# Massive stars in the Sagittarius Dwarf Irregular Galaxy

Massive stars in the Sagittarius Dwarf Irregular Galaxy Abstract Low metallicity massive stars hold the key to interpret numerous processes in the past Universe including re–ionization, starburst galaxies, high-redshift supernovae, and γ-ray bursts. The Sagittarius Dwarf Irregular Galaxy [SagDIG, 12+log(O/H) = 7.37] represents an important landmark in the quest for analogues accessible with 10-m class telescopes. This Letter presents low-resolution spectroscopy executed with the Gran Telescopio Canarias that confirms that SagDIG hosts massive stars. The observations unveiled three OBA-type stars and one red supergiant candidate. Pending confirmation from high-resolution follow-up studies, these could be the most metal-poor massive stars of the Local Group. stars: early-type, stars: massive, galaxies: individual: SagDIG, galaxies: stellar content 1 INTRODUCTION Understanding the evolution of very metal-poor massive stars is key to interpret star formation and feedback at previous Cosmic epochs, and a necessary step towards the physics of the first stars of the Universe. This has stimulated an enthusiastic effort to discover resolved massive stars in environments of decreasing metal content, mostly dwarf irregular galaxies of the Local Group. The Small Magellanic Cloud (SMC) is the current standard for the low-Z regime, but metal-poorer massive stars are now at reach with multi-object spectrographs at 8–10m telescopes. First quantitative analyses were directed to the IC 1613, NGC 3109, and WLM galaxies (Bresolin et al. 2007; Herrero et al. 2010; Tramper et al. 2011) with encouragingly low 1/7O⊙  oxygen abundances, but subsequent studies indicate that their iron content is SMC-like (∼1/5Fe⊙) instead (Garcia et al. 2014; Hosek et al. 2014; Bouret et al. 2015). Following efforts were directed to Sextans A (Camacho et al. 2016) with both 1/10Fe⊙ and 1/10O⊙ (Kaufer et al. 2004). However, this value is still far from the ∼1/30 Z⊙ average metallicity of the Universe at the peak of star formation history and from measurements of the intergalactic medium at higher redshifts (Madau & Dickinson 2014). The Sagittarius Dwarf Irregular Galaxy (aka Sgr dIG, ESO 594-4, UKS 1927-17.7; SagDIG from now on) makes a promising environment to find very metal-poor, resolved massive stars. Its intermediate-age population stands out as one of the most Fe-poor among the Local Group galaxies that sustain star formation (e.g. McConnachie 2012), with values inferred from the location of the red giant branch (RGB) that range from [Fe/H] = −2.0 to −2.45 (Beccari et al. 2014; Karachentsev, Aparicio & Makarova 1999). Regarding the youngest populations, the oxygen abundance determined from the only confirmed H  ii region SagDIG-H ii#3 (Strobel, Hodge & Kennicutt 1991) is also remarkably low 12+log(O/H) = 7.37$$\rm ^{ + 0.13}_{ - 0.11}$$ (Saviane et al. 2002). Scaled by the Solar photospheric abundance (Asplund et al. 2009), this value is equivalent to 1/21O⊙. While SagDIG is considered one of the extremely metal-poor galaxies of the Local Universe, rivaling I Zw18 (Filho et al. 2015), it has never been explored for massive stars. This Letter presents the first spectroscopic observations of young massive stars in the galaxy. Section 2 describes the observations and data reduction. Target identification, spectral types (SpT), and membership are assessed in Section 3. The discovered red supergiant (RSG) candidate is discussed in the context of the Local Group in Section 4. Finally, summary and some prospects for future work are provided in Section 5. 2 DATA AND REDUCTION The observations were executed with the OSIRIS spectrograph at the 10.4-m Gran Telescopio Canarias (GTC), programme GTC67-13A (PI M. Garcia). The 1.2 arcsec slit and R2000B VPH were used to achieve resolution R ∼ 1000 at ∼4000–5500 Å. The slit was aligned on targets SD-1 and SD-7 to cross the galaxy (Fig. 1) and enclose several ultraviolet (UV)-bright sources. A total of 2 h of grey sky, 1.2 arcsec seeing exposure time were devoted to the slit. Figure 1. View largeDownload slide SagDIG: GTC–OSIRIS acquisition image (1.2 arcsec seeing, filter–free), and programme stars. Figure 1. View largeDownload slide SagDIG: GTC–OSIRIS acquisition image (1.2 arcsec seeing, filter–free), and programme stars. The data were reduced following standard iraf1 procedures adapted for the team’s OSIRIS runs (Garcia & Herrero 2013). The spectra were flux–calibrated against the flux standard GD190, achieving about 5 per cent accuracy for λ > 4250 Å. Finally, the spectra were set to the heliocentric frame and corrected by radial velocity. The latter were calculated from the Doppler shift measured at the core of the Balmer lines. The reduced spectra are shown in Figs 2–3. Figure 2. View largeDownload slide GTC–OSIRIS spectra of blue massive stars in SagDIG. The data have been rebinned to three pixels for clarity, and corrected by heliocentric and radial velocity. The three spectra show a clear Balmer series. Figure 2. View largeDownload slide GTC–OSIRIS spectra of blue massive stars in SagDIG. The data have been rebinned to three pixels for clarity, and corrected by heliocentric and radial velocity. The three spectra show a clear Balmer series. Figure 3. View largeDownload slide Red programme stars. The flux-calibrated spectra were smoothed to five points, and corrected by heliocentric and radial velocity. In both panels the upper spectrum has been shifted by a constant for clarity (SD-4: 1.5E − 17 erg cm − 2 s − 1  Å − 1, SD-6: 3E − 17 erg cm − 2 s − 1  Å − 1). Figure 3. View largeDownload slide Red programme stars. The flux-calibrated spectra were smoothed to five points, and corrected by heliocentric and radial velocity. In both panels the upper spectrum has been shifted by a constant for clarity (SD-4: 1.5E − 17 erg cm − 2 s − 1  Å − 1, SD-6: 3E − 17 erg cm − 2 s − 1  Å − 1). The programme faced the challenges related to observations of faint, distant stars: contamination by nearby sources and small number of counts over the background level. For instance the spectrum of SD-1, with V ∼ 20.2, registered a factor ∼2.5 less counts than the extracted sky spectrum. The resulting signal to noise ratio (S/N) is consequently poor even after careful sky subtraction and some solar features remained. These shortcomings discourage future optical range spectroscopic observations of targets fainter than V = 20 mag in grey time. Targets SD-1 to 4 are found in an area of extended continuum emission (detected in far-UV imaging as well, e.g. Momany et al. 2014) that may contaminate the extracted spectra. In order to minimize blends, the sources were extracted discarding the wings of the spatial distribution of photons in the 2-D images. 