A FEROS spectroscopic study of the extreme O supergiant He 3-759
aa r X i v : . [ a s t r o - ph . S R ] J u l Astronomy&Astrophysicsmanuscript no. 12631 c (cid:13)
ESO 2018October 10, 2018
A FEROS spectroscopic study of the extreme O supergiantHe 3–759 ⋆ (Research Note) P. A. Crowther & C. J. Evans Department of Physics & Astronomy, Hicks Building, University of Sheffield, Hounsfield Road, Sheffield, S3 7RH, UK UK Astronomy Technology Centre, Royal Observatory Edinburgh, Blackford Hill, Edinburgh, EH9 3HJ, UK
ABSTRACT
We present a study of the extreme O-type supergiant He 3–759 using new high-resolution FEROS data, revealing that it is a nearspectroscopic twin of HD 151804 (O8 Iaf). We investigate the extinction towards He 3–759 using a variety of methods, revealing A V ∼ . m . If we assume He 3–759 has an identical absolute K -band magnitude to HD 151804 we find that it lies in the Sagittarius-Carina spiral arm at a distance of ∼ T ∗ = 30.5 kK, log L/L ⊙ = 5.9 and ˙ M = 10 − . M ⊙ yr − for a clumped wind whose terminal velocity is estimated at 1000 km s − . The at-mosphere of He 3–759 is enriched in helium ( X He = 49%) and nitrogen ( X N = 0.3%). A reanalysis of HD 151804 and HD 152408(WN9ha) reveals similar parameters except that the WN9ha star possesses a stronger wind and reduced surface hydrogen content.HD 151804 and HD 152408 lie within the Sco OB1 association, with initial masses of ∼ M ⊙ and ages ∼ . Myr, consistent withNGC 6231 cluster members using standard Geneva isochrones. Improved agreement with observed surface abundances are obtainedfor similar initial masses with more recent Geneva group predictions from which higher ages of ∼ Key words. stars: early-type – stars: fundamental parameters – stars: individual: He 3–759, HD 151804, HD 152408
1. Introduction
In normal star-forming galaxies, massive O-type stars dominateboth the Lyman continuum ionizing budget and the feedback ofmechanical energy through their intense stellar winds and, ulti-mately, as core-collapse supernovae. The bulk of their short (3–10 Myr) lives are spent on the main sequence as an unevolveddwarf or giant, before rapidly shedding their hydrogen envelopeduring either the Red Supergiant, Luminous Blue Variable orWolf-Rayet phase. O-type supergiants represent the transitionbetween these stages for the highest mass stars, with character-istic emission lines of He II λ α due to their relativelystrong stellar winds. N III λλ α surveys such asIPHAS (Drew et al. 2005; Witham et al. 2008) and VPHAS+(Arnaboldi et al. 2007) are in the process of remedying thisdeficit, at least for sources detected optically.Still, many sources from the extensive Michigan-Mt Wilsonsouthern H α survey (Henize 1976) remain largely neglected.He 3–759 is one such source, and is the focus of the presentstudy. This was first reported in the catalogue of Galactic Wolf-Rayet (WR) stars by Roberts (1962, Star Send offprint requests to : Paul.Crowther@sheffield.ac.uk ⋆ Based on observations made with ESO telescopes at the La Sillaobservatory under program ID 082-D.0136
Table 1.
Published coordinates of He 3–759, including astrome-try from Tycho-2. α (J2000) δ (J2000) Reference12 11.3 –62 29 Roberts (1962); Henize (1976)12 11.3 –62 30 Carlson & Henize (1979)12 12 08.56 –62 29 00.6 Th´e et al. (1994)12 12 08.6 –62 29 01 de Winter et al. (2001)12 11 18.54 –62 29 43.5 Tycho-2 H α emission. The intensity and sharpness of H β , He II λ III λλ
2. Observations
Previously unpublished spectroscopy of He 3–759 was ob-tained with the Double Beam Spectrograph (DBS) mountedat the Australian National University (ANU) 2.3m tele-
P. A. Crowther & C. J. Evans: FEROS spectroscopy of He 3–759 (RN) scope in April 1996. Subsequent high-resolution spectroscopywas obtained with the Fibre-fed Extended Range OpticalSpectrograph (FEROS) at the 2.2-m Max Planck Gesellschaft(MPG)/European Southern Observatory (ESO) telescope inMarch 2009.
