The Runaways and Isolated O-Type Star Spectroscopic Survey of the SMC (RIOTS4)
J. B. Lamb, M. S. Oey, D. M. Segura-Cox, A. S. Graus, D. C. Kiminki, J. B. Golden-Marx, J. Wm. Parker
aa r X i v : . [ a s t r o - ph . GA ] D ec ***Accepted for publication in ApJ *** Preprint typeset using L A TEX style emulateapj v. 5/2/11
THE RUNAWAYS AND ISOLATED O-TYPE STAR SPECTROSCOPIC SURVEY OF THE SMC (RIOTS4) * J. B. Lamb , M. S. Oey , D. M. Segura-Cox , A. S. Graus , D. C. Kiminki , J. B. Golden-Marx ,and J. Wm. Parker ***Accepted for publication in ApJ *** ABSTRACTWe present the Runaways and Isolated O-Type Star Spectroscopic Survey of the SMC (RIOTS4),a spatially complete survey of uniformly selected field OB stars that covers the entire star-formingbody of the SMC. Using the IMACS multislit spectrograph and MIKE echelle spectrograph on theMagellan telescopes, we obtained spectra of 374 early-type field stars that are at least 28 pc fromany other OB candidates. We also obtained spectra of an additional 23 field stars in the SMC baridentified from slightly different photometric criteria. Here, we present the observational catalog ofstars in the RIOTS4 survey, including spectral classifications and radial velocities. For three multi-slit fields covering 8% of our sample, we carried out monitoring observations over 9–16 epochs tostudy binarity, finding a spectroscopic, massive binary frequency of at least ∼
60% in this subsample.Classical Oe/Be stars represent a large fraction of RIOTS4 (42%), occurring at much higher frequencythan in the Galaxy, consistent with expectation at low metallicity. RIOTS4 confirmed a steep upperIMF in the field, apparently caused by the inability of the most massive stars to form in the smallestclusters. Our survey also yields evidence for in-situ field OB star formation, and properties of fieldemission-line star populations, including sgB[e] stars and classical Oe/Be stars. We also discuss theradial velocity distribution and its relation to SMC kinematics and runaway stars. RIOTS4 presents afirst quantitative characterization of field OB stars in an external galaxy, including the contributionsof sparse, but normal, star formation; runaway stars; and candidate isolated star formation.
Subject headings: galaxies: Magellanic Clouds – galaxies: stellar content – stars: early-type – stars:emission-line, Be – stars: fundamental parameters – binaries: spectroscopic – stars:kinematics INTRODUCTION
The standard model of star formation has been thatmost, if not all, stars form in clusters (e.g., Lada & Lada2003), with the most massive stars aggregating in thedense cores of clusters. It is intuitive that massive Ostars form preferentially from the plentiful gas reservoirsof giant molecular clouds (GMCs). However, another sig-nificant population of massive stars exists in an environ-ment of the opposite extreme. These massive stars are farremoved from dense clusters or OB associations and in-stead appear isolated within the sparse field population.The physical properties and origin of this field massivestar population remain unclear, despite the fact that itaccounts for 20–30% of the massive stars in star-forminggalaxies (Oey et al. 2004). The existence of such stars inisolation poses a challenge for theories of massive star for-mation, which suggest that the necessary gas conditionsare primarily or exclusively found in GMCs. Alterna-tively, rather than having formed in the field, these stars * This paper includes data gathered with the 6.5 meter Mag-ellan Telescopes located at Las Campanas Observatory, Chile. Astronomy Department, University of Michigan, 1085 S.University Ave., Ann Arbor, MI 48109-1107 Department of Physical Sciences, Nassau Community Col-lege, One Education Drive, Garden City, NY 11530 Department of Astronomy, University of Illinois, Urbana, IL61801 Department of Physics and Astronomy, University of Cali-fornia, Irvine, CA 92697 Department of Astronomy, University of Arizona, Tucson,AZ 85721 Southwest Research Institute, Department of Space Studies,Suite 300, 1050 Walnut Street, Boulder, CO 80302-5150, USA may have formed in clusters, and then been subsequentlyejected from their birth locations as runaway stars. Ineither case, field massive stars are a unique, understud-ied subset of a galaxy’s massive star population, probingboth extremely sparse and extremely dense star-formingconditions.The observational evidence for in situ field massivestar formation has grown in recent years. An opticaland UV photometric census of candidate O-type stars ina portion of the LMC suggests that approximately halfof these stars may be members of the field population(Parker et al. 2001). Some strong, direct evidence of for-mation in the field is work by Testi et al. (1997, 1998),who reported a sample of Herbig Ae/Be stars forming inisolation. At higher masses, Lamb et al. (2010) detectedsparse clusters associated with field OB stars in the SmallMagellanic Cloud, and Bressert et al. (2012) identified 15O stars that are candidates for isolated formation near30 Doradus, based on a variety of criteria. Additionalindividual candidates have been reported by Selier et al.(2011) and Oskinova et al. (2013). Oey et al. (2013) pre-sented a sample of 14 field OB stars centered in circularHII regions, thus implying that they are unlikely to havetransverse runaway velocities. Since these objects fur-thermore have non-runaway radial velocities, they mostlikely formed in situ. This growing observational datasetof massive stars that appear to have formed in sparseclusters or in isolation, without any indication of beingrunaways, strongly suggests that some component of thefield massive star population formed in situ. Even so,formation within clusters cannot be entirely ruled out Lamb, J. B., et al.for these stars. Gvaramadze et al. (2012) point out thatcluster dissolution, slow ejections, or multi-stage ejec-tions could all potentially mask the signatures that thesestars formed in clusters.This problem on the origin of field OB stars is centralto some outstanding controversies. Weidner & Kroupa(2006) suggest a deterministic relation between clustermass and the associated maximum stellar mass; whereasif it is indeed the case that massive stars can form insparse, low-mass clusters, it would suggest a large disper-sion in the relation between cluster mass and the asso-ciated maximum stellar mass, which is inconsistent withsuch a scenario. Furthermore, it would also imply thatindividual sparse clusters must necessarily have stellarinitial mass functions (IMFs) that grossly deviate fromany standard values. It remains unclear whether suchdeviations are real or whether they arise from stochas-tic sampling, so that an aggregate population of sparseclusters would yield a Salpeter-like IMF, as suggested byLamb et al. (2010).These issues are simply a consequence of the diffi-culties in understanding sparse massive star formationwithin the framework of current star formation mod-els. Two primary theories for massive star formationare the competitive accretion model and the core accre-tion model. In the competitive accretion model, molec-ular clouds fragment into star forming cores, which con-tinue to accrete matter from a shared reservoir of gas.In this scenario, massive stars form in locations wherethe gas density is highest, which is typically in the cen-ters of GMCs (Zinnecker 1982). Thus, it is implicit tothe competitive accretion model that massive stars mayonly form along with a significant population of lowermass stars (Bonnell et al. 2004). In contrast, core ac-cretion models suggest that the gas available for accre-tion is controlled by the mass of the fragmented coreitself (Shu et al. 1987). Thus in core accretion modelsit is possible, although difficult, to obtain gas conditionsthat would allow a massive star to form in isolation (e.g.Krumholz et al. 2009).A less controversial component of the field is therunaway population. Observationally, isolated massivestars with large space velocities are well-known to ex-ist. The typical definition for runaway stars is a pecu-liar space velocity >
30 km s − . Using this definition,runaway fractions ranging from 10% (Blaauw 1961) to50% (de Wit et al. 2005) have been observed for mas-sive stars within the Galaxy. However, other studies useevidence from bow shocks, the likelihood of slow run-away ejections, and the possibility of exotic multi-stageejection mechanisms to suggest that the true runawayfraction is much higher, up to 100% of the field popula-tion (Gvaramadze et al. 2012). In this scenario, the fieldpopulation is comprised primarily of stars that formed indense cluster cores, where the best conditions for massivestar ejections exist. Thus, the field population is a vitalprobe of the massive star formation process at both thedensest and least dense extremes.Other than the obvious kinematic signatures expectedfor runaway stars, it is not well known how the propertiesof massive stars formed in isolation vs runaways woulddiffer from stars in clusters. Observational studies do re-veal a few trends: for example, a study by van den Bergh(2004) compares the distribution of spectral types be- tween field and cluster O stars within the magnitude-limited Galactic O Star Catalog (Ma´ız-Apell´aniz et al.2004), finding that spectral types for field stars areskewed toward later types than stars in clusters. Thus,field O stars are either older or less massive as a popula-tion than O stars in clusters. A similar result was foundin the Magellanic Clouds, where Massey et al. (1995) andMassey (2002) discovered that the field population hasan extremely steep IMF in a few selected fields. The stel-lar IMF for stars in clusters is generally consistent withthe classical Salpeter slope of Γ = 1 .
35 for a power lawgiven by dn/d log m ∝ m − Γ , where n is the number ofstars of mass m . However, Massey et al. (1995) found ahigh-mass field IMF slope of Γ ∼ spatially complete and statistically sig-nificant sample of uniformly selected field massive stars.Here, we present an overview of the RIOTS4 survey andthe results to date. RIOTS4 TARGETS AND OBSERVATIONS
RIOTS4 targets a spatially complete sample of 374 uni-formly selected candidate field OB stars in the SMC.Our targets are identified by Oey et al. (2004; hereafterOKP04) according to the photometric criteria B ≤ . Q UBR ≤ − .