3 SOURCE IDENTIFICATION, SPECTRAL CLASSIFICATION AND MEMBERSHIP SagDIG is projected close to the Galactic Centre and the line of sight is severely contaminated with Milky Way stars. Hence, ensuring that the programme targets belong to the galaxy was an issue. This section discusses jointly target identification, spectral classification, membership, and their consistency. The location of the programme stars in the colour–magnitude diagram (CMD) is evaluated as a final check. The programme targets were identified in the Beccari et al. (2014, hereafter B14) and Momany et al. (2014, M14) catalogues by comparing the GTC acquisition images against HST–ACS archival observations (GO-9820 P.I. Y. Momany). Identification numbers and photometry are listed in Table 1. Table 1. Observed stars: SpT from this work, radial velocities (vrad) used in Figs 2–3, and cross-identification numbers and photometry from M14 and B14. MV was computed from apparent magnitudes, distance modulus (SagDIG stars: DM = 25.10 Momany et al. (2005); foreground stars: DM = 14.50), and Cox (2000)’s intrinsic colours to estimate E(V − I) and correct from extinction. The last column flags SagDIG membership. ID  SpT  vrad  M14  F475W  F606W  F814W  B14  V  I  E(V − I)  MV  Membership      (km s − 1)  ID        ID            SD-1  Late–O I  − 65  33466  20.189  20.125  20.024  8903  20.190  20.154  0.51  − 6.1  yes  SD-2  BA I  − 55  31722  20.355  20.278  20.120  b  20.38  20.24  0.12  − 5.1  yes  SD-3  OB V  − 95  32939  21.408  21.480  21.515  8822  21.442  21.585  0.28  − 4.3  yes  SD-4  G0–2 I  − 100  30481  21.558  20.503  19.394  8805  20.731  19.473  0.42  − 5.3  yes  SD-5  G4 III  − 90  3019  20.887  20.055  19.144  b  20.38  19.23  0.26  5.3  fg  SD-6  G5 V  − 70  a    18.59  17.82  7783  18.875  17.912  0.07  4.2  fg  SD-7  K4 V  − 50  a    18.76  17.74  8017  19.124  17.819  0.42  3.7  fga  ID  SpT  vrad  M14  F475W  F606W  F814W  B14  V  I  E(V − I)  MV  Membership      (km s − 1)  ID        ID            SD-1  Late–O I  − 65  33466  20.189  20.125  20.024  8903  20.190  20.154  0.51  − 6.1  yes  SD-2  BA I  − 55  31722  20.355  20.278  20.120  b  20.38  20.24  0.12  − 5.1  yes  SD-3  OB V  − 95  32939  21.408  21.480  21.515  8822  21.442  21.585  0.28  − 4.3  yes  SD-4  G0–2 I  − 100  30481  21.558  20.503  19.394  8805  20.731  19.473  0.42  − 5.3  yes  SD-5  G4 III  − 90  3019  20.887  20.055  19.144  b  20.38  19.23  0.26  5.3  fg  SD-6  G5 V  − 70  a    18.59  17.82  7783  18.875  17.912  0.07  4.2  fg  SD-7  K4 V  − 50  a    18.76  17.74  8017  19.124  17.819  0.42  3.7  fga  aStar without M14-counterpart. F606W and F814W magnitudes estimated from B14’s V- and I-magnitudes, using Rejkuba et al. (2005)’s transformations. bStar without B14-counterpart. VI-magnitudes estimated from M14, using Rejkuba et al. (2005)’s transformations. View Large The spectral classification of blue stars, whose most prominent spectral feature is the Balmer series, was carried out following classical criteria (see e.g. Garcia & Herrero 2013). Their spectra are shown in Fig. 2. The rest of the sample (Fig. 3) was classified after Dorda (2016)’s scheme for cool stars (G-types and later) in the Large Magellanic Cloud (LMC) and the SMC, and the spectral classification atlas by Gray (2000).2 They were also compared against SMC and MW standards. Membership of the programme stars with counterpart in M14’s catalogue, that registered relative shifts from multi-epoch HST visits, was ensured by the lack of proper motions. In addition, the estimated absolute magnitudes were compared against the calibrated values for the SpT assigned in this work. The consistency of the stellar radial velocities and SagDIG’s systemic velocity (−78.5 km s − 1, McConnachie 2012) was not determinant because of the low spectral resolution of these observations (∼300 km s − 1, hence typical vrad  uncertainties of ∼100 km s − 1). The following subsections provide more details on the targets that were positively identified as SagDIG massive stars. 3.1 SD-1 The HST–ACS images reveal that SD-1 is the blend of two stars ∼ 1.1 arcsec apart, SD-1-blue (M14-33466; B14-8903, V = 20.190) and SD-1-red (B14-8909, V = 21.774). SD-1 will be identified with the brightest star of the blend, SD-1-blue, from now on. The Balmer series is clearly detected in its spectrum (Fig. 2), and the comparison against the sky spectrum confirms that the stellar component dominates Hγand Hβ. The data also display He i+He ii 4026 Å, He i 4387 Å, and He i 4471 Å, but the detection of He ii lines is unclear. He ii 4541 Å is absent and the He ii 4200 Å and He ii 5411 Å absorptions overlap with sky features. The exception is He ii 4686 Å, detected in emission. Its width together with the absence of H i and He i emission lines renders a nebular origin unlikely, and the comparison against the subtracted sky spectrum discards the feature as an artefact. The He ii 4686 Å emission could be consistent with a late-O type, with the rest of He ii lines too weak for detection at the S/N of the spectrum. The non-detection of Mg ii 4481 Å and Si iii 4552 Å supports the late-O classification, although it could also be caused by the low metallicity of the star. The spectrum of SD-1 is reminiscent of star C1_31 in NGC 55 (Castro et al. 2008). C1_31 is mostly featureless except for the Balmer lines and a prominent broad He ii 4686 Å emission, with He i 4471 Å clearly seen but no He ii 4541 Å. It was classified as early-O supergiant. Similarly, SD-1 is classified as late-O I, with the He ii 4686 Å emission and narrow Balmer lines supporting the supergiant luminosity class. The estimated absolute magnitude MV = −6.1 agrees with the calibrated value for late-O supergiants (Massey 1998). 3.2 SD-2 SD-2 is the blend of M14-31722 (F606W = 20.278) and M14-31570 (F606W = 21.839), 0.7 arcsec apart and unresolved. SD-2 is identified with the brightest source M14-31722. Its spectrum shows the Balmer series, He i 4471 Å, He i 4387 Å, and He i 4922 Å (Fig. 2). There may be some Mg ii 4481 Å, but no lines of Si iii, Si iv, or He ii. The calibrated stellar flux corresponds to a colder object than SD-1. Since the width of the Balmer lines is similar to SD-1, and the strength of the He i lines is almost at noise level, SD-2 is classified as a BA supergiant. The estimated absolute magnitude MV  = −5.1 is consistent with a blue supergiant located at SagDIG. 