Chemical Composition of Young Stars in the Leading Arm of the Magellanic System0 Lan Zhang1,2,3, Christian Moni Bidin4, Dana I. Casetti-Dinescu5, R´ene A. M´endez3, Terrence M. Girard7, Vladimir I. Korchagin8, Katherine Vieira9, William F. van Altena6 & 7 1 Gang Zhao1 0 2 n Received ; accepted a J 3 ] Not to appear in Nonlearned J., 45. R S . h p - o r t s a [ 1 v 0 0Basedonobservationswiththe6.5mClaytelescopeatLasCampanasObservatory, Chile(programID: 9 5 CN2014A-057) 0 0 1KeyLabofOpticalAstronomy,NationalAstronomicalObservatories,CAS,20ADatunRoad,Chaoyang . 1 District, 100012 Beijing, China 0 7 2CASSouthAmericaCenterforAstronomy,CaminoElobservatorio#1515,LasCondes,Santiago,Chile 1 : v 3Departamento de Astronomia Universidad de Chile, Camino El observatorio #1515, Las Condes, Santi- i X ago, Chile r a 4Instituto de Astronom´ıa, Universidad Cat´olica del Norte, Av. Angomos 0610, Antofagasta, Chile 5DepartmentofPhysics,SouthernConnecticutStateUniversity,501CrescentSt.,NewHaven,CT06515, USA 6Astronomy Department, Yale University, 260 Whitney Ave. , New Haven, CT 06511, USA 714 Dunn Rd, Hamden, Connecticut, CT 06518, USA 8Institute of Physics, Southern Federal University, Stachki st/194, 344090, Rostov-on-Don, Russia 9Centro de Investigaciones de Astronomi´a, Apartado Postal 264, M´erida 5101-A, Venezuela – 2 – ABSTRACT Chemical abundances of eight O- and B-type stars are determined from high- resolution spectra obtained with the MIKE instrument on the Magellan 6.5m Clay telescope. The sample is selected from 42 candidates of membership in the Leading Arm of the Magellanic System. Stellar parameters are measured by two independent grids of model atmospheres and analysis procedures, confirming the consistency of the stellar parameter results. Abundances of seven elements (He, C, N, O, Mg, Si, and S) are determined for the stars, as are their radial velocities and estimates of distances and ages. Among the seven B-type stars analyzed, the five that have radial velocities compatible with membership to the LA have an average [Mg/H] of −0.42±0.16, significantly lower than the average of the remaining two [Mg/H] = −0.07±0.06 that are kinematical members of the Galactic disk. Among the five LA members, fourhaveindividual[Mg/H]abundancecompatiblewiththatintheLMC.Within errors, we can not exclude the possibility that one of these stars has a [Mg/H] consistentwiththemoremetal-poor, SMC-likematerial. Theremainingfifthstar has a [Mg/H] close to MW values. Distances to the LA members indicate that they are at the edge of the Galactic disk, while ages are of the order of ∼ 50−70 Myr, lower than the dynamical age of the LA, suggesting a single star-forming episode in the LA. V the LA members decreases with decreasing Magellanic LSR longitude, confirming the results of previous LA gas studies. Subject headings: – 3 – 1. Introduction The Magellanic Stream (MS, Mathewson et al. 1974; D’Onghia & Fox 2015), including the Bridge (Kerr 1957; Hindman et al. 1963), and the Leading Arm (LA) is a well-known, ∼ 200◦-long H I structure, evidencing the interaction between the Small and Large Magellanic Clouds (SMC and LMC, respectively) and the Milky Way (MW) (Nidever et al. 2010). In recent years, comprehensive studies of these structures have been carried out to understand the complex interactions between the Clouds and the MW, and hence to provide constraints on the gravitational potentials of these galaxies. The LA has a complex structure, consisting of as many as four substructures according to For et al. (2013) and Venzmer et al. (2012), situated above and below the Galactic plane and encompassing 60◦ width. Absolute proper-motion measurements of the Clouds include the HST-based determination by Kallivayalil et al. (2006, 2013) and the ground-based study of Vieira et al. (2010). These were used by Diaz & Bekki (2012) to explore a tidal model that simulates the Clouds’ interactions. In their modeling, a leading arm can be produced from the tidal interaction, but the observed multi-branch morphology of the LA and its kinematics are not reproduced. It has been thus suggested that the LA is hydrodynamically interacting with both the Galactic gaseous disk (McClure-Griffiths et al. 2008) and the hot halo (Diaz & Bekki 2011). If this is the case, newly formed stars in the LA may be expected. In order to explore this hypothesis, Casetti-Dinescu et al. (2012, hereafter CD12) searched for candidate OB-type stars over a ∼7900 deg2 region including the Magellanic Clouds, the Bridge, the Leading