84, where the reddening-free parameter Q UBR is given by, Q UBR = ( U − R ) − A U − A R A B − A R ( B − R )= ( U − R ) − . B − R ) , (1)where the A values correspond to extinction in the speci-fied bands. In the calculation of Q UBR , OKP04 adoptedthe ratio of total to selective extinction R V = 3.1 fromCardelli et al. (1989). These photometric criteria weredesigned to select stars with masses & M ⊙ , using the B magnitude to eliminate less massive main sequenceIOTS4 Survey 3stars, and the Q UBR criterion to identify only the blueststars; this corresponds to approximate spectral types ofB0 V, B0.5 I, and earlier. OKP04 applied these criteriato the
U BV R photometric survey data for the SMC ob-tained by (Massey 2002), which was optimized to iden-tify OB star candidates. This survey basically coveredthe full star-forming expanse of the galaxy, which en-sures uniform selection of a spatially complete sample ofmassive stars in the SMC. OKP04 further carried outa friends-of-friends analysis on this sample to identifyclusters. In this algorithm, stars are considered clus-ter members if their projected distances to other clustermembers are smaller than the given clustering length.The clustering length is the value that maximizes thenumber of identified clusters (Battinelli 1991), which is28 pc for the SMC sample. Thus the field OB targets forthe RIOTS4 survey correspond to all candidates from theOKP04 sample with no other candidates within a 28 pcradius.OKP04 also identified a sample of candidate field Ostars in a smaller region, covering the SMC bar, usingUV photometry from the
Ultraviolet Imaging Telescope(UIT) (Parker et al. 1998). These 91 field O star candi-dates were selected using reddening-free indices that in-clude UV and
U BV R photometry, along with the same B magnitude criteria as the main sample. Of these 91stars, there are 23 that were not identified by the opticalphotometric criteria above. We included these stars inour multi-object observations as described below.We observed the RIOTS4 survey targets over a five-year period from 2006 September to 2011 October usingspectrographs on the Magellan telescopes at Las Cam-panas Observatory. The majority of our observationswere obtained with the Inamori-Magellan Areal Cameraand Spectrograph (IMACS) in the f/4 multi-slit mode onthe Magellan Baade telescope (Bigelow & Dressler 2003).With 49 slit masks, we observed 328 of the 374 candidatefield OB stars, or over 7/8 of our total sample. We alsoobserved the 23 objects unique to the UV-selected sam-ple with this setup. We used the 1200 lines/mm gratingand slit widths of either 0.7 ′′ or 1.0 ′′ , yielding spectralresolutions of R ∼ R ∼ >
30 for our fainter targets. All observationsin our IMACS multi-object campaign occurred between2006 September to 2010 December. During our initialobserving run in 2006 September one of our 49 fields wasobserved with the 600 lines/mm grating, resulting in aspectral resolution of R ∼ ′′ slit width for a spectral resolution of R ∼ >
30. All MIKE ob-servations occurred in 2010 November. With IMACS f/4out of commission during our 2011 observations, we alsooperated IMACS in f/2 mode with a 300 lines/mm grismto observe a total of 27 objects. Depending on the see-ing, we used either a 0 . ′′ or 0 . ′′ slit width, which yieldspectral resolutions of R ∼ R ∼ § <
24 hoursto days, weeks, months, and years. Since these fieldsoverlap in area, a few stars were observed with up totwice as many observations.Initial reduction of RIOTS4 IMACS multi-slit obser-vations was completed with the Carnegie Observato-ries System for MultiObject Spectroscopy (COSMOS)data reduction package . COSMOS was custom de-signed for use with the IMACS instrument and 8-CCDarray setup. With COSMOS, we performed bias sub-traction, flat-fielding, wavelength calibration, and ex-traction of 2-D spectra following the standard COSMOSpipeline. For single-star spectra from MIKE and IMACS,we used standard IRAF procedures to do bias subtrac-tion, flat fielding, and wavelength calibration. Fromthe wavelength-calibrated 2-D spectra for both singlestar observations and multi-slit observations, we usedthe apextract package in IRAF to find, trace, and ex-tract 1-D spectra. We rectified the spectra using the continuum procedure and eliminated remaining cosmicrays or bad pixel values with the lineclean procedure,both of which belong to the onedspec package in IRAF. RIOTS4 DATA PRODUCTS
Catalog of Spectral Types
The first observational data product from RIOTS4is the catalog of spectral classifications for candidatefield OB stars. The completeness of RIOTS4 al-lows a full characterization of the distribution of stel-lar spectral types in the field. We classify the starsbased primarily on the atlas of OB spectra publishedby Walborn & Fitzpatrick (1990), and we also rely onWalborn et al. (2009) and Sota et al. (2011), especiallyfor identification of unique spectral features. However,these atlases present mostly Galactic stars at solar metal-licity ( Z ∼ . COSMOS was written by A. Oemler, K. Clardy,D. Kelson, G. Walth, and E. Villanueva. Seehttp://code.obs.carnegiescience.edu/cosmos. IRAF is distributed by the National Optical Astronomy Ob-servatory, which is operated by the Association of Universities forResearch in Astronomy (AURA), Inc., under cooperative agree-ment with the National Science Foundation (NSF).
Lamb, J. B., et al.metallicity ( Z ∼ . II λ I λ II λ I(+II) λ IV λ III λ II λ ∼ O8, we look for thepresence of emission features such as N II λλ II λ IV λ I λ III λ I λ C iii λ .
9; Harries et al. 2003) with the observed magnitudefrom Massey (2002). If the star is much brighter than ex-pected for its luminosity class, then we re-visit our lumi-nosity classification and adjust it to a more evolved classin more ambiguous cases. However, the existence of a bi-nary companion would also increase the observed bright-ness of an object. Therefore, we carefully re-examinesuch stars for evidence of spectroscopic binary compan-ions. Even so, detection of a secondary may often go un-noticed without multi-epoch observations, or they maybe unresolvable due to low inclination angle, small massratio, or long periods. Thus, undetected binaries may beexpected to bias our results slightly towards later spec-tral types and more evolved objects. In general, there is atendency that the magnitudes indicate brighter luminos-ity classes than derived spectroscopically; this is relatedto the known effect that SMC OB stars are observedto lie above theoretical evolutionary tracks on the H-Rdiagram, as discussed by, e.g., Lamb et al. (2013) andMassey (2002). However, for Be stars, we find more ex-treme discrepancies in luminosity class, and we thereforeomit these from the spectral classifications of Be stars inour catalog.The fraction of our objects that are undetected bina-ries is likely to be significant; we obtain a lower limitto the binary fraction of ∼
60% in the RIOTS4 multi-epoch campaign (see § III λ III lines.Another stellar population that creates issues for spec-tral typing is emission-line stars. In RIOTS4, this in-cludes classical Oe/Be stars, supergiant B[e] (sgB[e])stars (Graus et al. 2012), and Wolf-Rayet (WR) stars.These stars are often partially or wholly enshrouded incircumstellar disks or envelopes whose emission is su-IOTS4 Survey 5
Figure 1.
A sequence of spectral types from O4 V to O8 V stars from the RIOTS4 survey. We label the major spectral features in therange from 4000 − λ λ perimposed on the photospheric spectra. This resultsin weakened or absent absorption lines, which can dras-tically alter spectral types or make them impossible todetermine. Lesh (1968) classifications for Oe/Be starswere carried out by JBGM. In § § i linesin Oe stars (e.g., Negueruela et al. 2004), and Si iii λ iv λ ∼ R ∼ ∼
75 and R ∼ ∼
45 and R ∼ γ to determine luminosity classes due tothe relatively poor spectral resolution and S/N of theirdata. Coupling this with the expected high binary frac-tion and our careful treatment of binaries may explainthe differences. Stellar Radial Velocities and Multi-EpochObservations
Another important RIOTS4 data product is the mea-surement and distribution of radial velocities for SMCfield OB stars. Radial velocities are an important prop-erty of a stellar population, both for individual objectsand as an ensemble. Since runaways are a well knowncomponent of the field population, in principle, we canidentify many such objects using their radial velocities.For the field stars as a whole, the velocity distributionand dispersion probe the kinematics of this populationand on a large scale, the bulk motions of the SMC. For Lamb, J. B., et al.
Figure 2.
A sequence of spectral types from O9 V to B1.5 V from the RIOTS4 survey. We label the major spectral features in the rangefrom 4000 − λ λ multi-epoch observations, variability in the radial veloc-ity is a strong indicator of a massive binary system.We measure the radial velocities of RIOTS4 targetsusing the rvidlines package in IRAF. Velocities are ob-tained by fitting gaussian profiles to a combination ofH, He I , and He II absorption lines. We require a min-imum of 3 lines to determine the radial velocity, to en-sure that continuum fitting issues or odd line profilesdo not affect our measurements. Lines with velocitiesthat significantly deviate from all other lines for a sin-gle star are excluded from the radial velocity measure-ment. These spurious velocities are typically associatedwith lines close to the IMACS chip gaps, which can af-fect the continuum fitting and, therefore, the line profile.The uncertainties on our radial velocity measurementsare ∼ − for MIKE observations, ∼
10 km s − forIMACS f/4 observations, and ∼
25 km s − for IMACSf/2 observations.Since massive stars have a high binary frequency, itis likely that a large fraction of our radial velocity mea-surements are affected by variability. Thus, single-epochradial velocity measurements may cause erroneous iden-tification of binary systems as runaway stars. This vari-ability also adds scatter to the distribution of radial ve-locities for the full population. Our multi-epoch observa- tions are meant to address the magnitude of these effectsby measuring the scatter and estimating the field binaryfraction for 8% of the RIOTS4 sample ( § ∼ − . RESULTS
Stellar Catalog
Table 1 presents the basic catalog of the 374 objectsin the RIOTS4 survey. In columns 1 – 3, we list thestellar ID numbers and
B, V magnitudes from Massey(2002), respectively; column 4 contains the reddening-free Q UBR calculated by Oey et al. (2004). In column5, we provide an extinction estimate using the SMC ex-tinction maps from Zaritsky et al. (2002). Column 6 con-tains the spectral classification derived from the RIOTS4data. Columns 7 and 8 list our measured radial veloc-ity of the star and the radial velocity of the nearest (invelocity space) H I kinematic component with brightnesstemperature >
20 K (see § V = 13 .
0, and0.04 at V = 15 .
0. Table 2 provides the same data for theIOTS4 Survey 7
Figure 3.
A sequence of evolved stars from O6 to B1.5 from the RIOTS4 survey. We label the major spectral features in the range from4000 −
23 additional stars we observed from the UV-selectedsample. In what follows, we consider only the original,optically selected sample so that our analysis is appliedstrictly to a uniformly selected sample. However, giventhat there are 23 additional stars out of 91 identified withthe alternate criteria, we can infer that our base sampleis incomplete at least at the 25% level for identifying allactual OB stars.
Field IMF
Previous studies of the field massive star IMF in theMagellanic Clouds indicate a slope steeper than the tra-ditional Salpeter slope of Γ = 1 .
35. The observed slopesrange from Γ = 1.80 ± .
09 (Parker et al. 1998) to Γ ∼ ± . > M ⊙ , we followKoen (2006) to derive the cumulative mass distribu-tion for the SMC field and compare it with evolvedpresent-day mass functions from Monte Carlo modelswith ages up to 10 Myr, the lifetime of 20 M ⊙ stars.Using this method, we estimate that the field massivestar IMF slope is Γ=2.3 ± . − M ⊙ stars, using astochastic approach that models the uncertainties in stel-lar positions on the H-R diagram. With further MonteCarlo modeling, we determine that undetected binariesor a unique star formation history are unable to explainthis steep field IMF. Thus, we conclude that the steepobserved IMF is a real property of the SMC field. In § Figure 4.
A sample of synthetic binary spectra derived from actual RIOTS4 spectra. We label the major spectral features in the rangefrom 4000 − Table 1
RIOTS4 Catalog a ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)107 14.96 15.00 -0.95 0.82 Be – – MIKE 111024298 15.18 15.12 -0.91 1.03 B1e – – IMACS f/4 0709201037 15.15 15.28 -0.85 0.44 B0.5 V 110 110 IMACS f/4 0709201600 14.42 14.60 -0.87 0.32 O8.5 V 93 103 IMACS f/4 0709201631 15.19 15.15 -0.99 1.11 B1e
120 120 IMACS f/4 070920 a This table is published in its entirety in the electronic edition of the
Astrophysical Journal . A portionis shown here for guidance regarding its form and content. b From Massey (2002). c From Zaritsky et al. (2002). d Measured from Stanimirovi´c et al. (1999).
IOTS4 Survey 9
Table 2
Additional UV-Optically Selected Stars in the SMC BarID a B a V a Q UBR A V b Sp Type RV star RV HIc
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)5391 13.36 13.31 -1.00 1.13 O8.5 III 44 98 IMACS f/4 060913 d d d d - - IMACS f/4 09082515690 14.05 14.07 -0.99 0.89 O6 V((f)) 80 120 IMACS f/4 09082417963 15.12 15.21 -0.99 0.55 B0.2 V 115 120 IMACS f/4 09082418200 14.33 14.33 -0.87 0.90 B0e
111 120 IMACS f/4 09082424982 14.75 14.94 -0.85 0.26 O8 V 110 110 IMACS f/4 060913 d d d - - IMACS f/4 06091238302 14.64 14.81 -0.84 0.29 B1 V 154 154 IMACS f/4 09082540341 13.77 13.98 -0.92 0.24 O8.5 III((f)) - - IMACS f/4 09082541095 14.84 14.85 -0.92 0.92 O9.5-B0 V + Be - - IMACS f/4 06091144634 15.19 15.37 -0.85 0.27 O9.5-B0 V 150 150 IMACS f/4 09082545677 13.52 13.66 -0.92 0.47 O9.5 III 160 164 IMACS f/4 09082548672 14.34 14.52 -0.93 0.36 O7.5 V - - IMACS f/4 09082453373 14.08 14.20 -0.84 0.51 O9 V 119 122 IMACS f/4 090824 a From Massey (2002). b From Zaritsky et al. (2002). c Measured from Stanimirovi´c et al. (1999). d Observed multiple times for binary monitoring; see Table 3.