3.3 SD-3 Target SD-3 is faint and off-slit, resulting in extremely low spectral S/N. The sky contribution amounts to 83 per cent of the total flux at blue wavelengths, increasing to 91 per cent at 4800 Å. SD-3 overlaps with the red component of the SD-1 blend. However, there is a shift between SD-3’s strongest features and the same lines in SD-1 reflecting the different radial velocities of the two stars and confirming that the lines genuinely belong to SD-3. The spectrum of SD-3 (Fig. 2) exhibits the Balmer series and He i 4471 Å. The He ii lines, but also Si iii 4552 Å and Mg ii 4481 Å, are absent suggesting late-O or early-B SpT. The spectrum lacks any He ii 4686 Å, but contamination by SD-1-red begins to be significant in this region. The wings of Hγ  are dominated by the stellar component, and the profile is wider than SD-1’s, suggesting class III or V. Since the absolute magnitude agrees with Massey (1998)’s calibration for O9–B0 V (MV = −4.4 to −3.8), SD-3 is classified as an OB dwarf. 3.4 SD-4 SD-4 (M14-30481, F606W = 20.503) is contaminated by a nearby red star with similar magnitude (M14-30025, F606W = 20.360). According to their proper motions, both belong to SagDIG. Luminosity class was constrained in the first place, following Dorda (2016). Because the 5167 Å component is the strongest of the magnesium triplet Mg i 5167, 5172, 5184 Å, SD-4 was classified as a supergiant. The relative strength of Fe i 5247, 5250, 5255 Å versus Ca i+Fe i+Ti i 5270 Å, and Mn i 5433 Å versus Mn i 5447 Å support this luminosity class. The presence of the G band and its relative strength compared to Hγ  and Ca i 4227 Å suggests that the SpT is slightly later than G0. The relative strength of Fe i 4325 Å compared to both Hγand Fe i 4383 Å, on the other hand, suggests spectral type G0 or earlier. A G0–2 I type is finally assigned. It is noted that the estimated absolute magnitude MV  = −5.3 is slighlty fainter than the calibrated MV  = −6.4 for a G0 ∼ I star. Nevertheless, SD-4’s membership to SagDIG is supported by the lack of detection of proper motions by M14. Since RSGs stars may experience spectral variability (see Section 4), SD-4 is proposed as a RSG-candidate in SagDIG. 3.5 Target stars in the colour–magnitude diagram The location of the target stars in the CMD is consistent with the SpT assigned in this work (Fig. 4). The three OBA stars are found in the upper part of the blue plume. The red stars SD-5, SD-6, and SD-7 are enclosed in the vertical sequence of Galactic foreground contamination. SD-4, on the other hand, is above the tip of the RGB consistently with the star being a RSG. Figure 4. View largeDownload slide SagDIG’s CMD built from M14’s catalogue (grey dots). Marigo et al. (2017)’s isochrones for Z = 0.001 (∼1/15 Z⊙) are included for comparison, with colour coding log (age[yr]) = 6.0, 6.49, 7.0, 7.49, 8.0. Younger isochrones are not included because they overlap at F606W–F814W ∼ 0 in the F814W = 20–22 range. The isochrones were shifted by SagDIG distance modulus (m − M)0 = 25.10 and reddened by the equivalent to E(B − V) = 0.22 (AF606W = 0.646, AF814W = 0.410, after Sirianni et al. 2005). Figure 4. View largeDownload slide SagDIG’s CMD built from M14’s catalogue (grey dots). Marigo et al. (2017)’s isochrones for Z = 0.001 (∼1/15 Z⊙) are included for comparison, with colour coding log (age[yr]) = 6.0, 6.49, 7.0, 7.49, 8.0. Younger isochrones are not included because they overlap at F606W–F814W ∼ 0 in the F814W = 20–22 range. The isochrones were shifted by SagDIG distance modulus (m − M)0 = 25.10 and reddened by the equivalent to E(B − V) = 0.22 (AF606W = 0.646, AF814W = 0.410, after Sirianni et al. 2005). The loci of the youngest stars in the CMD is also consistent with the theoretical isochrones (Z = 0.001 or ∼1/15 Z⊙, Marigo et al. 2017) but require a larger correction than the reported E(B − V)fg  = 0.12 foreground extinction (e.g. Momany et al. 2005). Internal reddening can indeed be expected, since SD-1 to 4 are located in a concentration of neutral hydrogen (Young & Lo 1997, also B14). The E(B − V)  ≃0.22 colour excess calculated from the Balmer decrement of SagDIG-H ii#3 was adopted (Lee, Grebel & Hodge 2003; Tang et al. 2016). SD-3 is located at the main sequence where isochrones of log (age[yr]) = 4.5–7.0 overlap, in accordance with its OB V SpT. SD-2 is found between the log (age[yr]) = 7.0 and 7.49 isochrones, as expected considering that BA supergiants have evolved off the main sequence and are slightly older than O-stars. The location of SD-4 is consistent with a RSG with similar age log (age[yr]) = 7.49. SD-1 is found between the log (age[yr]) = 7.0–7.49 isochrones, however, it is more reddened than the other programme stars. If its E(B − V) = 0.38 spectroscopic colour excess is taken at face value, the isochrones would need an additional shift of Δ(F606W–F814W) ∼ 0.2 mag and ΔF814W ∼ 0.3 mag to account for the increased extinction. This would bring SD-1 close to the log (age[yr]) = 6.49 isochrone (3 Myr), that better agrees with the expected age of an O-type supergiant. The unknown amount of internal reddening in SagDIG hinders the interpretation of the CMD (SD-1,2,3 would be mistaken as intermediate-mass stars if E(B − V)fg was used) and may hide a significant population of young and massive stars. Extinction can explain the paucity of stars in the upper blue plume, which is at odds with the extended emission of the galaxy in the UV. These results may be considered a reminder that the total galactic stellar content may be underestimated from the luminosity function, and advices against selecting massive star candidates from their location in classical CMDs only. 4 THE MOST METAL-POOR RSG TO DATE RSGs are He-burning descendants of Mini  ≲ 30 M⊙ stars and the coolest stage in the life-cycle of massive stars. They are gathering increased interest because some RSGs (9 < Mini<16.5–23M⊙) are the alledged progenitors of Type-IIP supernovae (Groh et al. 2013). Besides, they are proving reliable metallicity probes optimally reached by current, powerful infrared spectrographs: Keck–NIRSPEC, GTC–EMIR, or VLT–KMOS (e.g. Davies et al. 2009; Gazak et al. 2015). RSGs are classically associated with M-types, but the average SpT shifts towards earlier types in metal-poorer galaxies (Elias, Frogel & Humphreys 1985; Levesque & Massey 2012). This trend is beyond a classification bias produced by the weakening of TiO bands when metallicity decreases, and it is believed to reflect the fact that the Hayashi limit shifts towards higher effective temperatures with decreasing metallicity (see discussion by Elias et al. 1985). For this reason, and because sometimes they exhibit spectroscopic variability, types as early as G0 are considered in the class of RSGs (see discussion in González-Fernández et al. 2015). According to the latest compilation of Local Group RSGs by Levesque & Massey (2012), the most frequent spectral SpT shifts from M2 I in the Milky Way and the LMC, to mid-K types in the SMC and to K0-1 in WLM (1/7–1/10 Z⊙). The RSG-candidate SD-4, with SpT G0–2 I, extends the spanned range of metallicity down to Z = 1/21 Z⊙  and supports the metallicity dependence of the RSG SpT. 5 SUMMARY AND CONCLUSIONS This Letter presents the first spectroscopic census and confirmation of massive stars in SagDIG. GTC–OSIRIS optical spectroscopy has unveiled three OBA-type stars in the galaxy, and one RSG candidate. Pending the quantitative analysis of high-resolution optical and UV spectroscopy (e.g. Garcia et al. 2014), these could be the most iron-poor massive stars of the Local Group. The G0–2 I SpT of the RSG-candidate SD-4 is consistent with the reported trend of earlier RSGs SpT in environments of lower metallicity (Levesque & Massey 2012). It extends the sequence down to Z = 1/10–1/20 Z⊙. The four massive stars have been found in the eastern part of the galaxy, matching extended UV emission, and close to the highest concentrations of neutral hydrogen. Their CMD location relative to theoretical isochrones is consistent with their SpT as long as colour excess E(B − V)  ≳ 0.22, larger than the foreground value, is considered. This demonstrates that internal reddening in SagDIG is not negligible, and that the selection of candidate massive stars exclusively from their location at the optical CMD could dismiss a significant fraction of good candidates. This Letter opens the way to use SagDIG as a benchmark for studies of metal-poor massive stars. The most urgent follow-up is high-S/N, mid-resolution spectra of the reported stars in order to confirm their low metal content. The census of massive stars must also be completed, and this requires deep multi-object spectroscopic observations. In this respect, SagDIG is an ideal target for VLT–MUSE. By providing spectra of all the stars in the field of view, MUSE will bypass the shortcomings of reddening and target-selection biases. Observations are already planned. ACKNOWLEDGEMENTS MG would like to thank F. Najarro for fruitful discussions and support from grants FIS2012-39162-C06-01, ESP2013-47809-C3-1-R, and ESP2015-65597-C4-1-R. Ricardo Dorda is thanked for his library of standard stars. This research used NASA’s Astrophysics Data System, the SIMBAD data base (Wenger et al. 2000), and the Aladin Sky Atlas (Bonnarel et al. 2000; Boch & Fernique 2014). Based on observations made with the GTC, installed in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofísica de Canarias, on the island of La Palma. Footnotes 1 iraf is distributed by the National Optical Astronomy Observatory, operated by the Association of Universities for Research in Astronomy (AURA) under agreement with the National Science Foundation. 2 https://ned.ipac.caltech.edu/level5/Gray/frames.html REFERENCES Asplund M., Grevesse N., Sauval A. J., Scott P., 2009, ARA&A , 47, 481 CrossRef Search ADS   Beccari G. et al.  , 2014, A&A, 570, AA  78, B14 Boch T., Fernique P., 2014, Astrono. Data Anal. Softw. Syst. XXIII , 485, 277 Bonnarel F. et al.  , 2000, A&AS , 143, 33 CrossRef Search ADS   Bouret J.-C., Lanz T., Hillier D. J., Martins F., Marcolino W. L. F., Depagne E., 2015, MNRAS , 449, 1545 CrossRef Search ADS   Bresolin F., Urbaneja M. A., Gieren W., Pietrzyński G., Kudritzki R.-P., 2007, ApJ , 671, 2028 CrossRef Search ADS   Camacho I., Garcia M., Herrero A., Simón-Díaz S., 2016, A&A , 585, A82 CrossRef Search ADS   Castro N. et al.  , 2008, A&A , 485, 41 CrossRef Search ADS   Cox A. N., 2000, in Arthur N., ed., Allen’s Astrophysical Quantities . Springer-Verlag, Berlin Davies B., Origlia L., Kudritzki R.-P., Figer D. F., Rich R. M., Najarro F., Negueruela I., Clark J. S., 2009, ApJ , 696, 2014 CrossRef Search ADS   Dorda R., 2016, PhD thesis , Universidad de Alicante Elias J. H., Frogel J. A., Humphreys R. M., 1985, ApJS , 57, 91 CrossRef Search ADS   Filho M. E., Sánchez Almeida J., Muñoz-Tun C., Nuza S. E., Kitaura F., Heß S., 2015, ApJ , 802, 82 CrossRef Search ADS   Garcia M., Herrero A., 2013, A&A , 551, A74 CrossRef Search ADS   Garcia M., Herrero A., Najarro F., Lennon D. J., Alejandro Urbaneja M., 2014, ApJ , 788, 64 CrossRef Search ADS   Gazak J. Z. et al.  , 2015, ApJ , 805, 182 CrossRef Search ADS   González-Fernández C., Dorda R., Negueruela I., Marco A., 2015, A&A , 578, A3 CrossRef Search ADS   Groh J. H., Meynet G., Georgy C., Ekström S., 2013, A&A , 558, A131 CrossRef Search ADS   Herrero A., Garcia M., Uytterhoeven K., Najarro F., Lennon D. J., Vink J. S., Castro N., 2010, A&A , 513, A70 CrossRef Search ADS   Hosek M. W., Jr, et al.  , 2014, ApJ , 785, 151 CrossRef Search ADS   Karachentsev I., Aparicio A., Makarova L., 1999, A&A , 352, 363 Kaufer A., Venn K. A., Tolstoy E., Pinte C., Kudritzki R.-P., 2004, AJ , 127, 2723 CrossRef Search ADS   Lee H., Grebel E. K., Hodge P. W., 2003, A&A , 401, 141 CrossRef Search ADS   Levesque E. M., Massey P., 2012, AJ , 144, 2 CrossRef Search ADS   Madau P., Dickinson M., 2014, ARA&A , 52, 415 CrossRef Search ADS   Marigo P. et al.  , 2017, ApJ , 835, 77 CrossRef Search ADS   Massey P., 1998, in Aparicio A., Herrero A., Sanchez F., eds, Stellar astrophysics for the Local Group: VIII Canary Islands Winter School of Astrophysics . Cambridge Univ. Press, Cambridge, New York, p. 95 McConnachie A. W., 2012, AJ , 144, 4 CrossRef Search ADS   Momany Y. et al.  , 2005, A&A , 439, 111 CrossRef Search ADS   Momany Y. et al.  , 2014, A&A, 572 , A42, M14 Rejkuba M., Greggio L., Harris W. E., Harris G. L. H., Peng E. W., 2005, ApJ , 631, 262 CrossRef Search ADS   Saviane I., Rizzi L., Held E. V., Bresolin F., Momany Y., 2002, A&A , 390, 59 CrossRef Search ADS   Sirianni M. et al.  , 2005, PASP , 117, 1049 CrossRef Search ADS   Strobel N. V., Hodge P., Kennicutt R. C., Jr, 1991, ApJ , 383, 148 CrossRef Search ADS   Tang J., Bressan A., Slemer A., Marigo P., Girardi L., Bianchi L., Rosenfield P., Momany Y., 2016, MNRAS , 455, 3393 CrossRef Search ADS   Tramper F., Sana H., de Koter A., Kaper L., 2011, ApJ , 741, L8 CrossRef Search ADS   Wenger M. et al.  , 2000, A&AS , 143, 9 CrossRef Search ADS   Young L. M., Lo K. Y., 1997, ApJ , 490, 710 CrossRef Search ADS   © 2017 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Monthly Notices of the Royal Astronomical Society: Letters Oxford University Press

# Massive stars in the Sagittarius Dwarf Irregular Galaxy

, Volume 474 (1) – Feb 1, 2018
5 pages

/lp/ou_press/massive-stars-in-the-sagittarius-dwarf-irregular-galaxy-ed4CcFO1Od
Publisher
journal_eissn:11745-3933
© 2017 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society
ISSN
1745-3925
eISSN
1745-3933
D.O.I.