Arm, and part of the Magellanic Stream. That study was based on photometry of the Galaxy Evolution Explorer Survey (GALEX, Bianchi et al. 2011), the Two Micron All Sky Survey (2MASS, Skrutskie et al. 2006), the Southern Proper Motion Program’s catalog 4 (SPM4, Girard et al. 2011), and the American Association of – 4 – Variable Star Observers All Sky Photometric Survey (APASS, Henden et al. 2011), and on proper motions from SPM4. The outcome was a list of 567 young OB-type candidates as possible members of the MS. In a subsequent study, Casetti-Dinescu et al. (2014, hereafter CD14) analyzed the kinematics and spectral properties of 42 of the above-mentioned OB candidates using intermediate-resolution spectra. These candidates were selected to be in three stellar overdensities in the LA region, above and below the plane. CD14 found a total of 19 young, OB-type stars. Of these, five young B-type stars had radial velocities (RVs) compatible with LA kinematics, and a low velocity dispersion of ∼ 33 km s−1. They also found one O6V star with an age of 1 − 2 Myr situated ∼ 39 kpc from the Galactic center. These discoveries suggested that recent star formation occurred in the LA as a likely consequence of the hydrodynamical interaction between the MS and the Galactic disk where the MS crosses the Galactic plane. The aim of the present work is to further study the detailed chemical abundances of these young stars. We wish to 1) understand the origin of these stars and thus further constrain the history of the LA, and 2) explore whether there is a chemical difference between young stars that are kinematical members of the LA and those that are not. To this end, we have selected a sample of eight stars: three above the Galactic plane and five below (including the O6V star). Also, of these eight stars, five are the aforesaid likely kinematical members of the LA, while three are not. We determine the abundances of elements Helium, Carbon, Nitrogen, Oxygen, Magnesium, Silicon, and Sulfur using high-resolution spectra obtained with the Magellan Inamori Kyocera Echelle (MIKE; Bernstein et al. 2003) on the Magellan 6.5m Clay telescope at Las Companas Observatory (Chile). The observations are described in § 2. The method and the procedures of the determination of stellar parameters and abundance analysis are described in § 3 and § 4. – 5 – The results are presented and discussed in § 5, § 6, and § 7, with a summary in § 8. 2. Observations & Data Reduction 2.1. Target selection McClure-Griffiths et al. (2008) derived a kinematic distance to the LA of d ∼ 21 kpc from the Sun. Also, H I gas velocities in the LA are in excess of 100 km s−1 (see e.g., Fig. 8 in Nidever et al. 2010). Therefore, we primarily selected stars whose distances are close to 21 kpc and RV > 100 km s−1 from the pilot, intermediated-resolution spectral study of CD14. In our sample, we also included three stars with RV < 100 km s−1 for the purpose of comparison. For observational reasons, faint stars (V > 16) were excluded. This leaves eight stars as targets in our study. We designate our stars as “CD14-A**” for those below the Galactic plane, at Magellanic coordinates (Λ , B ) ∼ (15◦, −22◦), and as “CD14-B**” M M for those above the plane, at (Λ , B ) ∼ (42◦, −8◦). The detailed spatial distribution M M of the stars can be seen in Fig. 1 of CD14, and we do not repeat it here. In our sample, CD14-A08 is the previously identified very hot, massive (∼40 M ), young (1−2 Myr) star (cid:12) with spectral type of O6V at a heliocentric distance of ∼39 kpc (CD14). The remaining seven stars are B-type stars. We list our stars in Tab. 1, which includes the current designation, the SPM4 identification number, equatorial coordinates, V magnitudes, as well as other observation details. – 6 – 2.2. High-resolution spectroscopy and data reduction High-resolution spectra were obtained with the MIKE instrument on the 6.5m Clay telescope in March of 2014 for these eight stars. The setup gave a resolution of R ∼ 33000 and R ∼ 29000 for blue (3200−5000 ˚A) and red (4900−10000 ˚A) sides, respectively. The average seeing and airmass during the observations were 1”.2 and 1.3, respectively. The spectra were binned during the data collection. MIKE binned 3×2 were used, where 3 and 2 are spatial and spectral direction, respectively. Table 1 summarizes the observation details and exposure times. The standard Carnegie Python Distribution (CarPy) routines for MIKE were used for data reduction, including order identification, wavelength