In Situ Formation of Field O Stars
As outlined earlier, the origin of the field massive starpopulation is an open question. In particular, it is un-known whether massive stars are capable of forming inisolation or within sparse clusters. Some theories ofmassive star formation, such as competitive accretion,suggest that the most massive star formed in a clus-ter depends on the cluster mass (Bonnell et al. 2004).Other theories, such as those based on core accretion,allow for the formation of massive stars in sparse en-vironments, or even in isolation (e.g., Krumholz et al.2009). The essential question is whether the formationof massive stars in sparse environments is merely improb-able (e.g., Elmegreen 2000) or actually impossible (e.g.,Weidner & Kroupa 2006).Using RIOTS4 spectra, along with data from the
Hubble Space Telescope (HST) (Lamb et al. 2010) andOGLE photometry (Udalski et al. 1998), we identify asample of unusually strong candidates for in-situ, fieldOB star formation. Lamb et al. (2010) discover threemassive stars that formed in sparse clusters containing ∼
10 or fewer companion stars with mass > M ⊙ andanother three candidates for truly isolated formation.Oey et al. (2013) present a sample of 14 field OB starsthat are centered on symmetric, dense H ii regions, whichminimizes the likelihood that these objects have trans-verse runaway velocities. In both studies, the RIOTS4spectra eliminate line-of-sight runaways, leaving strongcandidates for field massive stars that formed in situ. Weset further constraints on the degree to which these starsare isolated by examining their immediate stellar envi-ronments with the
HST and OGLE imaging, allowing usto evaluate the relationship between the most massivestars in any sparse clusters and the cluster mass. Ourresults imply that these two quantities are independent,and thus they favor the core collapse models for massivestar formation.
Radial Velocity Distribution
The distribution of radial velocities reveals informa-tion about the stellar population kinematics, as wellas the bulk motion of the SMC. Our velocity distribu-tion from RIOTS4 is generally consistent with earlierwork; Figure 5 is qualitatively similar to that found fromthe 2dF survey of OBA-type stars in the SMC foundby Evans & Howarth (2008). Both samples exhibit agaussian-like velocity distribution with a FWHM of ∼ − and a mean systemic velocity of ∼
150 km s − .As mentioned earlier, radial velocities for individual starsmay be affected by binary motions, and so we can onlymake inferences based on aggregate trends. We do seeevidence of a velocity gradient across the SMC, whichwe depict in Figure 6 by plotting velocity distributionsof three regions in the SMC. The Bar 1 and Bar 2 regionshave mean velocities of 140 km s − and 157 km s − , re-spectively, with corresponding respective velocity disper-sions of 32 km s − and 39 km s − . Note that althoughwe bisect the bar into two regions, it appears to have arelatively smooth velocity gradient. The SMC wing ismore redshifted than the SMC bar, having a mean ve-locity of 177 km s − with velocity dispersion of 29 km s − , but it does not appear to have a significant internalvelocity gradient. These observations of the large-scalemotions in the SMC agree with results based on stars inthe 2dF survey and on H i gas from Stanimirovi´c et al.(2004).
Runaway Stars
Runaway stars are a well-known component of thefield population, yet their relative contribution to thefield and ejection mechanisms from clusters remainpoorly understood. Observational estimates for therunaway frequency range from 10% (Blaauw 1961) to50% (de Wit et al. 2005), while some authors argue that all field massive stars are runaways (Gvaramadze et al.2012). One trend that seems to have emerged is thatO stars have a 2 − ∼
200 km s − , while dynamical ejectionscan attain higher velocities (Gvaramadze et al. 2009, andreferences therein). Both ejection scenarios are predictedto include binary runaways; however, the type of bina-ries differ significantly. For the binary supernova sce-nario, the compact object remnant of the primary starsometimes remains bound to the secondary as a runawaybinary system with an eccentric orbit (McSwain et al.2007). For dynamical ejections, tight binaries are some-times ejected as a single system, thus representing theonly mechanism that can form a runaway double-linedspectroscopic binary. Finally, while both mechanismsoriginate from binary systems, stars ejected from the bi-nary supernova scenario may be He-rich due to contam-ination from the supernova explosion (Hoogerwerf et al.2001).To estimate the fraction of runaway stars in the RI-OTS4 sample, we compare the observed stellar radial ve-locities of our OB stars with the H i gas velocity distribu-tion along the line of sight, using data from the AustraliaTelescope Compact Array (ATCA) and Parkes telescopescompiled and mapped by Stanimirovi´c et al. (1999). Weidentify runaway candidates as those objects with radialvelocities that are different by >
30 km s − from thoseof the nearest H i velocity components having a bright-ness temperature >
20 K in the same line of sight. Apair of examples are shown in Figure 7, with star 35491depicting an object consistent with the line-of-sight
H i gas velocity, and star 43724 meeting our criteria for arunaway star. We find that only 11% of the stars meetthese runaway criteria, 27 out of 238 stars with good ra-dial velocity determinations. This frequency is likely tobe overestimated due to false positives caused by binaryIOTS4 Survey 11
Figure 5.
The distribution of radial velocities from stars in the RIOTS4 survey.
Figure 6.
We split the RIOTS4 sample into three regions of the SMC, as shown in the upper left panel. Stars in our binary fields areplotted with asterisks while all other stars are plotted as crosses. In the other three panels, we plot the radial velocity distribution for starsin each separate region. The clear velocity gradient of RIOTS4 stars across the SMC agrees qualitatively with the velocity gradient of HIgas from Stanimirovi´c et al. (2004). motions, since the measured radial velocity may reflectthe orbital motion for a binary star, rather than the sys-temic velocity. While such motions will also sometimescause false negatives depending on the orbital configura-tion at the time of observation, false positives are morelikely to be observed. A more significant effect is thatradial velocities can only identify line-of-sight runawaymotions. We estimate that our observations miss 50%of runaways if the typical ejection velocity is ∼
60 km s − . Since only 8% of our survey has multi-epoch obser-vations, we are not yet able to correct for the effect ofbinaries on the stellar population kinematics. Therefore,we have initiated further follow-up, binary monitoringobservations to further minimize these degeneracies.We do find one runaway, star 5391, that we identify asa binary star from our multi-epoch observations ( § − is 55 km s − removedfrom the nearest significant component of H i gas. With2 Lamb, J. B., et al.
Figure 7.
Position-velocity diagrams for H I in the line of sight for two RIOTS4 stars, showing data from Stanimirovi´c et al. (2004).The solid, vertical line depicts the observed radial velocity of the RIOTS4 target, while the dashed line shows our brightness temperaturethreshold of 20 K. Stars 35491 (left) and 43724 (right) are examples for which the stellar and gas velocities are consistent and inconsistent,respectively. a semi-amplitude of 108 km s − in our observed varia-tions, the secondary cannot be a degenerate star. There-fore, if this binary system is indeed ejected from a cluster,then it must be due to the dynamical ejection mechanismrather than the binary supernova mechanism. Since dy-namical ejection frequently splits binaries, the existenceof a non-degenerate, runaway binary suggests a majorcontribution by this process to the runaway population.Another interesting object also points to the impor-tance of dynamical ejection: Star 49937 appears to bean extreme runaway that is unlikely to be the productof the binary supernova scenario. Its runaway velocityis ∼
200 km s − removed from the nearest H i veloc-ity component. While it is possible that this star’s run-away component is completely in the line of sight and/orfortuitously enhanced by binary motion, its high radialvelocity, taken at face value, is near the maximum ejec-tion speed possible from the binary supernova mecha-nism (Portegies Zwart 2000), as mentioned above. Thus,the existence of this star again suggests a significant rolefor dynamical ejection of runaways.
Binary Stars
Stellar multiplicity is a key parameter that probes theformation and dynamical evolution of a stellar popu-lation. For example, large protostellar disks may bedisrupted in high-density environments, thereby sup-pressing the formation of massive binaries (Kratter et al.2008, 2010). Recent studies of Galactic clustersand OB associations find observed massive-star bi-nary fractions ranging from ∼
60% to ∼
80% (e.g.,Sana et al. 2008, 2009, 2011; Kiminki & Kobulnicky2012; Kobulnicky et al. 2014). However, few studies havesystematically investigated the multiplicity of massivestars in the field. Early studies found that field massivestars have roughly half the binary frequency of massivestars found in clusters (e.g., Stone 1981; Gies 1987). Thisgeneral trend of a lower field binary frequency persists inlater studies, such as Mason et al. (1998, 2009), who usespeckle interferometry of objects in the Galactic O StarCatalog (Ma´ız-Apell´aniz et al. 2004) to compare the fre-quency of multiplicity between cluster and field O stars.In this magnitude-limited sample, they find a 39% bi- nary fraction for field O stars, compared to a 66% binaryfraction for O stars in clusters. When combining theirresults with data from the literature on spectroscopicallyidentified binaries, they obtain 51% and 75% binary frac-tions for field and cluster O stars, respectively. However,the spectroscopic data for these objects is non-uniformand therefore may not provide an accurate comparison ofthese statistics between cluster and field O stars. But itdoes suggest that the frequency of multiplicity for mas-sive stars in the Galactic field is lower than in clusters.With the RIOTS4 survey, we performed repeat obser-vations of three IMACS slit-mask fields over the 5-yearsurvey period, totaling 29 objects, to obtain an initialevaluation of the binary fraction of field massive starsin the SMC. We note that some of these stars belongto the UV-selected sample (Table 2), rather than thedefault sample. We have 9 −
10 epochs for each field,at intervals of days, weeks, months, and years apart;three stars appear in two of the three fields, yieldingup to twice the number of observations for these objects.As with the larger survey, these fields have a high frac-tion of Oe/Be stars, and we focus here primarily on the17 non-Oe/Be stars in these fields. We use three sepa-rate methods to identify potential binaries, which are de-scribed below. The first method identifies binaries usingmaximum observed radial velocity variations, the secondmethod is based on a statistical F-test analysis follow-ing Duquennoy & Mayor (1991), and the third methoduses the period power spectrum and searches for binaryorbital solutions from the radial velocity data. Table 3summarizes this binary monitoring sample: columns 1and 2 give the star ID and spectral type, respectively;column 3 gives the number of observations, and columns4 and 5 show the star’s binary status determined fromthe second and third methods; we note that the firstmethod yields the same results as the second. Column 6gives the systemic velocity based on the orbital solution,if available, or the mean of the minimum and maximummeasured radial velocities. Column 7 gives the largestvelocity variation observed within a 14-day interval ∆ v ,and column 8 provides the standard deviation σ obs ofthe radial velocity measurements for each star. Column9 lists the calculated P ( χ ), which is used to determineIOTS4 Survey 13binary status in the statistical F-test ( § Maximum radial velocity variation and timescale
To identify binary star candidates, we first compare theamplitude of radial velocity variations with the timescaleof the variations. Since the amplitude of radial veloc-ity variations is inversely correlated with the period ofa binary system, binaries with large-amplitude variationshould display variability on short timescales, providedthe eccentricity of the system is near zero. In Figure8, we plot the amplitude of the maximum observed ra-dial velocity variation over short timescales ( <
14 days;Table 3) versus the amplitude of the maximum radialvelocity variation over any time scale. Note that in Fig-ure 8 all objects must lie at or below the dashed identityline. In an ideal scenario, all short-period systems will liealong this locus; however, we cannot expect good sam-pling with .
