EDP Sciences logo
Subscriber Authentication Point
Search
Menu
Home All issues Volume 564 (April 2014) A&A, 564 (2014) L7 Full HTML
Free Access
Issue
A&A
Volume 564, April 2014
Article Number L7
Number of page(s) 5
Section Letters
DOI https://doi.org/10.1051/0004-6361/201323289
Published online 02 April 2014

© ESO, 2014

1. Introduction

Present understanding suggests that most stars form by disc accretion during pre-main sequence (PMS) evolution. In the favoured accretion model, the disc forms during protostellar collapse and is disrupted by a stellar magnetosphere, which channels gas from the disc onto the stellar surface (see Hartmann 2008). Practical limitations have restricted observations of PMS stars to nearby solar-metallicity star-forming regions. Consequently, the properties of low-metallicity PMS stars, located in extra-galactic star-forming regions remain largely unknown.

The formation of stars may be significantly influenced by the metallicity (Z) of the local medium. The extra cooling provided by increasing Z is thought to decrease stellar mass. This is perhaps exemplified in the contrast between the present-day low-mass stellar initial mass function (IMF) and the anticipated high-mass population III IMF (Larson 1998). This may imply higher mass accretion rates at lower Z (Omukai & Palla 2001).

thumbnail Fig. 1

Median FLAMES-Giraffe spectra of VFTS 822.
Open with DEXTER

The only current possibility of observing low Z PMS stars spectroscopically is in the Magellanic clouds. Medium-resolution spectra, rather than photometric observations, are necessary as high-mass PMS systems at large distances become increasingly confused. They cannot be separated from more evolved systems occupying similar regions of the Hertzsprung-Russell (H-R) diagram using photometry alone. Another driver for observing PMS stars spectroscopically is the possibility to measure mass accretion rates using different indicators.

Medium-resolution stellar spectroscopy with eight-to ten-metre class telescopes leads to a limiting magnitude of V~ 18. In the nearest low-Z galaxy, the Large Magellanic Cloud (LMC) located at a distance of 50 kpc (Schaefer 2008), this translates to a mid-late B spectral type. These stars are more massive than 4 M⊙ and reside near the main sequence with a very short PMS phase. Therefore, detection of even a single PMS star is contentious, although several candidates have been reported (de Wit et al. 2005; Clayton et al. 2010; De Marchi et al. 2010).

One approach to identifying high-mass PMS stars is to discover them serendipitously from large spectroscopic surveys. The VLT-FLAMES Tarantula Survey (VFTS; Evans et al. 2011) covering over 800 massive stars in the 30 Doradus starburst region of the LMC (Z ~ 0.5 Z⊙) provides an ideal dataset for this task. From the VFTS sample, we identify VFTS 822 (RA(J2000) = 05h39m38.49s, Dec(J2000) =  −69°09′00.5′′) as an accreting PMS candidate. We describe the characteristics of VFTS 822 that indicate it is in an accreting phase. However, since VFTS 822 exhibits the B[e] phenomenon (Lamers et al. 1998), its evolutionary status is unclear. Therefore, we explore the possibility that it could also be a B[e] supergiant.

2. Results

2.1. VFTS spectroscopy

Multi-epoch observations covering 3950–5050 Å (R ≈ 8000) and 6450–6800 Å (R = 16 000) were taken using the FLAMES multi-fibre spectrograph (Pasquini et al. 2002). Further details on the observations and data reduction techniques can be found in Evans et al. (2011). Spectra of VFTS 822 are plotted in Fig. 1. Strong Hα, Hβ, and Hγ emission are seen, together with emission lines from both permitted and forbidden metal transitions. The intensity of the Si II λ4128-30 and Mg II λ4481 absorption, combined with the relative weakness of the He I lines at, e.g., λλ4009, 4026, argue for an approximate classification of B8 (particularly since we expect that the He I λ4471 line is partly filled in by nebular contamination). Radial velocity (vrad) estimates were obtained by profile fitting the He I lines at λλ4009, 4026 and 4143 and the Mg II λ4481 doublet. A mean value of 262 ± 7 km s-1 was obtained. No evidence for binarity was found from peak-to-peak vrad variations.