The new observations highlighted discrepancies in previouslypublished coordinates of He 3–759. These are summarised inTable 1, with coordinates precessed to J2000 epochs using the
STARLINK COCO package where necessary. Note that the coor-dinates listed by the
SIMBAD database from Th´e et al. (1994)are incorrect. The positions published by de Winter et al. (2001)are presumably rounded values from Th´e et al.FEROS observations of He 3–759 were initially attemptedon 2009 March 18 using the SIMBAD coordinates. Howeverthe resulting spectrum was, surprisingly, of a cool M-typestar, namely the long period variable IRAS 12094–6212 fromCaldwell et al. (1991). We subsequently inspected the fits headerinformation from the ANU/DBS spectroscopy, which were con-sistent with Carlson & Henize (1979) values, although 5 . ′ We used DBS at the ANU 2.3m telescope to obtain blue, yel-low and red spectroscopy of He 3–759 on 1996 Apr 1–3. Thedetector for both arms of DBS were 1752 ×
532 pix SITE CCDswith blue and red 1200 l mm − gratings providing a dispersionof 0.5 ˚A pix − . The blue DBS arm was used on 1 Apr to ob-tain 1 ˚A (2 pix) resolution spectroscopy of λ λ λ ∼ ∼ m is estimated. Two 1800s exposures of He 3–759 were obtained with FEROSon the nights of 2009 March 19 & 20. FEROS is a cross-dispersed, fixed configuration instrument (Kaufer et al. 1999),which delivers R = 48 , on the 2.2-m, with continuous spec-tral coverage of ∼ The blue and green-red spectral regions of He 3–759 arepresented in Figures 1 and 2, respectively. Also shown arehigh-resolution spectra from Crowther & Bohannan (1997) ofHD 151804, HD 152408, and HDE 313846 from the Anglo-Australian Telescope (AAT) using the University CollegeLondon Echelle Spectrograph (UCLES). HD 151804 is a ‘nor- http://simbad.u-strasbg.fr/simbad/ mal’ Of star, classified as O8 Iaf (Conti & Alschuler 1971;Walborn 1972), with the more extreme sources HD 152408and HDE 313846 (WR108) reflected by their classification ofWN9ha (Crowther & Bohannan 1997; Bohannan & Crowther1999).These spectra illustrate an elegant morphological sequencein terms of increasing emission-line intensities. The N III , He II and H β emission in He 3–759 is slightly stronger than inHD 151804, but otherwise their blue (stellar) spectra are verysimilar (Figure 1), including the diagnostic He I λ II λ and we therefore adopt an O8 Iaf classifica-tion for He 3–759. Its appearance confirms the descriptionof Carlson & Henize (1979), with He 3–759 and HD 151804among the rare subset of O stars in which H β is observed inemission, signifying extreme mass-loss properties.The green-red region (Figure 2) reinforces the notion thatHe 3–759 is a slightly more extreme Of star than HD 151804,with stronger H α and He I IV λλ I λ
3. Reddening & Distance
Reliable photometry of He 3–759 is somewhat sparse in the lit-erature so we consider three different approaches to estimateits reddening drawn from (i) infrared photometry from the TwoMicron All Sky Survey (2MASS, Skrutskie et al. 2006); (ii) ul-traviolet International Ultraviolet Explorer (IUE) spectroscopyfrom Shore et al. (1990); (iii) the strength of Diffuse InterstellarBand (DIBs) observed in the FEROS spectroscopy. An estimateof the distance to He 3–759 then follows from comparison withHD 151804 which is a member of the Sco OB1 association (dis-tance modulus 11.4 m , Humphreys 1978). thence its distance, forquantitative analysis. A summary of visible and near-IR photometry for He 3–759 is presented in Table 2. Optical measurements are ratherheterogeneous, including the Tycho-2 (V T = 11.45), 2ndUSNO CCD (10.94) and USNO-B1.0 catalogues (B1=12.35,R1=10.85, I=10.38). We include visual and IR photometryof HD 151804 (O8 Iaf) drawn from Leitherer & Wolf (1984),Crowther & Bohannan (1997) and references therein. Intrinsicnear-IR colours are obtained from our analysis of He 3–759(Sect. 4) from which K s -band extinctions, A K s , may be obtainedusing the extinction relations from Indebetouw et al. (2005). Ourderived extinction of A K s = 0.58 ± . m for He 3–759 corre-sponds to E B − V =1.65, assuming a standard Galactic extinctionlaw. A similar approach for HD 151804 reveals A K s =0.19 m ,or E B − V =0.54 m , rather higher than results derived previously,such as E B − V =0.32 m from Crowther & Bohannan (1997). Alternatively, we can exploit archival IUE ultraviolet spec-troscopy of He 3–759 published by Shore et al. (1990). We havedownloaded low dispersion, large aperture datasets SWP 36664and LWP 15903 (obtained on 12 Jul 1989) from the IUE Newly We measure log 4471 / = 0.19 from our FEROS dataset forHe 3–759 versus 0.22 from our UCLES observations of HD 151804.. A. Crowther & C. J. Evans: FEROS spectroscopy of He 3–759 (RN) Fig. 1.