10 epochs of data. Nonetheless, we stillobserve a large fraction of high-variation systems alongthe identity line, which suggests there are no systematicvelocity offsets over time. Given the sampling of thesefields and our systematic errors, we conservatively iden-tify binaries as those objects with radial velocity varia-tions >
30 km s − including errors. This yields 10 prob-able binaries out of the 17 non-Oe/Be stars in our binarymonitoring fields. F-test: radial velocity variations relative to noise
We also use the approach of Duquennoy & Mayor(1991) who identified binary candidates in the nearbysolar neighborhood. This method compares the mean ofthe statistical measurement errors associated with eachradial velocity measurement ( σ ave ) with the standard de-viation in the measured radial velocities ( σ obs ; Table 3)for each star. For single objects with properly estimatedmeasurement errors, the ratio of σ obs / σ ave should ap-proximately equal unity. However, it is unclear wherethe cutoff ratio between single objects and binary starsshould occur. Thus, Duquennoy & Mayor (1991) use astatistical F-test to measure the probability P ( χ ) thatthe observed variations are due to statistical noise. Fol-lowing their work, we calculate χ , accounting for thenumber of observations n , with: χ = ( n − σ obs /σ ave ) . (2)Using the cumulative chi-square distribution given by F k ( χ ) = G ( k/ , χ /
2) (3)where G is the regularized Gamma function for a givendegree of freedom k = n −
1, we calculate P ( χ ) =1 − F k ( χ ), given in Table 3. In the case that all ob-jects are single, the distribution of P ( χ ) should be uni-form between values of 0 and 1. Binary systems, onthe other hand, should have very low values of P ( χ ),since their radial velocity variations are not due to sta-tistical noise. Thus, we can identify binaries as thoseobjects with P ( χ ) < .
01. We plot the distribution of P ( χ ) for the same 17 stars in Figure 9. Again, we finda high binary fraction with 10 out of 17 objects having P ( χ ) < . Period power spectrum
We used the radial velocities to search for credible or-bital solutions for all the stars in the binary monitoringsample based on the method described by Kiminki et al.(2012). In this approach, we generate the power spec-trum of periods for each object, and identify the mostlikely values, if any, with an IDL program created byA. W. Fullerton, which uses the CLEAN deconvolutionalgorithm of Roberts et al. (1987). We then apply theGudehus (2001) code for determining orbital solutions,the Binary Star Combined Solution Package, using thecandidate periods. We show the phase diagrams of thetwo best orbital solutions in Figure 10. These are forstars 10129 and 27600, with periods around 4.8 and 3.3days, respectively. Star 10129 appears to have a moder-ate eccentricity around e = 0 .
2, while 27600 is consistentwith a purely circular orbit. This approach again yields10 out of 17 probable binaries, although the identifiedcandidate binaries are not the exact same ones foundwith the preceding methods (Table 3).
Binary Fraction
All three binary identification methods suggest bina-rity in 10 out of 17 (59% ± ∼ −
50% as described above. However, the small num-ber statistics generate large errors, and our binary fre-quency is actually closer to values observed in Galacticclusters and OB associations. It is further difficult tocompare these frequencies because of the different obser-vational biases inherent in the different binary detectionmethods and sample properties; our frequencies are lowerlimits, representing results only for spectroscopic bina-ries. Sota et al. (2014) find a strong lower limit of 65%for the combined spectroscopic and visual binaries in thesouthern component of their Galactic O star survey. Al-most one quarter of these are identified exclusively byastrometric methods. We have started follow-up mon-itoring of additional RIOTS4 targets to confirm theseresults, and to obtain binary orbital parameters.We also applied the third binary identification methodto the remaining 12 stars in the monitoring fields, whichare classical Oe/Be stars. The radial velocities measuredfor these stars are more uncertain than for normal starsbecause of emission-line contamination in the H lines.We find that 6 out of the 12 Oe/Be stars appear to beprobable binaries.One of our binaries, star 27272, is a double-lined spec-troscopic binary (SB2) with B0.7 III and B star compo-nents (Figure 11). In our observations of this system, wefind that the stronger absorption line appears blueshiftedin all but 1–2 epochs. While this may be evidence of theStruve-Sahade (S-S) effect (Struve 1937; Sahade 1959),it is most likely caused by an unfortunate observing ca-dence, which impedes our ability to obtain a satisfactoryorbital solution.
Systemic Velocities
Estimated systemic velocities v sys for the 29 stars in themonitoring fields are given in Table 3. These are gener-ally given as the average of the minimum and maximum4 Lamb, J. B., et al. Table 3
Stars in binary monitoring fieldsID SpT N F-test Power Spec v sys (km s − ) ∆ v (km s − ) σ obs (km s − ) P ( χ ) Observation Dates a Normal OB Stars5391 O8.5 III 9 Y Y 44 144 75 < .
01 ABCEFGIJK6908 O9.5 – B0 III 9 Y N 128 93 25 < .
01 ABCEFGIJK6946 O9.5 V 9 N N 141 35 12 0.74 ABCEFGIJK7437 O6.5 I(f) 9 Y Y 151 29 33 < .
01 ABCEFGIJK7782 O8 V 9 Y Y 127 65 33 < .
01 ABCEFGIJK8257 B1.5 V 9 Y Y 96 61 21 < .
01 ABCEFGIJK8609 B0 III 9 N N 128 21 11 0.97 ABCEFGIJK10129 B0.2 V 9 Y Y 130 b
29 21 < .
01 ABCEFGIJK10671 B0.5 V 9 Y N 122 108 33 < .
01 ABCEFGIJK21844 O8 III((f)) 9 N N 151 36 13 0.09 BCDEFGHIK24213 B0 III 16 N N 126 6 9 0.99 ABCDEFGHIJK24982 O8 V 8 Y Y 110 59 32 < .
01 ADFGHIJK25912 O5 V 9 Y Y 150 103 45 < .
01 ADEFGHIJK27272 B0.7 III + B 9 Y Y 121 c
223 105 < .
01 ADEFGHIJK27600 B0.5 III 10 N Y 177 b
16 13 0.64 BCDEFGHIJK27712 B1.5 V 7 N N 127 8 7 0.46 ADFGHJK28841 B1 III 10 N Y 141 22 15 0.02 BCDEFGHIJKClassical Oe/Be Stars7254 O9.5 IIIe · · · Y 126 10 · · · · · ·
ABCEFGIJK21933 Be · · · N 130 57 · · · · · ·
AHIJ22321 O9.5 IIIpe · · · Y 167 b · · · · · · BCDEFGHIJK23710 O9–B0 pe · · · N 168 48 · · · · · ·
BCDEFGHIJK23954 B1.5e · · · N 130 69 · · · · · ·
ADFGHIJ24229 B1e · · · N 155 19 · · · · · ·
ADFGHJK24914 O9 III-Vpe · · · Y 81 20 · · · · · ·
AEHI25282 B0e · · · N 130 72 · · · · · ·
ABCDEFGHIJK25337 Be · · · Y 124 55 · · · · · ·
BCDEFGIJK27135 B1e · · · N 113 30 · · · · · ·
BCDEFGHIJK27736 B0e · · · Y 153 39 · · · · · ·
DEFGHJ27884 O7-8.5 Vpe · · · Y 156 32 · · · · · ·
BCDEFGHIJK a Dates of observation are coded as follows: (A) 2006 September 13, (B) 2007 September 19, (C) 2007 September 20,(D) 2008 September 24, (E) 2008 October 6, (F) 2008 October 7, (G) 2008 October 11, (H) 2008 November 21, (I) 2008November 22, (J) 2009 August 24, (K) 2010 December 20. b From orbital solution. c Average of SB2 components A and B.
Figure 8.
The observed maximum short-term ( <
14 days) radial velocity difference vs the largest radial velocity difference over anyperiod. The dashed line depicts the identity relationship. Objects with the highest observed velocity difference happening over a <
14 dayperiod will lie on this line. We expect real binary systems will exhibit velocity variations on both long and short term periods. Binariesidentified by having radial velocity variations >
30 km s − are plotted with a plus sign, while single stars are depicted as asterisks. IOTS4 Survey 15
Figure 9.
The distribution of P ( χ ) for non-Oe/Be stars in our binary fields. Binary systems that exhibit radial velocity variationssignificantly larger than expected from observational errors will have very low P ( χ ) values ( < . Figure 10.
Phase diagrams showing the solutions for two more securely identified binaries among the normal OB stars in the monitoringsample.
Figure 11.
The multi-epoch, RIOTS4 spectra of the double-lined spectroscopic binary 27272, with observation dates shown. of the N radial velocity measurements for each star. Forthree objects, more reliable values are available from fit-ted orbital parameters. The mean v sys is 131 km s − , ingood agreement with the value of 140 km s − for the Bar1 region, where these objects are located (Figure 6).Almost all of the stars in our monitoring sample have v sys within 2 σ of the mean. However, Star 5391 has v sys = 44 km s − , which is blueshifted by 87 km s − ,more than 3 σ from the mean and thus potentially a run-away star. This O8.5 III star is also identified as a binaryby our three methods (Table 3). Emission-line stars
A large fraction of our RIOTS4 stars turn out to beemission-line stars, mostly classical Oe/Be stars. We alsoidentify four B supergiant stars that exhibit forbiddenemission lines (Graus et al. 2012). One of these, star29267 (AzV 154; Azzopardi et al. 1975) was a previouslyknown sgB[e] star (Zickgraf et al. 1989). The other threestars are 46398, 62661, and 83480 (R15, R38, and R48,respectively; Feast et al. 1960). SgB[e] stars are normallydefined as stars exhibiting forbidden emission lines alongwith strong IR dust excess. However, this strong dustemission is not present in the three RIOTS4 stars newlyshown to be B[e] stars. In Graus et al. (2012), we discussthese objects in detail, demonstrating that they do showmore modest, free-free near-IR emission. We propose that they represent a new, transition class of dust-poorsgB[e] stars.There are two Wolf-Rayet stars included in the RI-OTS4 survey. They are stars 22409 and 30420, which areboth identified as WN3 + abs stars by Massey & Duffy(2001). In our RIOTS4 spectra, we detect only H ab-sorption lines for 22409, while 30420 also exhibits He II absorption (Figure 12). Massey & Duffy (2001) identifyHe II absorption in both objects and use the lack of He I to estimate that the absorption components correspondto O3-O4 stars.The rest of the emission-line stars are classicalOe/Be stars, comprising ∼
25% of the O stars(Golden-Marx et al. 2015) and ∼
50% of the B starsin the RIOTS4 survey. These objects exhibit emissionin their Balmer lines due to ‘decretion disks’ of ma-terial that are likely caused by rapid stellar rotation(e.g., Porter & Rivinius 2003). Oe/Be stars are morecommon at lower metallicities, with a Galactic Oe/O-star fraction of 0 . ± .
01 as measured from Galac-tic O Star Spectroscopic Survey (GOSSS; Sota et al.2011, 2014) and a 0 . ± .
09 Oe/O-star fraction inSMC clusters (Martayan et al. 2010). The denomina-tors here represent all O stars, including Oe stars. Simi-larly, the Be/B frequency of 30 −
40% in SMC clusters isabout twice the Galactic frequency (Maeder et al. 1999;Wisniewski & Bjorkman 2006). This metallicity effectIOTS4 Survey 17
Figure 12.