The estimation of atmospheric parameters is difficult because of the emission line spectrum that implies that a non-LTE photospheric model may not be appropriate. However, we have tried to provide constraints using the tlusty model atmospheres grid (Lanz & Hubeny 2007). The Hδ and Hγ profiles provide loci of estimates in the effective temperature and logarithmic gravity plane (Teff, log g). Possible solutions include (12 000 K, 2.6) and (15 000 K, 2.9). Assuming a normal helium abundance, the profile of the He I λ4026 line is consistent with the former. The only well-observed metal line is the Mg II doublet, which leads to an upper magnesium abundance estimate (assuming zero microturbulence) of 6.87 dex for the lower Teff. This is slightly smaller than found by Hunter et al. (2007) for the LMC using the same model atmosphere grid. As a result, Teff is unlikely to be significantly lower than 12 000 K. We adopt Teff = 12 000 ± 3000 K and log g = 2.6 ± 0.4) as our best estimates. For spectra of this quality, the errors would normally be of the order of (±1000 K, ±0.1), but the uncertainties in the appropriateness of the modelling implies that there might be additional uncertainties.

The Hα equivalent width EWHα = −67.5 ± 4 Å. The error reflects the difference obtained by fitting either the wings or the core. The EWHα is greater than maximum measured values for classical Be stars of ≳−50Å (Jones et al. 2011). The Hβ equivalent width EWHβ = −2.72 ± 0.5 Å.

2.2. Photometry

Broad-band photometry for VFTS 822 was adopted from the published catalogues summarised in Table 1. The broadband photometry was not used to estimate extinction, because the emission line contribution to the continuum, although small, cannot be well constrained. We therefore employed equivalent width measurements of the λ4428 and λ6614 diffuse interstellar band features. From van Loon et al. (2013), AV = EW4428/0.7 and AV = EW6614/0.03. This leads to an estimated extinction of AV = 1.75 ± 0.05 (where the observed scatter is not accounted for).

Using an ATLAS9 model (Castelli & Kurucz 2004) having Teff = 12 000 K, log g = 2.6, [M/H] = −0.5, we calculated a bolometric correction, BC = −0.673. We estimate the logarithm of the luminosity, log L = 4.04 ± 0.25 L⊙. The error accounts for the variation in BC due to the uncertainty in Teff. This is most likely an upper limit, since the disc contribution is considered negligible. The luminosity is greater than expected for a mid-late B-type star, but not unprecedented for Herbig stars. For example, HD 85567, V594 Cas, and V921 Sco each have log L≳ 4 L⊙ and are all classified as Herbig B[e] stars (Lamers et al. 1998; Kraus et al. 2012).

2.3. Infrared properties

thumbnail Fig. 2

Observed fluxes in optical (circles) and IR (asterisks). The dashed line is the model SED and the solid line the best-fit YSO model.
Open with DEXTER

The spectral energy distribution (SED) slope (α) in the infrared is an essential tool for diagnosing the nature of any infrared excess. We follow Lada et al. (1987), where α = d log(λFλ)/d logλ and λ > 3 μm. WISE photometry was not used to calculate α because it is possibly affected by nebulosity at λ > 10 μm. We therefore calculated α based on Spitzer photometry alone. For VFTS 822, α = −0.84 ± 0.1, which is greater than α~−3, as expected for normal OB stars (Sung et al. 2009). It is a Class II PMS object based on the scheme devised by Lada et al. (1987). The SED slope closely resembles the one for objects with a dusty disc, λFλ∝λ− 2/3, i.e., a much slower decrease with wavelength than for most evolved stars.

A diagnostic feature of dusty discs is (K − 24 μm) ≳ 5 (e.g. Hernandez et al. 2006). For VFTS 822, Ks − 22 μm ~ 9 mag, indicating the likely presence of a dusty disc. In Fig. 2, we plot the optical-infrared SED for VFTS 822. Overplotted is the extinction corrected adopted ATLAS9 model spectrum. The solid line is the best-fit young stellar object (YSO) model (χ2 = 85) from Robitalle et al. (2007). The SED slope and infrared colours indicate that VFTS 822 has a dusty circumstellar disc characterised by a disc temperature of ~950 K.

3. Evolutionary status

In low-mass stars, emission line spectra and dust discs are unique to the PMS phase. However, at Herbig Be masses, B[e]-type stars also display similar characteristics. B[e]-type stars are rare and heterogeneous in terms of their evolutionary status. They include both PMS objects and evolved supergiants (sg). Based on the spectral type, emission line features, and infrared properties, we consider VFTS 822 to be a B[e]-type star (see Lamers et al. 1998). Identifying the evolutionary status of VFTS 822 is challenging because the observed properties of sgB[e]-type stars and PMS Herbig-B[e] stars that have L~ 4–4.5 L⊙ are similar. Therefore, in the following sections we discuss the possibility that VFTS 822 might be either a PMS B[e] star or an sgB[e] star.