Blue region FEROS spectrum of He 3–759, compared with the AAT-UCLES spectra of HD 151804, HD 152408 andHDE 313846 from Crowther & Bohannan (1997). Spectral lines identified in HD 151804 are He I λλ IV λλ III λλ λλ ǫ , H δ , H γ and H β Balmer lines; He II λλ λ IV λλ λλ Fig. 2.
Green-red region FEROS spectrum of He 3–759, compared with the AAT-UCLES spectra of HD 151804, HD 152408 andHDE 313846 from Crowther & Bohannan (1997). Emission lines identified in HDE 313846 are C
III λ I λλ α , and Si IV λλ IV λλ II λ λλ P. A. Crowther & C. J. Evans: FEROS spectroscopy of He 3–759 (RN)
Table 2.
Visual (Tycho-2 B T and V T in parenthesis) and near-IR photometry for the O8 Iaf stars He 3–759 and HD 151804, includinga distance estimate to He 3–759. Star V B-V K s J–K s H–K s (J–K s ) (H–K s ) A J − K K s A H − K K s A K s DM M K s He 3–759 (11.45) (1.01) 7.88 0.66 0.32 –0.09 –0.04 0.50 0.65 0.58 14.0 ← –6.7HD 151804 ø5.22 0.07 4.86 ‡ ‡ ‡ –0.09 –0.04 0.19 0.18 0.19 11.4 → –6.7 Note 1. ‡ JHK magnitudes of Leitherer & Wolf (1984) are preferred to 2MASS due to a low quality index in this instance.
Extracted Spectra archive . We have reddened the spectral en-ergy distribution of our He 3–759 model from Sect. 4 and ob-tain an optimum fit to the combined UV spectrophotometryand IR photometry with E B − V =1.4 using a standard R V = A V /E B − V = 3.2 extinction law. This is presented in Figure 3,for which overall agreement is satisfactory, including the com-parison with Spitzer GLIMPSE (Benjamin et al. 2005) at mid-IRwavelengths. Intrinsic colours from our He 3–759 model include(K s – [8.0]) = 0.37 m and ([3.6] – [4.5]) = 0.1 m . We may also exploit the strong DIB features in the visual spec-trum of He 3–759 with respect to other moderately reddenedstars in Figures 1–2, notably λλ λ ± E B − V in excess of 1.0 according toSnow et al. (2002).Weaker DIB lines are seen to correlate reasonably well with E B − V , in particular λλ λ λ λ W λ ) estimates forthese three DIB features are given in Table 3, with uncertain-ties of ±
10% (sufficient for the purposes of the current investi-gation). The average of these three estimates is E B − V =1.46 m .Finally, our FEROS spectroscopy confirms the claim fromCarlson & Henize (1979) that the Ca II H line is broadened, al-beit owing to stellar H ǫ , rather than being of interstellar origin. Table 3.
Equivalent widths ( W λ ) of selected diffuse inter-stellar bands (DIBs) and the resulting estimates of E B − V .Uncertainties on the widths are ±
100 m ˚A for λ ± Line ( ˚A) W λ (m ˚A) E B − V Calibration4428 2500 > The three methods outlined above provide the following esti-mates of A V = R V E B − V . IR photometry results in A V = 5.1(for R V = 3.1), UV spectrophotometry implies A V = 4.5 and the http://sdc.laeff.inta.es/ines/ Fig. 3.
Reddened model spectral energy distribution of He 3–759( E B − V =1.4, R V =3.2) overlaid upon UV (IUE) spectrophotome-try, plus IR photometry from 2MASS (Skrutskie et al. 2006) andGLIMPSE (Benjamin et al. 2005)line strengths of DIB features also suggest A V = 4.5, yielding A V ∼ m or A K s = 0.53 m . If we assume that He 3–759 has asimilar absolute K s -band magnitude to HD 151804 (O8 Iaf) wemay estimate its distance.HD 151804 is a member of Sco OB1 (distance 1.9 kpc,Humphreys 1978) from which M K s = − m is obtained(Table 2), giving a distance modulus of 14.06 ± m or distanceof 6.5 +1 . − . kpc for He 3–759. For an adopted Solar galactocen-tric distance of 8.0 kpc (Reid 1993), He 3–759 would lie in theSagittarius-Carina arm, close to the Solar circle ∼ +0 . − . kpcfrom the Galactic Centre.