Spectra of the Wolf-Rayet and Be/X-ray binary stars in the RIOTS4 catalog. is consistent with the decretion disk scenario, since themetal-poor SMC stars have weak stellar winds, therebyimpeding the loss of stellar angular momentum throughthe winds. The high rotation rates therefore promotethe formation of decretion disks, leading to the Be phe-nomenon.Our RIOTS4 field Oe stars and their statistics are pre-sented by Golden-Marx et al. (2015), yielding an Oe/Oratio of 0 . ± .
04. We also find that the Oe spectral typedistribution extends to earlier types than in the Galaxy,both in terms of conventional classifications and hot Oestars whose spectral types are uncertain but are appar-ently of extremely early type. One extreme star, 77616,has He ii in emission from the disk, showing that eventhe hottest O stars can present the Oe/Be phenomenon;this supports theoretical models predicting that fast rota-tors can reach higher effective temperatures (Brott et al.2011). Our large sample of Oe stars in the SMC stronglysupports the metallicity effects predicted by the decre-tion disk model and characterizes the properties of earlyOe stars.Regarding the Be stars, the RIOTS4 Be/B fraction ap-pears to be even higher than found in previous studies.This result should be treated with caution because oursample selection criteria may be biased to favor selectionof Be stars. These objects emit strongly in H α , whichresults in a brightening of their R magnitude, thus low- ering Q UBR . Therefore, our sample selection criterionof Q UBR ≤ − .
84 is especially useful for selecting Bestars. Given our additional limiting B magnitude crite-rion, it is unclear whether our completeness limit for Bestars extends to later spectral types than normal stars,or whether it provides more complete identification of Bstars by including more Be stars down to the magnitudelimit. A comprehensive treatment of the Be stars, in-cluding detailed investigation of the selection effects andestimates of the luminosity classes, will be presented ina future publication. For now, we include Lesh (1968)classifications (Table 1) for these stars, which are a mea-sure of the magnitude of the Be phenomenon, and alsoindicate the presence of Fe II emission. In total, theOe/Be stars account for 157 of the 374 stars (42%) inthe RIOTS4 sample.We also observed three previously known Be/X-ray bi-nary systems within our survey, whose spectra are plot-ted in Figure 12. Object 52865 is reported to be a B0-0.5 III-Ve star in a binary system with a 967-s pulsar(Schurch et al. 2007; Haberl et al. 2008), and our spec-tral type for 52865 agrees with this spectral classification.Coe et al. (2012) report object 64194 to be a B0.5-1 Vestar in a binary system with a presumed neutron star, al-though no pulsar has been identified; we find a spectraltype of B0e for this star. Object 77458 is an eclipsingX-ray binary with a period of ∼ . . . DISCUSSION
The RIOTS4 survey provides a first, quantitative char-acterization of the field massive star population basedon a complete, uniformly selected sample of OB stars.It is also the first complete survey of field massive starsin an external galaxy. The resulting characterization ofthis population is necessarily sensitive to our definitionof field stars, recalling that our criterion requires thatmembers be at least 28 pc from other OB candidates, re-gardless of the presence of lower-mass stars. Thus, mostof our objects can be expected to represent the “tip ofthe iceberg” on low-mass clusters. On the other hand,we note that our 28-pc requirement is a more stringentcriterion for isolation than is often used in other stud-ies. This clustering length is derived from the spatialdistribution of the entire OB population and representsa characteristic value for the SMC as a galaxy (Oey et al.2004). In contrast, other studies often use more arbitrarydefinitions, for example, “field” OB stars in the vicinityof the 30 Doradus giant star cluster (Bressert et al. 2012)correspond to a different concept of field stars.Oey et al. (2004) showed that the clustering law forSMC OB stars follows an N − ∗ power law extending downto N ∗ = 1, which corresponds to our individual RIOTS4field OB stars, where N ∗ is the number of OB stars percluster. This basically confirms that most of our sam-ple corresponds to the “tip of the iceberg” objects, asexpected. However, as discussed in detail by Oey et al.(2004), the magnitude of the N ∗ = 1 bin does suggesta slight, but difficult to quantify, enhancement above asimple extrapolation of the power law distribution. Con-servatively, it is < . m up . This is at oddswith the steep upper IMF for the field stars found in § N ∗ for normal clusters, N ∗ & M cl & M ⊙ for a Kroupa (2001) IMF. Since typically m up & M cl here, it is apparent that in this regime itbecomes physically impossible, on average, to fully sam-ple the IMF up to m up . Therefore, the maximum stellarmasses in the sparsest clusters must necessarily be lower,on average, than in normal clusters. Since our RIOTS4sample is dominated by such stars in sparse clusters, thesteeper IMF is a natural consequence.
This effect alsoprovides a natural explanation for the value of the steeperSalpeter IMF slope in clusters, compared with a simple –2 power law expected from simple Bondi-Hoyle accretion (Oey 2011).Our RIOTS4 field stars therefore consist of both “tipof the iceberg” stars that dominate small, but normal,clusters and “deep field” objects that are substantiallymore isolated. The former correspond to objects thatare consistent with stochastic sampling of the IMF andclustering mass function, as described above; while thelatter correspond to objects that formed in greater isola-tion, if any, and runaway stars.As discussed in § § § ∼ ∼ § . ± . § ∼ CONCLUSIONS
The Runaways and Isolated O-Type Star SpectroscopicSurvey of the SMC (RIOTS4) provides a spatially com-plete, spectroscopic dataset for the field massive stars inthe Small Magellanic Cloud obtained from uniform crite-ria applied to the entire star-forming body of this galaxy.This survey sample is identified using photometric selec-tion criteria combined with a friends-of-friends algorithmto identify the most isolated objects (Oey et al. 2004).Over the course of five years, we obtained spectra for alltargets using the IMACS and MIKE spectrographs onthe Magellan Telescopes. From these spectra, we deriveeach star’s spectral classification and radial velocity.Using RIOTS4, we derived physical parameters suchas the stellar effective temperatures and masses, allowingus to investigate the shape of the field IMF above 20 M ⊙ (Lamb et al. 2013). We find that the slope of the fieldmassive star IMF is significantly steeper (Γ=2.3 ± . ∼ − HST and ground-based imaging, weidentify sparse clusters associated with target OB starsin the RIOTS4 sample. With cluster mass estimates andRIOTS4 stellar masses, we examine the relationship be-tween the most massive star in a cluster and the mass ofthe parent cluster. Our results are consistent with clus-ter mass being independent of the most massive memberstar. This applies unless the total cluster masses are sosmall that stars near the upper-mass limit cannot form.This suppression of the most massive stars in the small-est clusters explains the steep field IMF observed above.We also identify a compelling sample of candidate fieldOB stars that may have formed in situ, given their appar-ent lack of runaway velocities and central location withindense
H ii regions (Oey et al. 2013).We use the radial velocities of RIOTS4 stars to exam-ine the large-scale velocity structure of the SMC, and foran initial look at the kinematics of the field OB popu-lation and runaway frequency. We find that the kine-matics mirror those of other surveys of massive stars(Evans & Howarth 2008) and gas (Stanimirovi´c et al.2004). We find the systemic velocity of the SMC is ∼ − , with a large velocity gradient as a function ofposition that roughly follows the gradient observed in H i gas (Stanimirovi´c et al. 2004). Given this large ve-locity gradient, we must consider the line-of-sight SMCsystemic velocity as given by the gas kinematics whenidentifying runaway stars within our survey. Thus, wecompare the stellar radial velocity for each RIOTS4 starwith the local
H i gas velocity in the line of sight fromStanimirovi´c et al. (1999). Runaway candidates are de-fined to be those objects with a difference >
30 km s − between stellar and H i radial velocities. We find that11% of the sample meets this criterion, which is a lowerbound due to our inability to detect transverse runaways.The identification of a binary runaway system and a can-didate high-velocity (200 km s − ) runaway star suggestthat dynamical ejection is a significant and possibly dom-inant contributor to the runaway OB population.To identify binary stars within our sample, we look forstellar radial velocity variations using 9 −
16 epochs ofdata for three IMACS multi-slit fields encompassing 29stars. We use three methods to identify binary stars.First, binaries are likely to be those objects that exhibitlarge radial velocity variations whose amplitudes corre-late with time interval. Second, we identify binary candi-dates using a statistical F-test, comparing the observedvelocity variation with that expected from observationaluncertainties (Duquennoy & Mayor 1991). Third, weidentify candidates using the periodicity power spectrumand then fitting for orbital solutions. All three methodsfind 10 out of 17 normal OB stars (59% ± ∼ − ∼ − . ± .