3.1. B[e] supergiant?

The best fit log g (2.6 ± 0.4) is larger than the values measured for many late B-type sg (cf. Firnstein & Przybilla 2012, where log g~ 2 is typical). Owing to the emission lines spectra, abundances are challenging to determine, but we do not see the compelling evidence of chemical evolution that we might expect for an sgB[e] star (e.g. no N II λλ3995, 4630 features are seen). However, the lack of evidence for He enrichment suggests that nitrogen enrichment need not necessarily be present. Therefore, one cannot exclude the possibility that VFTS 822 is a supergiant. One promising avenue forward would be to compare the enrichment of CO13 to CO12 infrared bands (Kraus et al. 2009).

If VFTS 822 is an sgB[e] star, it has an age of 36 Myr, with a lower error bar of 4 Myr based on the single star isochrones of Marigo et al. (2008). This is significantly older than the estimated age of R136, the central cluster of 30 Dor (1–2 Myr), and the high-mass population in the surrounding nebula (e.g. Walborn & Blades 1997). The vrad of VFTS 822 is similar to other stars in the region (Sana et al. 2013), making it unlikely to be an interloper. Nonetheless, the possibility of a genuine 36 Myr old sgB[e] cannot be ruled out because of the complex star-formation scenario and age spread present within 30 Dor. However, the inherent rarity of sgB[e] stars means that a large initial population is expected to yield them (Clark et al. 2013). Such a population, which should be readily identifiable via the resulting IR bright red supergiant/super-AGB cohort, is not known to be present within the immediate vicinity (Bonanos et al. 2009).

thumbnail Fig. 3

H-R diagram showing the position of VFTS 822 (circle), LMC sgB[e] (squares), Galactic PMS B[e] (asterisks), model birthlines (dotted lines), MS (solid line), and PMS (dash-dotted line) isochrones.
Open with DEXTER

3.2. Or Herbig B[e] star?

The position of VFTS 822 in the H-R diagram is compared to known LMC sgB[e] (Zickgraf 2006) and Galactic Herbig B[e] stars (Lamers et al. 1998) in Fig. 3. No Herbig B[e] stars have been discovered yet in the LMC. Birthlines with mass accretion rates (Ṁacc) of 10-4 and 10-3 M⊙ yr-1 from Hosokawa & Omukai (2009) are plotted, along with the zero-age main sequence from Schaerer et al. (1992) and 104.5 yr PMS isochrone from Bressan et al. (2012). The position of VFTS 822 on the H-R diagram is coincident with a low-luminosity LMC sgB[e], but it sits in an H-R diagram position that is also covered by known Galactic Herbig B[e] stars. The comparable positions of sgB[e] and Herbig stars at log L < 4.5 L⊙ make establishing the evolutionary status of VFTS 822 non-trivial.

If VFTS 822 is a PMS, it has a Ṁacc of 3 ± 1 × 10-4 M⊙ yr-1, a mass (M∗) of 10.7 ± 1 M⊙, and age ≃30 000 yr according to the PMS models. PMS stars at such masses have extremely short PMS lifetimes, but a few have been detected (Drew et al. 1997). VFTS 822 is expected to be past its protostellar phase and to be accreting mass. At this age and mass, the Ṁacc of the object is expected to be relatively high, as is necessary to overcome the radiative pressure. Hosokawa & Omukai (2009) predict that at high Ṁacc (>10-4 M⊙ yr-1), massive stars (>10 M⊙) are seen in the PMS phase. We find that the theoretically predicted Ṁacc are similar to the observed accretion rates calculated in Sect. 3.2.1.

The estimated log g of VFTS 822 (2.6 ± 0.4) is somewhat lower than expected for PMS stars. However profile fitting of Herbig stars for log g determination is not common practice, with a recent result showing a Galactic Herbig star to have a log g value of 3.5, similar to giant stars (Ochsendorf et al. 2011). Given the modest extinction toward VFTS 822, this implies a relatively large stellar radius of over 10 R⊙. This may, however, be symptomatic of a bloated star, where very high Ṁacc causes stars to bloat.

VFTS 822 is located near the edge of the 30 Dor nebula at 5.8′ from R136 (approximately 84.4 pc at the distance to the LMC) as shown in Fig. 4. This region contains previously identified loosely grouped Class II PMS objects, which includes the Spitzer counterpart of VFTS 822 (Kim et al. 2007), whereas YSOs or Class I objects are situated nearer to R136. It is located on the edge of filamentary structure, where one might expect to find Class II PMS objects (Hartmann 2008).

