind emission lines. We use atomic models containing H i, He i-ii, C iii-iv, N iii-v, O iii-vi, Si iv, P iv-v, S iv-vi, Fe iv-vii and Ni iv-vi. We made the following assumptions. We adopt a metallicity of half solar with respect to Asplund et al. (2005), and assume N and He to have been enriched by the CNO process. We adopt a 12-fold enhancement of the N mass fraction. We employ a β parameter of 1.6 for the wind velocity law and a wind volume filling factor of f = 0.25, assuming that the clumping starts at 10 km s-1. In order to estimate the error bars in the stellar temperature (T ⋆ ) and mass-loss rate (Ṁ), we computed a grid of models around the preferred values.

thumbnail Fig. 2

Relative flux vs. wavelength in Å. Grey solid line: MEDUSA spectrum of VFTS 682. Black dashed line: model spectrum.

Table 1

Stellar parameters of VFTS 682.

The core temperature (T ⋆ ) at high optical depth of 54   500 ± 3000 K is based on the N iv λ4060 and N v doublet λ4604 and 4620 in emission (Fig. 2). N iv is also sensitive to the mass-loss rate, which is obtained from other lines. The hydrogen and the helium lines are used to determine both the helium abundance (45% by mass) and the mass-loss rate, with log (Ṁ/M⊙   yr-1) =  −4.4 ± 0.2, where the volume filling factor corrected mass-loss rate is given in Table 1. This value is larger than the log (Ṁ/M⊙   yr-1) =  −4.69 obtained from the mass-loss recipe of Vink et al. (2001) for non-clumped winds. The values can be brought into agreement for a volume filling factor of f = 0.1. In the absence of UV spectroscopy of the λ1548, λ1551 C iv doublet, we estimate the terminal velocity (3∞ = 2600   km   s-1) from the broadening of the He ii and Hα lines. An overview of the stellar parameters and abundances is provided in Table 1.

The stellar parameters of VFTS 682 are similar to those of R136a3 (with a log (L/L⊙) = 6.58) (Crowther et al. 2010). We note that there are other luminous WNh objects in 30 Dor, for instance, VFTS 617 (=Br 88) which has the same spectral type as VFTS 682 but with weaker emission lines. The spectral analysis of the additional VFTS WNh stars is currently in progress (Bestenlehner et al., in prep.).

3. Luminosity and reddening

In its spectral appearance, VFTS 682 is almost identical to that of R136a3 in the core of the dense cluster R136. The luminous objects in this cluster are thought to be the most massive stars known, with initial masses up to 300   M⊙ (Crowther et al. 2010). As VFTS 682 is apparently isolated, and shows no signs of binarity, it offers the opportunity to study such a luminous WN5h object in isolation. The precise luminosity of VFTS 682 is thus of key relevance for this work. Its derivation is however complicated by the large extinction.

Table 2

Apparent magnitude of VFTS 682.

Matching the optical photometry with a standard extinction law, i.e., a “Galactic average” extinction parameter RV = 3.1, we obtain a luminosity of log (L/L⊙) = 5.7 ± 0.2. With this relatively low luminosity we can explain the optical flux distribution, but we fail to match the observed spectral energy distribution (SED) at longer wavelengths. In the near–mid infrared (IR) the observed flux is  ~3 times higher than the corresponding model flux. Below, we show that we can successfully explain this “excess” IR emission, including its exact shape, with a reddening parameter RV = 4.7, which leads to a much higher stellar luminosity. This effect is visualised in the colour–magnitude diagram (CMD) of Fig. 4. We note that high RV parameters, that deviate from the Galactic average, are not extra-ordinary for massive stars in the LMC (e.g. Bonanos et al. 2011)

3.1. Modelling the near IR flux

Because the extinction in the IR is very low, we can avoid the problems resulting from the uncertain RV parameter, by choosing the near-IR photometry as our flux standard. In more detail, we derive the luminosity of VFTS 682 by matching B, and V from Parker (1993), and Ks from the “InfraRed Survey Facility (IRSF) Magellanic Clouds Point Source Catalog” (Kato et al. 2007, see Table 2). For this purpose we extract the intrinsic B, V, and Ks magnitudes from our model, using appropriate filter functions. Based on the resulting values of E(B − V) = 0.94 ± 0.02, and E(V − Ks) = 3.9, we derive RV = AV/E(B − V) for two oft-used extinction laws.