04 in the SMC is significantly greater than theMilky Way value, and the SMC spectral type distribu-tion also extends to the hottest effective temperatures,in contrast to Milky Way objects (Golden-Marx et al.2015). These results support the decretion disk modelfor the Be phenomenon, since metal-poor stars rotatefaster due to their inability to remove angular momen-tum via stellar winds. Similarly, our frequency of Be/Bstars is higher than Galactic values, but this result maybe biased by our photometric selection criteria. We willexamine the RIOTS4 Be stars in a future work.Work is also underway to evaluate the fraction of deepfield objects relative to “tip of the iceberg” stars, whichwill further clarify the statistics of OB star formationin the sparsest regime. In addition, we have initiatedfollow-up spectroscopic monitoring to obtain binary starproperties, including systemic velocities. These observa-tions will yield reliable statistics for runaway stars, dataon v sin i , and Oe/Be star variability.Many individuals helped make this publication a real-ity, including the referee, who provided thoughtful com-ments. Thanks to Nidia Morrell and Phil Massey foradvice on radial velocity measurements, and to ThomasBensby, Tom Brink, and Jess Werk for advice on thedata reduction pipelines. Thanks to Mario Mateo forhelp with scheduling the binary monitoring runs and ob-serving advice. We thank Fred Adams, Rupali Chandar,Xinyi Chen, Oleg Gnedin, Lee Hartmann, Wen-hsin Hsu,Anne Jaskot, Mario Mateo, Eric Pellegrini, and Jordan0 Lamb, J. B., et al.Zastrow for helpful discussions. This work was supportedby the National Science Foundation grants AST-0907758,AST-1514838; NASA grant NAG4-9248; and the Univer-sity of Michigan, Rackham Graduate School. REFERENCESAzzopardi, M., Vigneau, J., & Macquet, M. 1975, A&AS, 22, 285Battinelli, P. 1991, A&A, 244, 69Bernstein, R., Shectman, S. A., Gunnels, S. M., Mochnacki, S., &Athey, A. E. 2003, in Society of Photo-Optical InstrumentationEngineers (SPIE) Conference Series, Vol. 4841, InstrumentDesign and Performance for Optical/Infrared Ground-basedTelescopes, ed. M. Iye & A. F. M. Moorwood, 1694–1704Bigelow, B. C., & Dressler, A. M. 2003, in Society ofPhoto-Optical Instrumentation Engineers (SPIE) ConferenceSeries, Vol. 4841, Society of Photo-Optical InstrumentationEngineers (SPIE) Conference Series, ed. M. Iye &A. F. M. Moorwood, 1727–1738Blaauw, A. 1961, Bull. Astron. Inst. Netherlands, 15, 265Bonnell, I. A., Vine, S. G., & Bate, M. R. 2004, MNRAS, 349, 735Bressert, E., et al. 2012, A&A, 542, A49Brott, I., et al. 2011, A&A, 530, A115Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345,245Coe, M. J., et al. 2012, MNRAS, 424, 282de Wit, W. J., Testi, L., Palla, F., & Zinnecker, H. 2005, A&A,437, 247Duquennoy, A., & Mayor, M. 1991, A&A, 248, 485Elmegreen, B. G. 2000, ApJ, 530, 277Evans, C. J., & Howarth, I. D. 2008, MNRAS, 386, 826Evans, C. J., Howarth, I. D., Irwin, M. J., Burnley, A. W., &Harries, T. J. 2004, MNRAS, 353, 601Feast, M. W., Thackeray, A. D., & Wesselink, A. J. 1960,MNRAS, 121, 337Garmany, C. D., Conti, P. S., & Massey, P. 1987, AJ, 93, 1070Gies, D. R. 1987, ApJS, 64, 545Golden-Marx, J. B., Oey, M. S., Lamb, J. B., Graus, A. S., &White, A. S. 2015, ApJ, submittedGraus, A. S., Lamb, J. B., & Oey, M. S. 2012, ApJ, 759, 10Gudehus, D. H. 2001, in Bulletin of the American AstronomicalSociety, Vol. 33, American Astronomical Society MeetingAbstracts
IOTS4 Survey 21
Webster, B. L., Martin, W. L., Feast, M. W., & Andrews, P. J.1972, Nature Physical Science, 240, 183Weidner, C., & Kroupa, P. 2006, MNRAS, 365, 1333Wisniewski, J. P., & Bjorkman, K. S. 2006, ApJ, 652, 458Zaritsky, D., Harris, J., Thompson, I. B., Grebel, E. K., &Massey, P. 2002, AJ, 123, 855 Zickgraf, F.-J., Wolf, B., Stahl, O., & Humphreys, R. M. 1989,A&A, 220, 206Zinnecker, H. 1982, Annals of the New York Academy of Sciences,395, 226 r X i v : . [ a s t r o - ph . GA ] D ec – 1 –Table 1. RIOTS4 Catalog a ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)107 14.96 15.00 -0.95 0.82 Be – – MIKE 111024298 15.18 15.12 -0.91 1.03 B1e – – IMACS f/4 0709201037 15.15 15.28 -0.85 0.44 B0.5 V 110 110 IMACS f/4 0709201600 14.42 14.60 -0.87 0.32 O8.5 V 93 103 IMACS f/4 0709201631 15.19 15.15 -0.99 1.11 B1e
120 120 IMACS f/4 0709201830 13.84 13.94 -0.85 0.49 B0.5 III – – IMACS f/4 0709201952 14.98 14.93 -0.91 1.09 B1e
80 97 IMACS f/4 0709202034 14.51 14.57 -1.01 0.78 B – – MIKE 1110242093 15.12 15.19 -0.95 0.68 B1e – – IMACS f/4 0709202666 15.15 15.18 -0.88 0.70 B1.5e
117 117 IMACS f/4 0810073173 14.66 14.37 -1.73 0.99 O3-4 + neb 111 111 IMACS f/4 0810113224 14.86 14.70 -0.92 1.39 B1e
102 102 IMACS f/4 0709203459 13.32 13.46 -0.93 0.48 O9.5 I 201 169 IMACS f/4 0810113722 15.20 14.83 -1.08 0.46 B – – IMACS f/4 0810073815 14.97 14.92 -1.02 1.13 Be – – IMACS f/4 0810074294 14.96 15.00 -0.93 0.78 Be – – MIKE 1110244424 14.70 14.82 -0.90 0.49 B0 III – – IMACS f/4 0810114919 13.66 13.85 -0.95 0.33 B0 III 118 118 MIKE 1110245041 15.09 15.09 -0.96 0.96 B0 V 130 130 MIKE 1110245063 14.66 14.60 -0.87 1.02 B1e
116 116 IMACS f/4 0810075313 14.89 15.11 -0.87 0.23 O8.5 V 135 135 IMACS f/4 0810075905 14.87 14.90 -0.91 0.75 B0e – – IMACS f/4 0810076908 14.77 14.53 -1.02 1.71 O9.5-B0 III 128 128 IMACS f/4 081006 e
131 131 IMACS f/4 0810117254 14.75 14.74 -0.86 0.86 O9.5 IIIe
126 126 IMACS f/4 081006 e – – MIKE 1110247437 12.93 13.12 -0.94 0.33 O6.5 I(f) 151 151 IMACS f/4 081006 e e
165 165 MIKE 1110248609 13.98 14.11 -0.87 0.48 B0 III 128 128 IMACS f/4 081006 e – – IMACS f/4 08101110421 14.81 14.99 -0.87 0.28 B1 V 152 161 IMACS f/4 07091910556 14.60 14.63 -0.89 0.74 B0.2 V 100 120 IMACS f/4 09082510671 14.88 14.89 -0.84 0.85 B0.5 V 122 122 IMACS f/4 081006 e – – IMACS f/4 090825 ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)11213 14.88 14.77 -0.97 1.29 Be – – IMACS f/4 09082411238 14.41 14.30 -0.96 1.21 O + B – – MIKE 11102411280 14.15 14.16 -0.94 0.90 B1e
125 138 IMACS f/4 07091911623 14.12 14.13 -0.86 0.83 O9 III: 170 170 IMACS f/4 06091211677 14.47 14.46 -1.01 1.04 O9 III:e – – IMACS f/4 08101111777 13.76 13.88 -0.92 0.53 B0.2 V 168 168 IMACS f/4 06091211802 14.62 14.59 -0.86 0.96 B0.2 IV 155 155 IMACS f/4 09082412102 14.64 14.69 -0.98 0.84 O9 IIIe – – IMACS f/4 08101113075 12.93 13.05 -0.89 0.52 O9.5 I 131 131 IMACS f/4 06091213314 14.76 14.77 -0.87 0.84 B1 III 168 215 IMACS f/4 07091913682 14.94 14.92 -0.96 1.03 Be – – IMACS f/4 09082513774 15.07 15.07 -0.84 0.85 – – – IMACS f/4 06091313831 13.91 14.06 -0.87 0.39 B0.5 V 172 172 IMACS f/4 06091213896 13.56 13.76 -0.98 0.30 O8 III((f)) 159 163 IMACS f/4 08101114324 14.17 14.11 -0.97 1.12 O6 V((f))e
152 152 IMACS f/4 09082414878 14.11 14.25 -0.89 0.46 O9 III 147 147 IMACS f/4 09082515060 14.11 14.29 -0.88 0.30 B0.2 III-IV 155 155 IMACS f/4 06091215102 15.17 15.01 -0.89 1.40 B1.5 V 103 123 IMACS f/4 08101115256 15.09 15.00 -0.90 1.14 B1 V 157 157 IMACS f/4 06091215263 15.05 15.22 -0.87 0.28 B0.2 V 109 156 IMACS f/4 07091915271 13.36 13.54 -0.94 0.39 O6 III((f))e
143 153 IMACS f/4 07091915618 14.84 14.95 -0.92 0.53 B 128 128 IMACS f/4 09082515742 13.51 13.64 -0.92 0.53 O9 III 168 168 IMACS f/4 09082515854 15.21 14.51 -1.17 0.68 B1e
107 117 IMACS f/4 09082416039 14.53 14.54 -1.03 0.99 B1e – – IMACS f/4 09082516147 14.03 14.15 -0.91 0.53 B0 V 147 147 IMACS f/4 06091216230 13.01 13.20 -0.90 0.29 O9 III: 71 136 IMACS f/4 06091216481 14.16 14.26 -0.94 0.62 O9.5 V 160 160 IMACS f/4 07091916518 13.77 13.90 -0.96 0.53 B0 V 176 171 IMACS f/4 07091916587 13.99 13.91 -1.00 1.24 B0 V – – IMACS f/4 07092016616 14.18 14.29 -0.86 0.50 – – – IMACS f/4 06091317240 14.37 14.47 -0.85 0.55 O7.5 V 105 108 IMACS f/4 06091217813 14.83 14.84 -0.86 0.86 B0 V 192 184 IMACS f/4 07092018187 14.95 15.00 -0.95 0.79 Be – – IMACS f/4 07091918301 14.40 14.58 -0.86 0.31 B0.2 V 155 155 IMACS f/4 07091918329 14.64 14.65 -0.98 0.91 O9.5 IIIe pec 119 123 IMACS f/4 07092018373 13.48 13.41 -1.05 1.26 Be
160 160 MIKE 111025 ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)19382 14.27 14.31 -0.96 0.81 B0 V 153 153 IMACS f/4 07092019728 12.49 12.59 -0.89 0.59 B1 I 166 174 IMACS f/4 07091920939 14.14 14.18 -0.86 0.73 B1 II 155 155 IMACS f/4 06091221844 14.09 14.18 -0.86 0.57 O8 III((f)) 151 151 IMACS f/4 080920 e