calibration, flat-field correction, background subtraction, one-dimensional spectra extraction and flux calibration. The radial velocities (RV) of targets were measured by cross-correlation with a synthetic template whose temperature and gravity is similar to those of the targets after continuum rectification. We used the IRAF1 task FXCOR2 based on the standard algorithm of Tonry & Davis (1979), where the synthetic template was taken from the synthetic spectra library of Munari et al. (2005). Two telluric oxygen A bands: the band 6884 ˚A, transition (0,1) and the band 7621 ˚A, transition (0,0) were also adopted to correct the RV zero point, where the molecular data were taken from the HIgh-resolution TRANsmission molecular absorption (HITRAN) database3. The signal-to-noise ratios (S/Ns) of the spectra at 4000 and 5800 ˚A is also presented in Tab. 1. Examples of a portion of spectra are shown in Fig. 1 and the 1IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation. (Tody 1986, 1993) 2http://iraf.noao.edu/projects/rv/fxcor.html 3https://www.cfa.harvard.edu/hitran/ – 7 – obtained RVs are listed in Tab. 2. 3. Stellar Atmosphere Parameters 3.1. Measurements The effective temperature (T ), surface gravity (logg), and atmospheric helium eff abundance of the target stars were measured by fitting the observed hydrogen and helium lines with synthetic spectra. The stellar model atmospheres we employed for target stars are interpolated from comprehensive grids of metal line-blanketed, non local thermodynamic equilibrium (NLTE), plane-parallel, hydrostatic model atmospheres of O-type (Lanz & Hubeny 2003a, OSTAR2002) and B-type (Lanz & Hubeny 2007, BSTAR2006) stars, with opacity sampling which is a simple Monte Carlo-like sampling of the superline cross sections and solar abundance. Both sets of grids were generated by TLUSTY, a program for calculating plane-parallel, horizontally homogeneous model stellar atmospheres in radiative and hydrostatic equilibrium (Hubeny & Lanz 1995). The OSTAR2002 grid contains 12 values of T , 27,500 K < T < 55,000 K with eff eff 2500 K steps, eight logg, 3.0 < logg < 4.75 with 0.25 cm s−2 steps, and 10 chemical compositions, from metal-rich relative to the Sun to metal-free, while the BSTAR2006 includes 16 values of T , 15,000 K < T < 30,000 K with 1000 K steps, 13 logg, eff eff 1.75 < logg < 4.75 with 0.25 cm s−2 steps, six chemical compositions and a microturbulent velocity of 2 km s−1, from twice to one-tenth of the solar metallicity and metal-free. For more details on the grids and computations of model atmospheres, we refer the reader to Lanz & Hubeny (2003a, 2007) and references therein. For all our targets, the solar-metallicity (Z/Z = 1) grid is adopted. However, for CD14-A05 and CD14-A12, (cid:12) which show strong helium lines, an additional super-solar metallicity (Z/Z = 2) grid was (cid:12) – 8 – also employed. The spectrum synthesis program we adopted for line synthesis is the SYNSPEC, a general spectrum synthesis program (developed by Ivan Hubeny & Thierry Lanz) 4. It can compute line formation under both LTE and NLTE conditions. In our analysis, the input model atmospheres are NLTE, therefore, the automatic NLTE treatment mode of SYNSPEC is employed, in which the program automatically decides which levels are treated in NLTE, and assigns proper NLTE populations to the lower and upper level of a given transition. The input model atoms and ions (Lanz & Hubeny 2003b) are provided by TLUSTY5. The form of the intrinsic line profiles is a Voigt function, in which natural, Stark, van der Waals, and thermal Doppler broadening are all included, while a Gaussian function is considered for rotational broadening. Stellar parameters are determined using a spectrum synthesis procedure that includes the Balmer series lines from H to H (excepting H , α 10 (cid:15) which is blended with a Ca II H line) and nine He I lines, i.e., 4009 ˚A, 4026 ˚A, 4120 ˚A, 4143 ˚A, 4388 ˚A, 4471 ˚A, 4922 ˚A, 5876 ˚A, and 7065 ˚A. For CD14-A08, He II (4686 ˚A) is also considered because the line is sensitive to T . eff The best fitted stellar parameters are found by χ2 test. n Fig. 2 the sensitivity to T is shown by displaying synthetic spectra at ±∆T from the best-fit value of T . eff eff eff ∆T = 2000 K for CD14-A08, CD14-A12, and CD14-B14. While for the rest stars, eff Tbest − 2000 K is too low to generate He I profiles correctly, therefore, in this case, eff ∆T = 1500 K. In Fig. 3 the sensitivity to logg is illustrated by showing spectra at logg eff