4. Physical and Wind Parameters
We have derived the physical and wind properties of He 3–759using CMFGEN (Hillier & Miller 1998), and re-analysed opticalspectroscopy of HD 151804 (O8 Iaf) and HD 152408 (WN9ha)from Crowther & Bohannan (1997) for comparison.
CMFGEN solves the radiative transfer equation in the co-moving frame, under the additional constraint of statistical equi- . A. Crowther & C. J. Evans: FEROS spectroscopy of He 3–759 (RN) Fig. 4.
Spectroscopic fits (dotted, red) to FEROS observations(solid, black) of He 3–759.librium. The temperature structure is determined by radiativeequilibrium. Since CMFGEN does not solve the momentumequation, a density or velocity structure is required. For the su-personic part, the velocity is parameterized with a classical β -type law, with an exponent derived from fits to H α . This is con-nected to a hydrostatic density structure at depth, such that thevelocity and velocity gradient match at the interface. The sub-sonic density structure is set by a corresponding log g = 3 . fully line-blanketed plane-parallel TLUSTY model (v.200, seeLanz & Hubeny 2003). The atomic model is similar to thatadopted by Crowther et al. (2002), including ions from H, He,C, N, O, Si, P, S and Fe.We have assumed a depth-independent Doppler profile forall lines when solving for the atmospheric structure in the co-moving frame, while in the final calculation of the emergentspectrum in the observer’s frame, we have adopted a uniform tur-bulence of 50 km s − . Incoherent electron scattering and Starkbroadening for hydrogen and helium lines are adopted. Finally,we convolve our synthetic spectrum with a rotational broaden-ing profile for which v sin i ∼
100 km s − . Clumping is incorpo-rated using a volume filling factor, f , as described in Hillier et al.(2003), with a typical value of f =0.1 resulting in a reduction inmass-loss rate by a factor of p (1 /f ) ∼ . Table 4.
Physical and wind properties of He 3–759 with respectto HD 151804 and HD 152408, allowing for an uncertainty in ab-solute magnitude of ± m . Clumped mass-loss rates are quotedhere for volume filling factors of f =0.1. Star He 3–759 HD 151804 HD 152408O8 Iaf O8 Iaf WN9ha T eff (kK) 30.5 29.0 31.8 R ∗ ( R ⊙ ) 32.1 +8 . − . T / (kK) 29.3 28.1 31.3 log L/L ⊙ ± v sin i (km s − ) 100 104 80 v ∞ (km s − ) 1000 1445 970 ˙ M ( M ⊙ yr − ) 10 − . ± . − . − . X H (%) 49 43 27 X He (%) 49 56 72 X N (%) 0.3 0.25 0.6 M K s (mag) –6.7 ± Note 2.
Formal uncertainties in T eff are ± ±
5% (H and He) or a factor of two (N)
We derive the stellar temperature of He 3–759 using diagnosticHe I λ λ II λ λ α and H β for the mass-loss rate and velocity structure. We haveestimated a terminal wind velocity of 1000 ±
300 km s − basedupon low-resolution IUE observations of C IV λ β =2 is used for the supersonic velocity structure is usedsince this provides an excellent fit to the H α profile. Regardingwind clumping in Of supergiants, either He II λ α aresuitable for determination of the volume filling factor f , if thevelocity law is known. However, since H α is used to estimate thevelocity law and the peak emission of He II λ f is not achievable.Spectroscopic fits to FEROS observations are presented inFigure 4, with a summary of physical and wind parameters pre-sented in Table 4. Overall, the fits are satisfactory, with the ex-ception of He II λ I λ I λ λ I lines in O stars, such that triplets (e.g. λ λ X He = 49% by mass. The promi-nent N III λλ λλ X N = 0.3%, correspond-ing to an enrichment of 4 times the solar value. However, N III λ III λ III λ IV λλ λλ X C = 0.2% (0.7 timesthe solar case), although large uncertainties are admitted. Foroxygen, solely O III λ X O = 0.2% (0.5 times the solar value). For silicon, sulphur andiron we adopt solar values.We have also reanalysed two of the reference stars –HD 151804 (O8 Iaf) and HD 152408 (WN9ha) – based upon our P. A. Crowther & C. J. Evans: FEROS spectroscopy of He 3–759 (RN)