10.1093/mnrasl/slx194
Publisher site
See Article on Publisher Site

### Abstract

Abstract Low metallicity massive stars hold the key to interpret numerous processes in the past Universe including re–ionization, starburst galaxies, high-redshift supernovae, and γ-ray bursts. The Sagittarius Dwarf Irregular Galaxy [SagDIG, 12+log(O/H) = 7.37] represents an important landmark in the quest for analogues accessible with 10-m class telescopes. This Letter presents low-resolution spectroscopy executed with the Gran Telescopio Canarias that confirms that SagDIG hosts massive stars. The observations unveiled three OBA-type stars and one red supergiant candidate. Pending confirmation from high-resolution follow-up studies, these could be the most metal-poor massive stars of the Local Group. stars: early-type, stars: massive, galaxies: individual: SagDIG, galaxies: stellar content 1 INTRODUCTION Understanding the evolution of very metal-poor massive stars is key to interpret star formation and feedback at previous Cosmic epochs, and a necessary step towards the physics of the first stars of the Universe. This has stimulated an enthusiastic effort to discover resolved massive stars in environments of decreasing metal content, mostly dwarf irregular galaxies of the Local Group. The Small Magellanic Cloud (SMC) is the current standard for the low-Z regime, but metal-poorer massive stars are now at reach with multi-object spectrographs at 8–10m telescopes. First quantitative analyses were directed to the IC 1613, NGC 3109, and WLM galaxies (Bresolin et al. 2007; Herrero et al. 2010; Tramper et al. 2011) with encouragingly low 1/7O⊙  oxygen abundances, but subsequent studies indicate that their iron content is SMC-like (∼1/5Fe⊙) instead (Garcia et al. 2014; Hosek et al. 2014; Bouret et al. 2015). Following efforts were directed to Sextans A (Camacho et al. 2016) with both 1/10Fe⊙ and 1/10O⊙ (Kaufer et al. 2004). However, this value is still far from the ∼1/30 Z⊙ average metallicity of the Universe at the peak of star formation history and from measurements of the intergalactic medium at higher redshifts (Madau & Dickinson 2014). The Sagittarius Dwarf Irregular Galaxy (aka Sgr dIG, ESO 594-4, UKS 1927-17.7; SagDIG from now on) makes a promising environment to find very metal-poor, resolved massive stars. Its intermediate-age population stands out as one of the most Fe-poor among the Local Group galaxies that sustain star formation (e.g. McConnachie 2012), with values inferred from the location of the red giant branch (RGB) that range from [Fe/H] = −2.0 to −2.45 (Beccari et al. 2014; Karachentsev, Aparicio & Makarova 1999). Regarding the youngest populations, the oxygen abundance determined from the only confirmed H  ii region SagDIG-H ii#3 (Strobel, Hodge & Kennicutt 1991) is also remarkably low 12+log(O/H) = 7.37$$\rm ^{ + 0.13}_{ - 0.11}$$ (Saviane et al. 2002). Scaled by the Solar photospheric abundance (Asplund et al. 2009), this value is equivalent to 1/21O⊙. While SagDIG is considered one of the extremely metal-poor galaxies of the Local Universe, rivaling I Zw18 (Filho et al. 2015), it has never been explored for massive stars. This Letter presents the first spectroscopic observations of young massive stars in the galaxy. Section 2 describes the observations and data reduction. Target identification, spectral types (SpT), and membership are assessed in Section 3. The discovered red supergiant (RSG) candidate is discussed in the context of the Local Group in Section 4. Finally, summary and some prospects for future work are provided in Section 5. 2 DATA AND REDUCTION The observations were executed with the OSIRIS spectrograph at the 10.4-m Gran Telescopio Canarias (GTC), programme GTC67-13A (PI M. Garcia). The 1.2 arcsec slit and R2000B VPH were used to achieve resolution R ∼ 1000 at ∼4000–5500 Å. The slit was aligned on targets SD-1 and SD-7 to cross the galaxy (Fig. 1) and enclose several ultraviolet (UV)-bright sources. A total of 2 h of grey sky, 1.2 arcsec seeing exposure time were devoted to the slit. Figure 1. View largeDownload slide SagDIG: GTC–OSIRIS acquisition image (1.2 arcsec seeing, filter–free), and programme stars. Figure 1. View largeDownload slide SagDIG: GTC–OSIRIS acquisition image (1.2 arcsec seeing, filter–free), and programme stars. The data were reduced following standard iraf1 procedures adapted for the team’s OSIRIS runs (Garcia & Herrero 2013). The spectra were flux–calibrated against the flux standard GD190, achieving about 5 per cent accuracy for λ > 4250 Å. Finally, the spectra were set to the heliocentric frame and corrected by radial velocity. The latter were calculated from the Doppler shift measured at the core of the Balmer lines. The reduced spectra are shown in Figs 2–3. Figure 2. View largeDownload slide GTC–OSIRIS spectra of blue massive stars in SagDIG. The data have been rebinned to three pixels for clarity, and corrected by heliocentric and radial velocity. The three spectra show a clear Balmer series. Figure 2. View largeDownload slide GTC–OSIRIS spectra of blue massive stars in SagDIG. The data have been rebinned to three pixels for clarity, and corrected by heliocentric and radial velocity. The three spectra show a clear Balmer series. Figure 3. View largeDownload slide Red programme stars. The flux-calibrated spectra were smoothed to five points, and corrected by heliocentric and radial velocity. In both panels the upper spectrum has been shifted by a constant for clarity (SD-4: 1.5E − 17 erg cm − 2 s − 1  Å − 1, SD-6: 3E − 17 erg cm − 2 s − 1  Å − 1). Figure 3. View largeDownload slide Red programme stars. The flux-calibrated spectra were smoothed to five points, and corrected by heliocentric and radial velocity. In both panels the upper spectrum has been shifted by a constant for clarity (SD-4: 1.5E − 17 erg cm − 2 s − 1  Å − 1, SD-6: 3E − 17 erg cm − 2 s − 1  Å − 1). The programme faced the challenges related to observations of faint, distant stars: contamination by nearby sources and small number of counts over the background level. For instance the spectrum of SD-1, with V ∼ 20.2, registered a factor ∼2.5 less counts than the extracted sky spectrum. The resulting signal to noise ratio (S/N) is consequently poor even after careful sky subtraction and some solar features remained. These shortcomings discourage future optical range spectroscopic observations of targets fainter than V = 20 mag in grey time. Targets SD-1 to 4 are found in an area of extended continuum emission (detected in far-UV imaging as well, e.g. Momany et al. 2014) that may contaminate the extracted spectra. In order to minimize blends, the sources were extracted discarding the wings of the spatial distribution of photons in the 2-D images. 