To determine RV, we use the relations RV = 1.1994 × E(V − Ks)/E(B − V) − 0.183, as inferred from the extinction law by Cardelli et al. (1989), and RV = 1.12 × E(V − Ks)/E(B − V) + 0.016, from Fitzpatrick (1999). This way, we obtain RV = 4.7 ± 0.1, AV = 4.45 ± 0.12, and AKs = 0.55 ± 0.15, where the uncertainties refer to the differences in the adopted extinction laws. As AV is much larger than AKs, the derived luminosity mostly relies on Ks, while the observed values of B, and V mostly determine RV. Small uncertainties in B − V will thus mainly affect RV, in a way that the absolute values of B, and V stay the same. As RV = AV/E(B − V), this introduces an anti-correlation between the derived RV, and E(B − V), where AV is preserved. The resulting luminosity is log (L/L⊙) = 6.5 ± 0.2, where the error includes uncertainties in the stellar temperature, mass-loss rate, and extinction. The resulting SED fit is presented in Fig. 3.

thumbnail Fig. 3

Log flux vs. log wavelength. Squares: B,   V,   J,   H,   Ks photometry, Diamonds: Spitzer SAGE. Circles: Spitzer 5.8   μm and 8.0   μm from Gruendl & Chu (2009). Solid line: SED with anomalous RV parameter. Dashed line: standard RV parameter. Dashed-Dotted line: attempt to reproduce the NIR excess with a 1500 K black body.

thumbnail Fig. 4

VFTS objects. Rectangles are WN-stars. Triangles are WC-stars. The arrows in the top right corner are for two different RV values, with a length corresponding to AV = 1. The longer lines on the left indicate the transformations between reddened and unreddened CMD positions.

In an independent analysis of the available UBVIJHK photometry, using the CHORIZOS code (Maíz-Apellániz 2007) for CHi-square cOde for parameteRized modelIng and characteriZation of phOtometry and Spectrophotometry, we obtain E(B − V) = 0.98 ± 0.03, and RV = 4.55 ± 0.17, which, due to the aforementioned anti-correlation between E(B − V) and RV, results in an almost unchanged value of AV = 4.46 ± 0.06. The resulting luminosity, and high RV are thus confirmed.

Table  2 gives a summary of the available photometry for VFTS 682. In analogy to our previous findings, the differences between the optical photometry of Evans et al. (2011), and Parker (1993) mainly affect RV, but not the derived luminosity. Based on the WFI photometry by Evans et al. (2011), we obtain RV = 5.2 ± 0.1 with E(B − V) = 0.84 ± 0.02. The near-IR photometry however has a direct influence on our results. Based on the fainter Two Micron All Sky Survey (2MASS) photometry the derived luminosity decreases by about 0.1 dex to log (L/L⊙) = 6.4. In this work, we use the IRSF values, as they connect better to the observed Spitzer MIR photometry.

If we alternatively adopt the standard reddening parameter RV = 3.1, and attribute the NIR excess to a second component, such as a cool star and/or a warm (1000–2000 K) dust component (see Fig. 3), we obtain a luminosity of only log (L) = 5.7 ± 0.2. We note that just one extra component (cool star and/or dust black body) cannot fit the entire SED slope, which implies that we would need a multitude of additional components. These would need to conspire in such a way as to precisely follow our SED. Although we cannot disprove such a configuration, it would be highly contrived.

To summarise, the most likely luminosity of VFTS 682 is log (L) = 6.5 ± 0.2, which would support the high luminosities that have been derived by Crowther et al. (2010) for the VMS in the core of R 136.

3.2. The MIR excess

We also investigate the issue of whether VFTS 682 has a mid-IR excess in Spitzer data. We compared two sources of MIR photometry, the “Surveying the Agents of a Galaxy’s Evolution (SAGE) IRAC Catalog” for point sources (Meixner et al. 2006) as well as the recent YSO catalog of Gruendl & Chu (2009). In contrast to the SAGE IRAC Catalog, Gruendl & Chu (2009) included spatially more extended sources. The MIR photometry data from both sources are listed in Table 2. Gruendl & Chu (2009) detected a mid-IR excess at 5.8 and 8.0 μm and categorised VFTS 682 as a YSO candidate. We are not able to confirm the MIR excess on the basis of the SAGE IRAC Catalog photometry, but the difference in the two data sources may be explained in the following way.

The MIR point-source data of Meixner et al. (2006) basically follow the slope of our optical-to-NIR fit continued into the MIR, whilst the larger scale MIR data of Gruendl & Chu (2009) show a clear MIR excess. Whereas the MIR excess could be consistent with an extended circumstellar shell, such as recently detected in Spitzer 24 micron images of the Galactic Centre WNh star WR102ka (Barniske et al. 2008), there is no evidence for such a shell (or bowshock) in Spitzer 8 micron images of VFTS 682 at similar angular scales as for WR102ka. However, given the resolution limit of  ~2 arcsec the IR excess might be due to an unresolved circumstellar shell with a diameter of  <0.5 pc.