AAT UCLES datasets presented in Figs 1–2 and the methodoutlined above. A TLUSTY log g = 3 . model at depth wasadopted for HD 152408 since log g = 3 . models were not avail-able for T eff = 32.5kK. For current stellar masses of ∼ M ⊙ (see Sect. 4.3), surface gravities are log g ∼ . , while effectivegravities, corrected for radiation pressure, are log g eff ∼ . . Fits are of comparable quality to those presented here forHe 3–759, also failing to reproduce He II λ A comparison between the physical properties of He 3–759 and non-rotating, solar metallicity Geneva models fromMeynet et al. (1994, see also Lejeune & Schaerer 2001) suggestsan age of 2.7 Myr and initial mass of ∼ M ⊙ . Similar resultsare obtained for HD 151804 and HD 152408, in good agreementwith the age of the NGC 6231 cluster within Sco OB1, as de-rived by Crowther et al. (2006) using the same set of isochrones.However, these standard evolutionary models are well knownnot to predict the observed helium enrichment at such phases.In contrast, comparisons with the evolutionary models ofMeynet & Maeder (2000) allowing for rotation and contempo-rary mass-loss rate prescriptions enable reasonable matches toboth the surface hydrogen abundance ( ∼ %) and location inthe H-R diagram. For a distance of 6.5 kpc to He 3–759, ini-tial 60 M ⊙ models rotating at 300 km s − suggest a greater ageof 3.9 Myr, while a slightly lower age of 3.6 Myr is obtainedfor a non-rotating 60 M ⊙ model. At these ages, current stellarmasses lie in the range 35–45 M ⊙ , from which we adopt 40 M ⊙ for surface gravity estimates. Lower mass evolutionary modelsfrom Meynet & Maeder (2000) fail to predict the combinationof surface hydrogen content and its position in the H-R diagram,favouring our preferred distance to He 3–759.In summary, He 3–759 appears to be a very high mass star ata relatively young age, but unlike HD 151804 and HD 152408it does not reside within a known cluster or OB association.According to Larson (2003), the most massive star of a clus-ter (of mass M clu ) scales with cluster mass according to 1.2 M . suggesting a lower limit of ∼ M ⊙ for its birth clus-ter. He 3–759 does not possess a high radial velocity so it wouldbe expected to be located close to its natal cluster. Alternatively,Parker & Goodwin (2007) have proposed that some massivestars may form in relatively low mass clusters. Such clusterswould not necessarily be easily identified at large distances, asis the case for He 3–759.
5. Summary
We have presented a high quality FEROS spectrum of the poorlystudied, early-type emission line supergiant He 3–759, from The Eddington parameter – the ratio of radiation pressure to gravity– is Γ e ∼ which an O8 Iaf classification is obtained, and clarified its co-ordinates. We have used three methods to estimate its high in-terstellar extinction, namely fitting a stellar model to its IUEultraviolet spectrophotometry and 2MASS and GLIMPSE pho-tometry; obtaining its near-IR extinction from comparison withintrinsic colours; deriving its visual extinction from measuredstrengths of DIBs. Combining these approaches implies A V =4.7 m or A K s = 0.53 m . If we assume that He 3–759 has a similarabsolute K s -band magnitude to HD 151804 (O8 Iaf) its distanceis estimated as 6.5 kpc, within the Sagittarius-Carina arm. Thepresence of such a high-mass ( ∼ M ⊙ ) star in isolation is curi-ous given the lack of a nearby cluster, which would be expectedto be relatively massive ( ≥ M ⊙ ).No doubt, many other emission-line OB supergiants awaitdiscovery, in view of large optical surveys such as IPHAS andVPHAS+. Alternatively, visibly obscured extreme early-type su-pergiants may be identified by their infrared free-free excess fol-lowing the approach of Hadfield et al. (2007). Acknowledgements.
We thank John Hillier for his development of CMFGEN,Hugues Sana for his reprocessing of the data, and Martin Cordiner and Keith T.Smith for helpful discussion regarding the interstellar features. This publicationis based in part upon INES data from the IUE satellite, 2MASS which is a jointproject of the University of Massachusetts and the IPAC/CalTech, funded by theNASA and the NSF, and Spitzer datasets from NASA/IPAC Infrared ScienceArchive (IRSA). IRSA is operated by JPL, CalTech under contract with NASA.