3 SOURCE IDENTIFICATION, SPECTRAL CLASSIFICATION AND MEMBERSHIP SagDIG is projected close to the Galactic Centre and the line of sight is severely contaminated with Milky Way stars. Hence, ensuring that the programme targets belong to the galaxy was an issue. This section discusses jointly target identification, spectral classification, membership, and their consistency. The location of the programme stars in the colour–magnitude diagram (CMD) is evaluated as a final check. The programme targets were identified in the Beccari et al. (2014, hereafter B14) and Momany et al. (2014, M14) catalogues by comparing the GTC acquisition images against HST–ACS archival observations (GO-9820 P.I. Y. Momany). Identification numbers and photometry are listed in Table 1. Table 1. Observed stars: SpT from this work, radial velocities (vrad) used in Figs 2–3, and cross-identification numbers and photometry from M14 and B14. MV was computed from apparent magnitudes, distance modulus (SagDIG stars: DM = 25.10 Momany et al. (2005); foreground stars: DM = 14.50), and Cox (2000)’s intrinsic colours to estimate E(V − I) and correct from extinction. The last column flags SagDIG membership. ID  SpT  vrad  M14  F475W  F606W  F814W  B14  V  I  E(V − I)  MV  Membership      (km s − 1)  ID        ID            SD-1  Late–O I  − 65  33466  20.189  20.125  20.024  8903  20.190  20.154  0.51  − 6.1  yes  SD-2  BA I  − 55  31722  20.355  20.278  20.120  b  20.38  20.24  0.12  − 5.1  yes  SD-3  OB V  − 95  32939  21.408  21.480  21.515  8822  21.442  21.585  0.28  − 4.3  yes  SD-4  G0–2 I  − 100  30481  21.558  20.503  19.394  8805  20.731  19.473  0.42  − 5.3  yes  SD-5  G4 III  − 90  3019  20.887  20.055  19.144  b  20.38  19.23  0.26  5.3  fg  SD-6  G5 V  − 70  a    18.59  17.82  7783  18.875  17.912  0.07  4.2  fg  SD-7  K4 V  − 50  a    18.76  17.74  8017  19.124  17.819  0.42  3.7  fga  ID  SpT  vrad  M14  F475W  F606W  F814W  B14  V  I  E(V − I)  MV  Membership      (km s − 1)  ID        ID            SD-1  Late–O I  − 65  33466  20.189  20.125  20.024  8903  20.190  20.154  0.51  − 6.1  yes  SD-2  BA I  − 55  31722  20.355  20.278  20.120  b  20.38  20.24  0.12  − 5.1  yes  SD-3  OB V  − 95  32939  21.408  21.480  21.515  8822  21.442  21.585  0.28  − 4.3  yes  SD-4  G0–2 I  − 100  30481  21.558  20.503  19.394  8805  20.731  19.473  0.42  − 5.3  yes  SD-5  G4 III  − 90  3019  20.887  20.055  19.144  b  20.38  19.23  0.26  5.3  fg  SD-6  G5 V  − 70  a    18.59  17.82  7783  18.875  17.912  0.07  4.2  fg  SD-7  K4 V  − 50  a    18.76  17.74  8017  19.124  17.819  0.42  3.7  fga  aStar without M14-counterpart. F606W and F814W magnitudes estimated from B14’s V- and I-magnitudes, using Rejkuba et al. (2005)’s transformations. bStar without B14-counterpart. VI-magnitudes estimated from M14, using Rejkuba et al. (2005)’s transformations. View Large The spectral classification of blue stars, whose most prominent spectral feature is the Balmer series, was carried out following classical criteria (see e.g. Garcia & Herrero 2013). Their spectra are shown in Fig. 2. The rest of the sample (Fig. 3) was classified after Dorda (2016)’s scheme for cool stars (G-types and later) in the Large Magellanic Cloud (LMC) and the SMC, and the spectral classification atlas by Gray (2000).2 They were also compared against SMC and MW standards. Membership of the programme stars with counterpart in M14’s catalogue, that registered relative shifts from multi-epoch HST visits, was ensured by the lack of proper motions. In addition, the estimated absolute magnitudes were compared against the calibrated values for the SpT assigned in this work. The consistency of the stellar radial velocities and SagDIG’s systemic velocity (−78.5 km s − 1, McConnachie 2012) was not determinant because of the low spectral resolution of these observations (∼300 km s − 1, hence typical vrad  uncertainties of ∼100 km s − 1). The following subsections provide more details on the targets that were positively identified as SagDIG massive stars. 3.1 SD-1 The HST–ACS images reveal that SD-1 is the blend of two stars ∼ 1.1 arcsec apart, SD-1-blue (M14-33466; B14-8903, V = 20.190) and SD-1-red (B14-8909, V = 21.774). SD-1 will be identified with the brightest star of the blend, SD-1-blue, from now on. The Balmer series is clearly detected in its spectrum (Fig. 2), and the comparison against the sky spectrum confirms that the stellar component dominates Hγand Hβ. The data also display He i+He ii 4026 Å, He i 4387 Å, and He i 4471 Å, but the detection of He ii lines is unclear. He ii 4541 Å is absent and the He ii 4200 Å and He ii 5411 Å absorptions overlap with sky features. The exception is He ii 4686 Å, detected in emission. Its width together with the absence of H i and He i emission lines renders a nebular origin unlikely, and the comparison against the subtracted sky spectrum discards the feature as an artefact. The He ii 4686 Å emission could be consistent with a late-O type, with the rest of He ii lines too weak for detection at the S/N of the spectrum. The non-detection of Mg ii 4481 Å and Si iii 4552 Å supports the late-O classification, although it could also be caused by the low metallicity of the star. The spectrum of SD-1 is reminiscent of star C1_31 in NGC 55 (Castro et al. 2008). C1_31 is mostly featureless except for the Balmer lines and a prominent broad He ii 4686 Å emission, with He i 4471 Å clearly seen but no He ii 4541 Å. It was classified as early-O supergiant. Similarly, SD-1 is classified as late-O I, with the He ii 4686 Å emission and narrow Balmer lines supporting the supergiant luminosity class. The estimated absolute magnitude MV = −6.1 agrees with the calibrated value for late-O supergiants (Massey 1998). 3.2 SD-2 SD-2 is the blend of M14-31722 (F606W = 20.278) and M14-31570 (F606W = 21.839), 0.7 arcsec apart and unresolved. SD-2 is identified with the brightest source M14-31722. Its spectrum shows the Balmer series, He i 4471 Å, He i 4387 Å, and He i 4922 Å (Fig. 2). There may be some Mg ii 4481 Å, but no lines of Si iii, Si iv, or He ii. The calibrated stellar flux corresponds to a colder object than SD-1. Since the width of the Balmer lines is similar to SD-1, and the strength of the He i lines is almost at noise level, SD-2 is classified as a BA supergiant. The estimated absolute magnitude MV  = −5.1 is consistent with a blue supergiant located at SagDIG. 