130 130 IMACS f/4 081006 e
167 167 IMACS f/4 080920 e f
98 118 IMACS f/4 06091222451 13.98 14.14 -0.93 0.41 O9 III – – IMACS f/4 07092022623 15.19 15.21 -0.86 0.84 B0e – – IMACS f/4 07092023710 14.85 14.74 -0.97 1.31 O9-B0 pe
168 168 IMACS f/4 080920 e – – IMACS f/4 06091223954 14.59 14.58 -0.85 0.86 B1.5e
130 130 IMACS f/4 080920 e e
155 155 IMACS f/4 080920 e – – MIKE 11102424914 14.19 14.19 -1.00 0.95 O9 III-Vpe
81 105 IMACS f/4 081121 e
130 130 IMACS f/4 080920 e
124 124 IMACS f/4 080920 e – – IMACS f/4 06091225974 15.08 15.08 -0.85 0.75 B1.5 V 116 123 IMACS f/4 06091227135 15.20 15.32 -0.93 0.56 B1e
113 117 IMACS f/4 080920 e e e
153 153 IMACS f/4 080920 e
156 156 IMACS f/4 080920 e e ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)30420 15.12 15.24 -0.97 0.56 WN3 + abs f – – IMACS f/4 06091230472 14.55 14.67 -0.87 0.48 B1e – – IMACS f/4 06091230492 14.45 14.51 -0.92 0.72 B0.5 179 179 MIKE 11102430744 14.24 14.30 -0.88 0.66 B1 III + O9.5 Ve – – IMACS f/4 06091231574 15.17 15.10 -0.90 1.05 B1 V 149 145 IMACS f/4 06091231699 14.73 14.78 -0.90 0.73 B1e – – IMACS f/4 06091232449 14.19 14.29 -0.86 0.51 B1.5 III 121 121 IMACS f/4 06091232552 14.85 14.87 -0.96 0.88 B0e – – IMACS f/4 09082532752 14.63 14.57 -0.90 1.04 Be
270 186 MIKE 11102533135 15.20 15.26 -0.95 0.77 B0e + – – IMACS f/4 09082533245 15.10 15.23 -0.91 0.47 Be – – IMACS f/4 06091233823 14.45 14.62 -0.89 0.35 O9.5 III 170 170 MIKE 11102534005 14.80 15.01 -1.02 0.29 O9.5 V 166 166 IMACS f/4 09082534315 13.68 13.78 -0.88 0.54 B1 II 138 138 IMACS f/4 06091234457 13.74 13.93 -0.91 0.30 B0 III 168 168 MIKE 11102534651 15.01 14.99 -0.87 0.92 Be – – MIKE 11102534988 14.84 14.85 -0.95 0.89 Be
226 183 MIKE 11102535474 14.56 14.45 -0.93 1.53 B1 III 224 184 MIKE 11102535491 14.66 14.71 -0.90 0.77 O8 V 126 126 IMACS f/4 06091135598 14.88 15.15 -0.94 0.05 O8 V 162 162 IMACS f/4 06091136175 12.64 12.68 -0.85 0.73 B1 II 191 191 MIKE 11102436213 13.21 13.38 -0.90 0.38 O9.5 III + B 120 120 MIKE 11102536294 14.51 14.55 -0.95 0.80 B0.5e
130 130 IMACS f/4 06091136325 15.02 14.88 -0.88 1.33 O9.5 V 146 146 IMACS f/4 06091236514 15.01 15.15 -0.87 0.47 O9 V 141 141 IMACS f/4 06091136815 14.93 14.98 -0.96 0.78 Be
155 155 MIKE 11102536975 15.04 15.04 -0.89 0.87 Be
164 164 MIKE 11102637419 15.12 14.95 -0.98 1.54 Be
127 127 IMACS f/4 06091137502 14.65 14.62 -1.03 1.10 O9.5-B0: pe – – IMACS f/4 06091138024 14.32 14.53 -0.94 0.27 O4 V((f)) 103 151 IMACS f/4 09082538036 14.28 14.33 -0.98 0.82 O6.5-7: Vpe
152 152 IMACS f/4 06091138445 14.67 14.68 -1.09 1.02 Be – – IMACS f/4 09082538508 15.14 14.95 -0.96 1.52 B0.7 179 179 IMACS f/4 09082538893 14.74 14.89 -0.90 0.41 B0e
179 179 IMACS f/4 06091138921 14.07 14.12 -0.96 0.80 B0 III 154 154 IMACS f/4 06091139211 14.73 14.94 -0.86 0.18 O9.5 III 156 156 IMACS f/4 09082539904 15.13 15.17 -0.98 0.81 B0e – – IMACS f/4 090825 ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)40380 14.22 14.33 -0.92 0.58 O7 V 181 178 IMACS f/4 09082540504 13.98 14.15 -0.87 0.30 B1.5 III 95 150 IMACS f/4 09082541183 14.78 14.84 -0.95 0.77 Be – – IMACS f/4 10122241345 14.87 15.04 -0.89 0.34 B1 III – – IMACS f/4 06091141648 13.86 13.95 -0.90 0.61 B0.5 III 106 118 IMACS f/4 06091142260 14.43 14.54 -0.85 0.50 O9.5 III 140 140 IMACS f/4 06091142654 14.84 14.78 -0.99 1.17 Be – – IMACS f/4 09082542959 14.59 14.72 -0.91 0.48 B – – – –43411 14.90 14.97 -0.98 0.72 O9 V 132 150 IMACS f/4 08100643589 14.93 15.04 -0.89 0.56 B0e – – IMACS f/4 06091143724 14.05 14.25 -0.96 0.32 O7 III((f)) 100 148 IMACS f/4 10122244316 15.04 15.15 -0.93 0.58 Be
192 169 MIKE 11102644336 14.78 14.72 -1.01 1.20 B1.5e – – IMACS f/4 09082544965 14.65 14.69 -1.06 0.90 B0e – – IMACS f/4 07092045640 13.95 14.01 -0.97 0.74 B1e
142 150 IMACS f/4 08100646022 14.74 14.85 -0.84 0.52 O9.5 V 148 153 IMACS f/4 10122146035 14.05 14.24 -0.95 0.34 O7 V – – IMACS f/4 10122146090 14.58 14.66 -0.88 0.64 B0e – – IMACS f/4 07092046241 15.17 15.11 -0.86 1.00 B1 V 117 117 IMACS f/4 06091246317 14.53 14.76 -0.90 0.17 O8.5 V 174 174 IMACS f/4 07092046392 14.39 14.56 -0.93 0.35 B0.7 IV 138 156 IMACS f/4 10122246398 12.67 12.64 -0.94 1.00 B[e] – – MIKE 11102446831 15.05 15.15 -0.87 0.54 O7.5-9 V 145 148 IMACS f/4 08100647029 15.08 15.18 -0.84 0.53 B0.2 V 188 188 IMACS f/4 10122147459 15.10 15.17 -0.85 0.55 B0 V + B 155 155 IMACS f/4 06091247478 14.74 14.96 -0.98 0.28 O7 V 143 154 IMACS f/4 10122247668 15.04 15.16 -0.97 0.56 Be – – – –47908 14.64 14.19 -1.10 0.52 Be – – IMACS f/4 09082448037 14.56 14.54 -1.00 1.03 B0e – – IMACS f/4 09082448057 14.91 14.93 -0.97 0.87 B0.2e – – IMACS f/4 09082448170 14.16 14.29 -0.96 0.53 O9.5 V 145 145 IMACS f/4 08101148266 15.19 15.17 -1.08 1.06 B0e – – IMACS f/4 08100648432 14.94 15.07 -0.90 0.37 B0III + B 201 186 MIKE 11102548601 13.58 13.71 -0.92 0.49 B0 III 133 133 IMACS f/4 06091248882 15.19 15.16 -0.99 1.04 B1e
148 158 IMACS f/4 09082449014 14.53 14.60 -0.95 0.69 B0e
133 133 IMACS f/4 09082449450 14.07 14.22 -0.86 0.37 O9.5 V 150 156 IMACS f/4 101222 ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)49517 15.02 14.97 -0.96 1.11 B1.5 V 138 138 IMACS f/4 06091249580 14.22 14.38 -0.88 0.39 O9.5 IV 124 124 IMACS f/4 09082449825 14.75 14.94 -0.88 0.29 B0.2 III 140 156 IMACS f/4 10122249937 15.14 15.19 -0.90 0.74 B0V 395 192 MIKE 11102550095 14.48 14.51 -0.94 0.87 O9 Ve – – IMACS f/4 06091250153 14.64 14.46 -1.47 1.92 H II region – – MIKE 11102550331 12.83 12.99 -0.92 0.41 – – – IMACS f/4 06091250396 15.17 15.16 -1.00 1.00 Be – – IMACS f/4 09082450609 12.43 12.55 -0.90 0.53 B1 I 149 141 IMACS f/4 09082450791 15.04 15.19 -0.93 0.47 O8 V 190 183 IMACS f/4 08100750825 14.58 14.84 -0.92 0.08 O8 V 230 186 IMACS f/4 10122251036 15.10 15.16 -0.87 0.71 Be – – IMACS f/4 10122251214 14.72 14.76 -1.03 0.87 Be – – IMACS f/4 08100651234 13.58 13.66 -0.86 0.58 B1.5 III 187 187 IMACS f/4 07092051236 14.93 14.92 -0.89 0.89 Be – – IMACS f/4 09082451373 13.72 13.79 -1.08 0.84 O8 IIIze – – IMACS f/4 09082451384 13.26 13.42 -0.92 0.44 O9.5 Ia 82 123 IMACS f/4 07092051419 15.15 15.29 -0.90 0.45 O9 III – – IMACS f/4 06091251424 14.99 15.02 -0.94 0.82 B1e
151 153 IMACS f/4 08100751435 14.92 15.12 -0.89 0.29 O9 III-V 195 189 IMACS f/4 09082451500 14.72 14.97 -0.93 0.15 O6 V 171 171 IMACS f/4 10122152257 14.57 14.61 -0.87 0.77 B0e – – IMACS f/4 09082452363 14.80 14.98 -0.97 0.38 O7 III((f))e
170 170 IMACS f/4 10122252410 13.69 13.75 -1.07 0.85 O8 III: ze pec – – IMACS f/4 09082452865 14.48 14.56 -1.12 0.85 B0-B0.5e
192 181 MIKE 11102552959 15.03 15.25 -0.87 0.14 B0 V – – MIKE 11102653042 14.67 14.85 -0.86 0.28 O9.5 V 154 163 IMACS f/4 09082453319 14.66 14.90 -0.87 0.15 B0 V 173 173 IMACS f/4 08100753360 15.04 15.09 -0.95 0.77 O7-9p:e – – IMACS f/4 10122253480 13.99 14.24 -1.15 0.03 – – – – –54456 13.35 13.55 -0.88 0.25 B0 III 138 161 IMACS f/4 10122254721 12.69 12.88 -0.95 0.35 O9 III 118 151 IMACS f/4 08100755417 14.58 14.62 -0.99 0.83 B1e – – IMACS f/4 07092055952 14.43 14.64 -0.89 0.21 B0.2 V 251 196 IMACS f/4 08100756503 14.87 14.95 -0.92 0.68 O9 Ve – – IMACS f/4 07092056587 14.68 14.75 -0.97 0.73 B0 V – – IMACS f/2 11072156662 14.87 11.85 -1.30 0.54 B 182 182 MIKE 111026 ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)57397 14.66 14.69 -0.96 0.82 B1e
158 163 IMACS f/4 07092058168 14.94 14.87 -0.89 1.10 – – – – –58756 14.79 14.74 -0.94 1.08 Be – – IMACS f/4 08101158864 15.14 15.14 -0.91 0.90 B0e – – IMACS f/4 07092058947 13.63 13.63 -0.85 0.74 B1 III 156 164 IMACS f/4 08101159319 14.23 14.32 -0.99 0.68 early Ope – – IMACS f/4 08100759421 13.64 13.78 -0.86 0.42 B1 III 155 146 IMACS f/4 08101159867 14.83 14.76 -1.06 1.25 B0e – – IMACS f/4 07092059897 15.19 15.28 -0.93 0.66 B0e
154 148 MIKE 11102659977 14.81 14.81 -0.98 0.94 B0e – – IMACS f/2 11072160439 13.73 13.94 -0.90 0.22 O + B – – MIKE 11102660460 14.74 14.98 -0.88 0.09 O8.5 V 173 173 IMACS f/4 08101161039 14.87 15.11 -0.85 0.09 O9.5 V 184 184 IMACS f/4 08100761543 15.11 15.19 -0.85 0.61 B – – MIKE 11102561842 13.38 13.47 -0.87 0.59 B1 III – – IMACS f/2 11071962416 14.59 14.68 -0.92 0.61 O9 V 135 135 IMACS f/4 07092062638 14.57 14.67 -0.98 0.65 O9.5 III-Ve – – IMACS f/4 08100762661 13.01 12.98 -0.85 0.91 B[e] 89 159 MIKE 11102662981 14.66 14.86 -0.87 0.24 B0.2 V 160 166 IMACS f/4 07092063112 14.97 14.99 -0.90 0.82 B0.5e