4. Variability on timescales of years

To study the stability of VFTS 682 we show Optical Gravitational Lensing Experiment (OGLE-III) lightcurves (Udalski et al. 2008) in V and I in Fig. 5. On top of the short-term jitter, the object clearly dimmed by  ~0.1 mag in both the V and I band during the years 2006–2009. Furthermore, the object shows an increase in the K-band by  ~0.15 mag during 2010 (Evans et al., in prep.). This kind of long-term variability is unprecedented for Wolf-Rayet stars, and is more characteristic for Luminous Blue Variables (LBVs). However, we note that the nature of these changes is not the same as those of bona-fide LBVs with S Doradus cycles of 1–2 mag variations (Humphreys & Davidson 1994).

5. Discussion on the origin of VFTS 682

VFTS 682 is located in an active star-formation region of 30 Dor (Johansson et al. 1998). It is not anywhere close to the core of R136, or – as far as we can tell – to any other nearby cluster. This might be considered a surprise as VMS are normally found in the cores of large clusters, such as the Arches cluster or R136. By contrast, VFTS 682 is rather isolated at a projected distance of 29 pc from R136. This raises the question of whether the object formed in situ or is a slow runaway object from R136 instead.

In order to address these issues, we provide some velocity estimates. The measured RV shift of VFTS 682 is vr ≈ 300 km s-1, which is somewaht higher than the average velocity in 30 Dor (~270 ± 10 km s-1, Sana et al., in prep.; Bosch et al. 2009). If VFTS 682 is a runaway from R136, it would require a tangential velocity of 30 km s-1 to appear at a projected distance of 30 pc within approximately 1 Myr. Together with an RV offset of  ~30 km s-1, VFTS 682 would then require a true velocity of  ~40 km s-1, i.e. at the lower range of velocities for classical OB runaway stars (e.g. Philp et al. 1996). Still, this would make it the most massive runaway star known to date.

thumbnail Fig. 5

OGLE III V and I lightcurves for VFTS 682 during 2001–2009.

In case the object is a runaway star, a bow shock might potentially be visible as VFTS 682 is surrounded by dust clouds. Whilst there is currently no evidence for a bow shock, one of the possible explanations for the MIR excess is the presence of a bow-shock (e.g. Gvaramadze et al. 2010). Alternative explanations for the MIR excess may involve its (line-of-sight) association with an active star forming region, or a nebula formed by vigorous mass loss since the object started to burn hydrogen in its core. In this context, it may or may not be relevant that VFTS 682 shows slow photometric variations suggestive of LBVs that may suffer from mass ejections.

The most probable luminosity of VFTS 682 is log (L) = 6.5 ± 0.2. The question is what stellar mass the object corresponds to, and what is its most likely evolutionary age. The high temperature places the object right on the zero-age main sequence (ZAMS), which can be best understood as a result of chemically homogeneous evolution (CHE). Such an evolution has also been proposed for two WR stars in the SMC (Martins et al. 2009). Note that this type of evolution has been suggested for the production of long gamma-ray bursts (Yoon & Langer 2005). In order to obtain a meaningful mass limit, we employ the recent mass-luminosity relationships by Gräfener et al. (2011) for homogeneous hydrogen burners. Utilising the derived He abundance (Y = 0.45), the most likely present-day mass is  ~150 M⊙. In detailed stellar evolution models of Brott et al. (2011) and Friedrich et al. (in prep), CHE is achieved through rapid rotation, with  km s-1. A value of Y = 0.45 is obtained at an age of 1–1.4 Myr and the initial mass would be of  ~120–210 M⊙, where the mass range is defined through the uncertainty in the luminosity.

Finally, from our analysis it is clear that VFTS 682 is a very massive object. It is often assumed that such massive objects can only be formed in dense cluster environments, where they are normally found. The apparent isolation of VFTS 682 may thus represent an interesting challenge for dynamical ejection scenarios and/or massive star formation theory.