3.3 SD-3 Target SD-3 is faint and off-slit, resulting in extremely low spectral S/N. The sky contribution amounts to 83 per cent of the total flux at blue wavelengths, increasing to 91 per cent at 4800 Å. SD-3 overlaps with the red component of the SD-1 blend. However, there is a shift between SD-3’s strongest features and the same lines in SD-1 reflecting the different radial velocities of the two stars and confirming that the lines genuinely belong to SD-3. The spectrum of SD-3 (Fig. 2) exhibits the Balmer series and He i 4471 Å. The He ii lines, but also Si iii 4552 Å and Mg ii 4481 Å, are absent suggesting late-O or early-B SpT. The spectrum lacks any He ii 4686 Å, but contamination by SD-1-red begins to be significant in this region. The wings of Hγ  are dominated by the stellar component, and the profile is wider than SD-1’s, suggesting class III or V. Since the absolute magnitude agrees with Massey (1998)’s calibration for O9–B0 V (MV = −4.4 to −3.8), SD-3 is classified as an OB dwarf. 3.4 SD-4 SD-4 (M14-30481, F606W = 20.503) is contaminated by a nearby red star with similar magnitude (M14-30025, F606W = 20.360). According to their proper motions, both belong to SagDIG. Luminosity class was constrained in the first place, following Dorda (2016). Because the 5167 Å component is the strongest of the magnesium triplet Mg i 5167, 5172, 5184 Å, SD-4 was classified as a supergiant. The relative strength of Fe i 5247, 5250, 5255 Å versus Ca i+Fe i+Ti i 5270 Å, and Mn i 5433 Å versus Mn i 5447 Å support this luminosity class. The presence of the G band and its relative strength compared to Hγ  and Ca i 4227 Å suggests that the SpT is slightly later than G0. The relative strength of Fe i 4325 Å compared to both Hγand Fe i 4383 Å, on the other hand, suggests spectral type G0 or earlier. A G0–2 I type is finally assigned. It is noted that the estimated absolute magnitude MV  = −5.3 is slighlty fainter than the calibrated MV  = −6.4 for a G0 ∼ I star. Nevertheless, SD-4’s membership to SagDIG is supported by the lack of detection of proper motions by M14. Since RSGs stars may experience spectral variability (see Section 4), SD-4 is proposed as a RSG-candidate in SagDIG. 3.5 Target stars in the colour–magnitude diagram The location of the target stars in the CMD is consistent with the SpT assigned in this work (Fig. 4). The three OBA stars are found in the upper part of the blue plume. The red stars SD-5, SD-6, and SD-7 are enclosed in the vertical sequence of Galactic foreground contamination. SD-4, on the other hand, is above the tip of the RGB consistently with the star being a RSG. Figure 4. View largeDownload slide SagDIG’s CMD built from M14’s catalogue (grey dots). Marigo et al. (2017)’s isochrones for Z = 0.001 (∼1/15 Z⊙) are included for comparison, with colour coding log (age[yr]) = 6.0, 6.49, 7.0, 7.49, 8.0. Younger isochrones are not included because they overlap at F606W–F814W ∼ 0 in the F814W = 20–22 range. The isochrones were shifted by SagDIG distance modulus (m − M)0 = 25.10 and reddened by the equivalent to E(B − V) = 0.22 (AF606W = 0.646, AF814W = 0.410, after Sirianni et al. 2005). Figure 4. View largeDownload slide SagDIG’s CMD built from M14’s catalogue (grey dots). Marigo et al. (2017)’s isochrones for Z = 0.001 (∼1/15 Z⊙) are included for comparison, with colour coding log (age[yr]) = 6.0, 6.49, 7.0, 7.49, 8.0. Younger isochrones are not included because they overlap at F606W–F814W ∼ 0 in the F814W = 20–22 range. The isochrones were shifted by SagDIG distance modulus (m − M)0 = 25.10 and reddened by the equivalent to E(B − V) = 0.22 (AF606W = 0.646, AF814W = 0.410, after Sirianni et al. 2005). The loci of the youngest stars in the CMD is also consistent with the theoretical isochrones (Z = 0.001 or ∼1/15 Z⊙, Marigo et al. 2017) but require a larger correction than the reported E(B − V)fg  = 0.12 foreground extinction (e.g. Momany et al. 2005). Internal reddening can indeed be expected, since SD-1 to 4 are located in a concentration of neutral hydrogen (Young & Lo 1997, also B14). The E(B − V)  ≃0.22 colour excess calculated from the Balmer decrement of SagDIG-H ii#3 was adopted (Lee, Grebel & Hodge 2003; Tang et al. 2016). SD-3 is located at the main sequence where isochrones of log (age[yr]) = 4.5–7.0 overlap, in accordance with its OB V SpT. SD-2 is found between the log (age[yr]) = 7.0 and 7.49 isochrones, as expected considering that BA supergiants have evolved off the main sequence and are slightly older than O-stars. The location of SD-4 is consistent with a RSG with similar age log (age[yr]) = 7.49. SD-1 is found between the log (age[yr]) = 7.0–7.49 isochrones, however, it is more reddened than the other programme stars. If its E(B − V) = 0.38 spectroscopic colour excess is taken at face value, the isochrones would need an additional shift of Δ(F606W–F814W) ∼ 0.2 mag and ΔF814W ∼ 0.3 mag to account for the increased extinction. This would bring SD-1 close to the log (age[yr]) = 6.49 isochrone (3 Myr), that better agrees with the expected age of an O-type supergiant. The unknown amount of internal reddening in SagDIG hinders the interpretation of the CMD (SD-1,2,3 would be mistaken as intermediate-mass stars if E(B − V)fg was used) and may hide a significant population of young and massive stars. Extinction can explain the paucity of stars in the upper blue plume, which is at odds with the extended emission of the galaxy in the UV. These results may be considered a reminder that the total galactic stellar content may be underestimated from the luminosity function, and advices against selecting massive star candidates from their location in classical CMDs only. 4 THE MOST METAL-POOR RSG TO DATE RSGs are He-burning descendants of Mini  ≲ 30 M⊙ stars and the coolest stage in the life-cycle of massive stars. They are gathering increased interest because some RSGs (9 < Mini<16.5–23M⊙) are the alledged progenitors of Type-IIP supernovae (Groh et al. 2013). Besides, they are proving reliable metallicity probes optimally reached by current, powerful infrared spectrographs: Keck–NIRSPEC, GTC–EMIR, or VLT–KMOS (e.g. Davies et al. 2009; Gazak et al. 2015). RSGs are classically associated with M-types, but the average SpT shifts towards earlier types in metal-poorer galaxies (Elias, Frogel & Humphreys 1985; Levesque & Massey 2012). This trend is beyond a classification bias produced by the weakening of TiO bands when metallicity decreases, and it is believed to reflect the fact that the Hayashi limit shifts towards higher effective temperatures with decreasing metallicity (see discussion by Elias et al. 1985). For this reason, and because sometimes they exhibit spectroscopic variability, types as early as G0 are considered in the class of RSGs (see discussion in González-Fernández et al. 2015). According to the latest compilation of Local Group RSGs by Levesque & Massey (2012), the most frequent spectral SpT shifts from M2 I in the Milky Way and the LMC, to mid-K types in the SMC and to K0-1 in WLM (1/7–1/10 Z⊙). The RSG-candidate SD-4, with SpT G0–2 I, extends the spanned range of metallicity down to Z = 1/21 Z⊙  and supports the metallicity dependence of the RSG SpT. 5 SUMMARY