113 159 IMACS f/4 08100763284 13.38 13.49 -0.85 0.50 B1e
124 161 IMACS f/4 07092063413 13.97 13.90 -0.89 1.10 O9.5 Ia 144 163 IMACS f/4 08100763842 14.56 14.70 -0.87 0.41 B1 V 141 141 IMACS f/4 08100663877 14.78 14.98 -0.87 0.23 B0.2 V 129 166 IMACS f/4 07092064032 14.98 14.99 -0.91 0.84 B1e – – IMACS f/4 09082664194 14.61 14.60 -0.98 0.99 B0e
168 168 IMACS f/4 09082664453 13.41 13.38 -1.06 1.14 B1e
194 194 IMACS f/4 09082664710 14.51 14.56 -0.85 0.68 B1.5e
117 161 IMACS f/4 08100764773 15.05 15.25 -0.87 0.23 O8.5 V 136 136 IMACS f/4 09082665103 15.03 15.09 -0.97 0.74 Be
125 125 IMACS f/4 08100665145 15.11 15.00 -0.85 1.19 B1.5e
175 175 IMACS f/4 08100665318 14.78 14.95 -0.89 0.35 O9 Ve
190 190 IMACS f/4 08100765346 14.99 15.19 -0.99 0.36 O8 V 156 163 MIKE 11102665355 14.71 14.91 -0.92 0.28 O8 V + O9 – – IMACS f/4 09082666160 12.92 13.09 -0.92 0.40 O + B 125 166 IMACS f/4 08100666302 14.78 14.80 -0.98 0.87 B1e
174 174 IMACS f/4 08100666415 13.08 13.25 -0.91 0.38 O9.7 Ia 166 166 IMACS f/4 081006 ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)66507 14.83 14.91 -0.96 0.68 B1e – – IMACS f/4 08100667029 15.08 15.21 -0.86 0.40 B1 V 148 141 IMACS f/4 08100667060 14.43 14.63 -0.94 0.28 O7 Vz 126 164 IMACS f/4 09082867269 14.56 14.73 -0.93 0.40 O7 V 144 140 IMACS f/4 09082667305 15.16 15.20 -0.93 0.77 B1.5e
180 180 IMACS f/4 08100667334 14.93 15.05 -0.94 0.53 B0 V 209 202 IMACS f/4 08100667673 14.80 14.84 -1.00 0.83 O9.5 V:e – – IMACS f/4 09082667893 13.92 13.97 -1.05 0.87 B0e – – IMACS f/4 08100668071 14.99 15.19 -0.91 0.28 O9.5 III 177 177 IMACS f/4 08100668157 15.14 15.37 -0.85 0.10 B0.2 V 204 201 IMACS f/4 08100668427 13.52 13.51 -1.06 1.09 B0.2e
157 157 IMACS f/4 09082668621 14.41 14.64 -0.94 0.20 O9.5 III 156 174 IMACS f/4 09082868756 14.30 14.48 -0.94 0.34 O7.5 III: 205 205 IMACS f/4 09082868963 15.20 15.39 -0.87 0.25 B0 V 163 163 IMACS f/4 08100669155 14.50 14.69 -0.85 0.28 B0.2 III 209 206 IMACS f/4 09082869460 14.82 14.95 -0.99 0.53 O6.5 III((f))e
130 146 IMACS f/4 09082669555 14.35 14.59 -0.96 0.17 O6.5 V 223 197 IMACS f/4 09082669598 15.03 15.23 -0.95 0.22 O9 V 175 175 IMACS f/4 09082869630 13.79 14.01 -0.85 0.16 B1 II – – IMACS f/4 08100669769 15.13 15.30 -0.93 0.37 B1.5 V 122 131 IMACS f/4 08100670149 14.41 14.61 -0.90 0.25 O9 V 155 155 IMACS f/4 09082670663 14.60 14.63 -0.89 0.80 B1.5e – – IMACS f/2 11072071002 13.75 13.97 -0.93 0.23 O9.5 III 154 163 IMACS f/4 09082671409 14.91 14.88 -1.05 1.14 Be – – IMACS f/4 09082671652 14.44 14.65 -0.88 0.18 B0.5e – – IMACS f/4 09082571815 15.20 15.43 -0.96 0.21 O8 V 172 172 IMACS f/4 09082672204 14.34 14.57 -0.92 0.20 O8 V + O8 V – – IMACS f/4 09082672208 12.41 12.54 -0.89 0.48 B1 II 131 133 IMACS f/4 09082872210 14.47 14.48 -1.02 0.99 B0e – – IMACS f/4 09082572535 13.42 13.45 -1.11 1.00 O8-9: IIIpe:
162 162 IMACS f/4 09082572656 14.85 14.99 -0.85 0.41 B0-1.5 III 135 168 IMACS f/4 09082672724 14.25 14.42 -0.87 0.34 B0 V 201 201 IMACS f/4 09082872884 14.65 14.90 -0.92 0.11 O9 V 222 204 IMACS f/4 09082672941 14.86 15.11 -0.90 0.10 O9 V 110 126 IMACS f/4 09082873185 15.19 15.27 -0.95 0.72 B – – IMACS f/2 11101273337 12.96 13.12 -0.96 0.47 O8.5 I((f)) 202 202 IMACS f/4 09082873355 14.26 14.36 -0.93 0.59 B0e – – IMACS f/4 090825 ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)73795 14.68 14.73 -1.03 0.82 mid Oe – – IMACS f/2 11072073913 13.57 13.63 -0.91 0.72 O9.5 I 184 184 IMACS f/4 09082573952 12.51 12.62 -0.88 0.52 B1.5 II 135 135 IMACS f/4 10122174828 12.97 13.08 -0.86 0.52 Be – – IMACS f/4 09082574946 15.13 15.14 -0.99 0.88 B 73 131 IMACS f/4 10122175061 14.17 14.24 -0.93 0.69 B1e
164 164 IMACS f/4 09082575126 14.65 14.39 -1.14 2.02 O9 V 139 139 IMACS f/4 09082575210 14.57 14.67 -0.89 0.59 O8.5 V 196 196 IMACS f/4 09082575626 14.87 15.14 -0.93 0.04 O9 III-V 159 159 IMACS f/4 09082575689 14.31 14.34 -1.08 0.94 Ope pec – – MIKE 11102575719 14.60 14.87 -0.84 0.00 – – – – –75919 14.32 14.39 -1.01 0.78 O9 IIIe
166 166 IMACS f/4 10122175980 14.12 14.17 -1.02 0.81 B0e – – IMACS f/4 09082675984 15.12 15.35 -0.85 0.07 B0 III 229 212 MIKE 11102676253 15.04 15.27 -0.90 0.13 B0.2 V + B 188 188 IMACS f/4 09082576371 13.76 13.96 -0.91 0.30 O9.5 III – – IMACS f/4 09082676654 14.74 14.75 -1.17 1.04 Be – – IMACS f/4 09082676657 15.14 15.31 -0.94 0.43 O9-9.5 V 198 198 IMACS f/4 09082676773 13.46 13.59 -0.86 0.45 Be – – IMACS f/4 09082576870 14.92 14.98 -0.99 0.77 Be – – IMACS f/2 11101377290 15.10 15.10 -1.04 1.00 B0.5e – – IMACS f/4 09082577368 13.76 14.00 -0.93 0.15 O6 V 222 202 IMACS f/4 09082677458 13.00 13.15 -0.92 0.45 B0.2e
204 204 IMACS f/4 09082677609 12.55 12.69 -0.91 0.48 B0.5 I 192 192 IMACS f/4 09082577616 14.11 14.08 -1.11 1.22 early Ope pec – – IMACS f/4 09082677734 14.87 14.93 -0.97 0.74 – – – IMACS f/4 09082677816 14.43 14.59 -0.85 0.34 B0.2 III 168 168 IMACS f/4 09082577851 14.09 14.14 -0.96 0.82 B0.2-1e – – IMACS f/4 09082678438 14.79 15.02 -0.86 0.11 B0 V – – IMACS f/2 11072078694 14.27 14.35 -0.98 0.72 O8.5 IIIe – – IMACS f/2 11072179225 14.65 14.69 -1.06 0.88 Be
150 171 MIKE 11102679248 14.40 14.60 -0.92 0.29 O8.5 V – – IMACS f/2 11072179326 14.90 14.94 -0.95 0.80 Be – – IMACS f/2 11071979513 14.71 14.88 -0.97 0.42 Be – – IMACS f/2 11071979587 14.98 15.13 -0.87 0.38 B – – IMACS f/2 11101279697 14.86 14.93 -1.04 0.79 Be
219 201 MIKE 11102679976 14.74 14.74 -1.01 1.01 Be + – – IMACS f/2 110720
10 –Table 1—Continued ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)80075 14.90 15.07 -0.87 0.35 B – – IMACS f/2 11101280352 14.23 14.28 -1.08 0.90 Be – – IMACS f/2 11101380412 13.33 13.46 -0.94 0.53 B0.7 II 184 150 IMACS f/4 09082880545 13.36 13.53 -0.89 0.36 B0.5 II – – IMACS f/4 09082880573 14.38 14.48 -0.85 0.54 B1 V 179 179 IMACS f/4 09082780579 14.66 14.81 -0.84 0.38 B0.7 V 168 163 IMACS f/4 09082780582 14.40 14.48 -0.92 0.66 B – – IMACS f/4 09082880998 14.24 14.46 -0.96 0.25 O8 V – – IMACS f/2 11072181019 14.24 14.43 -0.93 0.33 O9.5 V 162 162 IMACS f/4 09082781071 14.67 14.70 -0.99 0.88 Be – – IMACS f/2 11072181169 15.02 15.23 -0.86 0.19 B0.2 V 178 178 IMACS f/4 09082781258 14.96 15.18 -0.88 0.14 B0-1.5 V 247 174 IMACS f/4 09082781465 14.85 14.88 -1.02 0.89 Be – – IMACS f/4 09082681634 14.42 14.50 -0.86 0.57 B1.5 Ve
177 181 IMACS f/4 09082681646 14.15 14.36 -0.92 0.26 O8 V 185 185 IMACS f/4 09082681647 14.62 14.85 -0.89 0.17 B0.2 V 168 168 IMACS f/4 09082781696 15.07 15.17 -0.91 0.55 B1 V 194 187 IMACS f/4 09082681941 13.82 14.03 -0.89 0.24 O9.5 III 200 187 IMACS f/4 09082782322 14.12 14.37 -0.88 0.09 O9.5 III 171 171 IMACS f/4 09082782328 14.87 14.96 -1.01 0.71 B0e – – IMACS f/4 09082682387 15.07 15.26 -0.87 0.27 Be – – IMACS f/2 11101182408 14.39 14.59 -0.87 0.23 B1 III 196 176 IMACS f/4 09082682444 15.15 15.30 -0.95 0.44 B0 V 199 169 IMACS f/4 09082682489 14.17 14.22 -1.04 0.86 O9: IIIpe – – IMACS f/4 09082782572 14.45 14.40 -0.85 0.96 Be – – IMACS f/2 11101182711 14.32 14.36 -1.00 0.83 B1 V – – IMACS f/4 09082682783 14.88 15.12 -0.89 0.11 B0.5 V 160 160 IMACS f/4 09082783017 14.80 15.03 -0.87 0.15 O9.5 III 191 181 IMACS f/4 09082683073 15.10 15.27 -0.87 0.34 B0.7 V 158 215 IMACS f/4 09082683074 15.12 15.35 -0.85 0.10 Be – – IMACS f/2 11101383171 14.28 14.39 -0.94 0.58 B0e – – IMACS f/4 09082683202 12.61 11.53 -0.87 0.47 Be – – IMACS f/2 11071983224 14.60 14.65 -1.12 0.90 B1e – – IMACS f/4 09082683232 14.11 14.28 -0.84 0.29 B1.5 III – – IMACS f/4 09082683480 13.74 13.73 -0.90 0.88 B[e] 164 164 IMACS f/4 09082683510 15.16 15.45 -0.91 0.00 O8 V 153 159 IMACS f/4 09082683651 14.88 15.08 -0.86 0.24 B – – IMACS f/2 111012
11 –Table 1—Continued ID b B b V b Q UBR A V c Sp Type RV star RV HId
Instrument Observation Date(km s − ) (km s − ) (YYMMDD)83678 12.99 13.18 -0.95 0.36 O8.5 III 166 174 IMACS f/4 09082683962 14.52 14.61 -0.96 0.65 Be – – IMACS f/2 11072084277 14.42 14.57 -0.85 0.38 B1e – – IMACS f/2 11072084544 14.88 15.02 -0.91 0.46 Be – – IMACS f/2 111012 a This table is published in its entirety in the electronic edition of the
Astrophysical Journal . A portion isshown here for guidance regarding its form and content. b From Massey (2002). c From Zaritsky et al. (2002). d Measured from Stanimirovi´c et al. (1999). e Observed multiple times for binary monitoring; see Table 3. ff