References

  1. Asplund, M., Grevesse, N., & Sauval, A. J. 2005, ASPCS, 336, 25 [Google Scholar]
  2. Barniske, A., Oskinova, L. M., & Hamann, W. 2008, A&A, 486, 971 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  3. Bonanos, A. Z., Castro, N., Macri, L. M., & Kudritzki, R. 2011, ApJ, 729, L9 [NASA ADS] [CrossRef] [Google Scholar]
  4. Bonnell, I. A., Vine, S. G., & Bate, M. R. 2004, MNRAS, 349, 735 [NASA ADS] [CrossRef] [Google Scholar]
  5. Bosch, G., Terlevich, E., & Terlevich, R. 2009, AJ, 137, 3437 [NASA ADS] [CrossRef] [Google Scholar]
  6. Brott, I., de Mink, S. E., Cantiello, M., et al. 2011, A&A, 530, A115 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  7. Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245 [NASA ADS] [CrossRef] [Google Scholar]
  8. Cioni, M., Clementini, G., Girardi, L., et al. 2011, A&A, 527, A116 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  9. Crowther, P. A., Schnurr, O., Hirschi, R., et al. 2010, MNRAS, 408, 731 [NASA ADS] [CrossRef] [Google Scholar]
  10. de Wit, W. J., Testi, L., Palla, F., & Zinnecker, H. 2005, A&A, 437, 247 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  11. DENIS. 2005, VizieR Online Data Catalog, 1, 2002 [Google Scholar]
  12. Evans, C. J., Taylor, W. D., Henault-Brunet, V., et al. 2011, A&A, 530, A108 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  13. Fitzpatrick, E. L. 1999, PASP, 111, 63 [NASA ADS] [CrossRef] [Google Scholar]
  14. Gräfener, G., Vink, J. S., de Koter, A., & Langer, N. 2011, A&A, submitted [Google Scholar]
  15. Gruendl, R. A., & Chu, Y. 2009, ApJS, 184, 172 [Google Scholar]
  16. Gvaramadze, V. V., Kroupa, P., & Pflamm-Altenburg, J. 2010, A&A, 519, A33 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  17. Hillier, D. J., & Miller, D. L. 1998, ApJ, 496, 407 [NASA ADS] [CrossRef] [Google Scholar]
  18. Humphreys, R. M., & Davidson, K. 1994, PASP, 106, 1025 [NASA ADS] [CrossRef] [Google Scholar]
  19. Johansson, L. E. B., Greve, A., Booth, R. S., et al. 1998, A&A, 331, 857 [NASA ADS] [Google Scholar]
  20. Kato, D., Nagashima, C., Nagayama, T., et al. 2007, PASJ, 59, 615 [NASA ADS] [Google Scholar]
  21. Kuiper, R., Klahr, H., Beuther, H., & Henning, T. 2010, ApJ, 722, 1556 [CrossRef] [Google Scholar]
  22. Lamb, J. B., Oey, M. S., Werk, J. K., & Ingleby, L. D. 2010, ApJ, 725, 1886 [NASA ADS] [CrossRef] [Google Scholar]
  23. Maíz-Apellániz, J. 2007, ASPCS, 364, 227 [Google Scholar]
  24. Martins, F., Hillier, D. J., Bouret, J. C., et al. 2009, A&A, 495, 257 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  25. Meixner, M., Gordon, K. D., Indebetouw, R., et al. 2006, AJ, 132, 2268 [NASA ADS] [CrossRef] [Google Scholar]
  26. Parker, J. W. 1993, AJ, 106, 560 [NASA ADS] [CrossRef] [Google Scholar]
  27. Parker, R. J., & Goodwin, S. P. 2007, MNRAS, 380, 1271 [NASA ADS] [CrossRef] [Google Scholar]
  28. Philp, C. J., Evans, C. R., Leonard, P. J. T., & Frail, D. A. 1996, AJ, 111, 1220 [NASA ADS] [CrossRef] [Google Scholar]
  29. Udalski, A., Szymanski, M. K., Soszynski, I., et al. 2008, Acta Astron., 58, 69 [NASA ADS] [Google Scholar]
  30. Vink, J. S., de Koter, A., & Lamers, H. J. G. L. M. 2001, A&A, 369, 574 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  31. Yoon, S., & Langer, N. 2005, A&A, 443, 643 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  32. Yorke, H. W., & Kruegel, E. 1977, A&A, 54, 183 [NASA ADS] [Google Scholar]

All Tables

Table 1

Stellar parameters of VFTS 682.

In the text
Table 2

Apparent magnitude of VFTS 682.

In the text

All Figures

thumbnail Fig. 1

Combined YJKs image from the VISTA Magellanic Clouds survey (Cioni et al. 2011). The arrows correspond to 5 pc in the Northern and Eastern directions. The projected distance of VFTS 682 to the cluster R136 is 29 pc (for an assumed LMC distance of 50 kpc).
In the text
thumbnail Fig. 2

Relative flux vs. wavelength in Å. Grey solid line: MEDUSA spectrum of VFTS 682. Black dashed line: model spectrum.
In the text
thumbnail Fig. 3

Log flux vs. log wavelength. Squares: B,   V,   J,   H,   Ks photometry, Diamonds: Spitzer SAGE. Circles: Spitzer 5.8   μm and 8.0   μm from Gruendl & Chu (2009). Solid line: SED with anomalous RV parameter. Dashed line: standard RV parameter. Dashed-Dotted line: attempt to reproduce the NIR excess with a 1500 K black body.
In the text
thumbnail Fig. 4

VFTS objects. Rectangles are WN-stars. Triangles are WC-stars. The arrows in the top right corner are for two different RV values, with a length corresponding to AV = 1. The longer lines on the left indicate the transformations between reddened and unreddened CMD positions.
In the text
thumbnail Fig. 5

OGLE III V and I lightcurves for VFTS 682 during 2001–2009.
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.