values ±0.2 dex from its best-fit value. It can be seen that T and logg values are sensitive eff to He I lines and Balmer series lines, respectively. We first used the same method as the one described in Moni Bidin et al. (2012, hearafter MB12) to estimate the errors on T , eff 4http://nova.astro.umd.edu/Synspec49/synspec.html 5http://nova.astro.umd.edu/Tlusty2002/tlusty-frames-data.html – 9 – logg and vsini. Specifically, the uncertainties estimated from the χ2 statistics (labeled as σ ) were multiplied by three to obtain the final error, since the errors propagated from χ2 the data reduction procedure, such as sky subtraction and the normalization, can’t be neglected. The details are described in MB12 and references therein. But the errors of T eff and logg are also affected by the noise level of spectra. In order to take this into account, the uncertainties propagated from the S/Ns (labeled as σ ) were estimated by resampling S/N the data (Andrae 2010), which is a Monte-Carlo method. For each data point F , we λ resample a new F(cid:48) from a Gaussian distribution with a mean value of F and a standard λ λ deviation σ , which is calculated from σ = Fλ . Then T (logg) of the best fit for the λ λ S/N eff new resampled spectrum is re-determined with fixing logg (T ), log NHe and vsini. The eff NH resampling and fitting processes are repeated 50 times for each star. Thus, a series of T eff (logg) for each star is obtained. Therefore, σ is given by the standard deviation of the S/N series. The uncertainties with typical S/N values are listed in Tab. 3. In conclusion, the final errors on T and logg are estimated by summing 3σ and σ in quadrature. For eff χ2 S/N the error of log NHe, we refer the reader to § 4.2. Finally, the stellar parameter results and NH their errors are listed in Tab. 4. 3.2. Comparisons with a MB12-like analysis In order to test the reliability of the stellar-parameter results, We also perform a separate, independent measuring process using different grids of model atmospheres and analysis code (specifically, that in MB12) for the stellar parameters. In the work of MB12, the grid of model spectra is computed with Lemke’s version6 of the LINFOR program (developed originally by Holweger, Steffen, and Steenbock at Kiel University), which is based 6http://a400.sternwarte.uni-erlangen.de/~ai26/linfit/linfor.html – 10 – on local thermodynamic equilibrium (LTE) model atmospheres of ATLAS9 (Kurucz 1993). The grid covers 7000 K < T < 35000 K, 2.5 < logg < 6.0, and −3.0 < log NHe < −1.0. eff NH For CD14-A08, whose T is higher than 40000 K (from the intermediate-resolution spectral eff study of CD14), a grid of metal-free NLTE model atmospheres of Moehler et al. (2004) is employed. In this analysis, Balmer series from H to H12, four He I lines (4026 ˚A, 4388 ˚A, β 4471 ˚A, 4922 ˚A), and two He II lines (4542 ˚A and 4686˚A) were fitted simultaneously. For more details on the grids of model atmospheres and analysis code, we refer the read to MB12 and references therein. Comparisons of the atmospheric parameters between the two analyses are presented in Tab. 5 and Fig. 4. For convenience, we label our default spectral analysis process described in the previous subsection as “SA I”, and the separate one ´a la MB12 as “SA II”. The comparison reveals a general good agreement between the two parameter sets, apart from two problematic cases. Specifically, the nature of the initially suspected hot O-type star CD14-A08 is still obscure (see Sect. 5.1.1), and the measurements for this star were challenging, while the SA II temperature of CD14-A05 is lower than the cooler end of the BSTARS grid, hence a direct comparison of the results is doubtful. With the exclusion of these two stars, the mean difference is only 74 K in T and −0.08 dex in logg. The offset eff in logg seems systematic, although negligible compared to errors. SA II logg estimates tend to be lower than SA I values by ≈ −0.1 dex. Rotational velocities are in good agreement (mean difference 8 km s−1), but it must be reminded to the reader that SA II estimates are relatively rough, because vsini in this method is an input value of the routine and not a fit parameter. The helium abundance shows no systematic offset between the two parameter sets, but the differences are very large in some cases. However, the SA II results may not be very reliable, because the upper end of the employed grid is at solar abundance (log NHe = −1), while most of the stars (six, according to SA I) have super-solar values. NH