AND CONCLUSIONS This Letter presents the first spectroscopic census and confirmation of massive stars in SagDIG. GTC–OSIRIS optical spectroscopy has unveiled three OBA-type stars in the galaxy, and one RSG candidate. Pending the quantitative analysis of high-resolution optical and UV spectroscopy (e.g. Garcia et al. 2014), these could be the most iron-poor massive stars of the Local Group. The G0–2 I SpT of the RSG-candidate SD-4 is consistent with the reported trend of earlier RSGs SpT in environments of lower metallicity (Levesque & Massey 2012). It extends the sequence down to Z = 1/10–1/20 Z⊙. The four massive stars have been found in the eastern part of the galaxy, matching extended UV emission, and close to the highest concentrations of neutral hydrogen. Their CMD location relative to theoretical isochrones is consistent with their SpT as long as colour excess E(B − V)  ≳ 0.22, larger than the foreground value, is considered. This demonstrates that internal reddening in SagDIG is not negligible, and that the selection of candidate massive stars exclusively from their location at the optical CMD could dismiss a significant fraction of good candidates. This Letter opens the way to use SagDIG as a benchmark for studies of metal-poor massive stars. The most urgent follow-up is high-S/N, mid-resolution spectra of the reported stars in order to confirm their low metal content. The census of massive stars must also be completed, and this requires deep multi-object spectroscopic observations. In this respect, SagDIG is an ideal target for VLT–MUSE. By providing spectra of all the stars in the field of view, MUSE will bypass the shortcomings of reddening and target-selection biases. Observations are already planned. ACKNOWLEDGEMENTS MG would like to thank F. Najarro for fruitful discussions and support from grants FIS2012-39162-C06-01, ESP2013-47809-C3-1-R, and ESP2015-65597-C4-1-R. Ricardo Dorda is thanked for his library of standard stars. This research used NASA’s Astrophysics Data System, the SIMBAD data base (Wenger et al. 2000), and the Aladin Sky Atlas (Bonnarel et al. 2000; Boch & Fernique 2014). Based on observations made with the GTC, installed in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofísica de Canarias, on the island of La Palma. Footnotes 1 iraf is distributed by the National Optical Astronomy Observatory, operated by the Association of Universities for Research in Astronomy (AURA) under agreement with the National Science Foundation. 2 https://ned.ipac.caltech.edu/level5/Gray/frames.html REFERENCES Asplund M., Grevesse N., Sauval A. J., Scott P., 2009, ARA&A , 47, 481 CrossRef Search ADS   Beccari G. et al.  , 2014, A&A, 570, AA  78, B14 Boch T., Fernique P., 2014, Astrono. Data Anal. Softw. Syst. XXIII , 485, 277 Bonnarel F. et al.  , 2000, A&AS , 143, 33 CrossRef Search ADS   Bouret J.-C., Lanz T., Hillier D. J., Martins F., Marcolino W. L. F., Depagne E., 2015, MNRAS , 449, 1545 CrossRef Search ADS   Bresolin F., Urbaneja M. A., Gieren W., Pietrzyński G., Kudritzki R.-P., 2007, ApJ , 671, 2028 CrossRef Search ADS   Camacho I., Garcia M., Herrero A., Simón-Díaz S., 2016, A&A , 585, A82 CrossRef Search ADS   Castro N. et al.  , 2008, A&A , 485, 41 CrossRef Search ADS   Cox A. N., 2000, in Arthur N., ed., Allen’s Astrophysical Quantities . Springer-Verlag, Berlin Davies B., Origlia L., Kudritzki R.-P., Figer D. F., Rich R. M., Najarro F., Negueruela I., Clark J. S., 2009, ApJ , 696, 2014 CrossRef Search ADS   Dorda R., 2016, PhD thesis , Universidad de Alicante Elias J. H., Frogel J. A., Humphreys R. M., 1985, ApJS , 57, 91 CrossRef Search ADS   Filho M. E., Sánchez Almeida J., Muñoz-Tun C., Nuza S. E., Kitaura F., Heß S., 2015, ApJ , 802, 82 CrossRef Search ADS   Garcia M., Herrero A., 2013, A&A , 551, A74 CrossRef Search ADS   Garcia M., Herrero A., Najarro F., Lennon D. J., Alejandro Urbaneja M., 2014, ApJ , 788, 64 CrossRef Search ADS   Gazak J. Z. et al.  , 2015, ApJ , 805, 182 CrossRef Search ADS   González-Fernández C., Dorda R., Negueruela I., Marco A., 2015, A&A , 578, A3 CrossRef Search ADS   Groh J. H., Meynet G., Georgy C., Ekström S., 2013, A&A , 558, A131 CrossRef Search ADS   Herrero A., Garcia M., Uytterhoeven K., Najarro F., Lennon D. J., Vink J. S., Castro N., 2010, A&A , 513, A70 CrossRef Search ADS   Hosek M. W., Jr, et al.  , 2014, ApJ , 785, 151 CrossRef Search ADS   Karachentsev I., Aparicio A., Makarova L., 1999, A&A , 352, 363 Kaufer A., Venn K. A., Tolstoy E., Pinte C., Kudritzki R.-P., 2004, AJ , 127, 2723 CrossRef Search ADS   Lee H., Grebel E. K., Hodge P. W., 2003, A&A , 401, 141 CrossRef Search ADS   Levesque E. M., Massey P., 2012, AJ , 144, 2 CrossRef Search ADS   Madau P., Dickinson M., 2014, ARA&A , 52, 415 CrossRef Search ADS   Marigo P. et al.  , 2017, ApJ , 835, 77 CrossRef Search ADS   Massey P., 1998, in Aparicio A., Herrero A., Sanchez F., eds, Stellar astrophysics for the Local Group: VIII Canary Islands Winter School of Astrophysics . Cambridge Univ. Press, Cambridge, New York, p. 95 McConnachie A. W., 2012, AJ , 144, 4 CrossRef Search ADS   Momany Y. et al.  , 2005, A&A , 439, 111 CrossRef Search ADS   Momany Y. et al.  , 2014, A&A, 572 , A42, M14 Rejkuba M., Greggio L., Harris W. E., Harris G. L. H., Peng E. W., 2005, ApJ , 631, 262 CrossRef Search ADS   Saviane I., Rizzi L., Held E. V., Bresolin F., Momany Y., 2002, A&A , 390, 59 CrossRef Search ADS   Sirianni M. et al.  , 2005, PASP , 117, 1049 CrossRef Search ADS   Strobel N. V., Hodge P., Kennicutt R. C., Jr, 1991, ApJ , 383, 148 CrossRef Search ADS   Tang J., Bressan A., Slemer A., Marigo P., Girardi L., Bianchi L., Rosenfield P., Momany Y., 2016, MNRAS , 455, 3393 CrossRef Search ADS   Tramper F., Sana H., de Koter A., Kaper L., 2011, ApJ , 741, L8 CrossRef Search ADS   Wenger M. et al.  , 2000, A&AS , 143, 9 CrossRef Search ADS   Young L. M., Lo K. Y., 1997, ApJ , 490, 710 CrossRef Search ADS   © 2017 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society

### Journal

Monthly Notices of the Royal Astronomical Society: LettersOxford University Press

Published: Feb 1, 2018

## You’re reading a free preview. Subscribe to read the entire article.

### DeepDyve is your personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month ### Explore the DeepDyve Library ### Search Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly ### Organize Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place. ### Access Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals. ### Your journals are on DeepDyve Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more. All the latest content is available, no embargo periods. DeepDyve ### Freelancer DeepDyve ### Pro Price FREE$49/month
\$360/year

Save searches from
PubMed

Create lists to

Export lists, citations