3.2.1. Ṁ acc

Balmer line emission, especially Hα, is a commonly used accretion tracer. It is thought to originate in the magnetospheric accretion columns (Calvet & Hartmann 1992). The line luminosity (Lline) can be used to measure the Ṁacc. To calculate Lline from Hα and Hβ, we assume that the excess flux in the continuum is negligible. This assumption is correct if there is no significant veiling. Since photospheric absorption lines are visible, we assume this is justified. The line flux (Fline) is given by (1)Here Fcont is the continuum flux. The continuum flux was measured using the adopted ATLAS9 model spectra. The average flux of the model over the 6556–6570 Åand 4856–4864 Å ranges was taken as the Hα and Hβ continuum flux, respectively. Multiplying by 4πd2, we obtain the respective Lline. Based on the accretion luminosity (Lacc)–Lline relation of Herczeg & Hillenbrand (2008), we determine log Lacc = 3.89 ± 0.6 L⊙ from Hα and log Lacc = 3.58 ± 0.6 L⊙ from Hβ. The errors include the RMS scatter in the Lacc–Lline relations.

Another commonly used accretion indicator is the Balmer jump. The excess luminosity is thought to be caused by an accretion shock on the stellar surface. The UV excess of VFTS 822, (U − I) − (U − I)0 = −0.742, can be used to measure its accretion rate. The observed (U − I) colour was corrected for extinction, and the (U − I)0 model colour was determined using the model spectrum and appropriate filter responses. The excess U-band luminosity, log L(U,excess) = 2.96 ± 0.3 L⊙, and the accretion luminosity Lacc = 3.96 ± 0.5 L⊙. The Ṁacc is given by the free-fall equation. Following the assumptions of Herczeg & Hillenbrand (2008), (2)Here Ṁacc is in units of M⊙ yr-1. We calculate a median Ṁacc of 3.16 ± 2 × 10-4 M⊙ yr-1 from the three indicators. The Lacc–Lline, Lacc–LU,excess relations adopted are tested well on Z⊙ PMS stars, up to Herbig Ae masses (≲4 M⊙). We therefore acknowledge that they may not be numerically applicable to VFTS 822, but only provide an order-of-magnitude estimate. The applied magnetospheric accretion model has been considered valid for spectral types up to late B, but not for the majority of Herbig Be stars (Mottram et al. 2007; Monnier et al. 2005).

The agreement between the Ṁacc calculated from the Hα, Hβ, U-band luminosities suggests that the Balmer lines and jump are produced in the same mass accretion phenomenon, favouring a PMS evolutionary phase.

4. Conclusions

VFTS 822 is a B8 star with estimated parameters of Teff = 12 000 ± 3000 K, log g = 2.6 ± 0.4, and log L = 4.04 ± 0.25 L⊙. Both Hα and Hβ are in emission, along with forbidden and

permitted metal lines. A UV and infrared excess is found when compared to model colours. It is considered to exhibit the B[e] phenomenon. Therefore, its location in the H-R diagram does not clarify its evolutionary status. Thus, we suggest that: (i) VFTS 822 may be an sgB[e] star, around 35 Myr, which we consider quite unlikely (see Sect. 3.1); (ii) alternatively, VFTS 822 could be a Herbig B[e] PMS star with an age of ≈30 000 years, a mass of 10 M⊙, and a Ṁacc ~ 10-4 M⊙ yr-1.

The lack of evolutionary abundance enrichment and a UV excess favour a PMS rather than post main sequence star. The low log g may be explained as the result of a bloated radius due to the large Ṁacc. Becasue no similar PMS stars have been discovered, comparing properties is not feasible at this point. At the measured and predicted Ṁacc, VFTS 822 is expected to have a short PMS lifetime. While the probability of observing a true PMS star in this very small gap of time is serendipitous, our observations suggest that VFTS 822 is a genuine PMS candidate.

We conclude that if VFTS 822 is a PMS B[e] star, it is a rare object and worthy of future study. Further data is required to clarify its evolutionary status.

References

Online material

thumbnail Fig. 4

Spitzer 8 μm image showing the position of VFTS 822 (square) with respect to R136 (cross). North is up and east is to the left. Circles and diamonds are known Class I and Class II sources, respectively.
Open with DEXTER

Table 1

Adopted photometry for VFTS 822.

All Tables

Table 1

Adopted photometry for VFTS 822.

In the text

All Figures

thumbnail Fig. 1

Median FLAMES-Giraffe spectra of VFTS 822.
Open with DEXTER
In the text
thumbnail Fig. 2

Observed fluxes in optical (circles) and IR (asterisks). The dashed line is the model SED and the solid line the best-fit YSO model.
Open with DEXTER
In the text
thumbnail Fig. 3

H-R diagram showing the position of VFTS 822 (circle), LMC sgB[e] (squares), Galactic PMS B[e] (asterisks), model birthlines (dotted lines), MS (solid line), and PMS (dash-dotted line) isochrones.
Open with DEXTER
In the text
thumbnail Fig. 4

Spitzer 8 μm image showing the position of VFTS 822 (square) with respect to R136 (cross). North is up and east is to the left. Circles and diamonds are known Class I and Class II sources, respectively.
Open with DEXTER
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.