The CHARA Array resolves the long-period Wolf-Rayet binaries WR 137 and WR 138
Noel D. Richardson, Tomer Shenar, Olivier Roy-Loubier, Gail Schaefer, Anthony F. J. Moffat, Nicole St-Louis, Douglas R. Gies, Chris Farrington, Grant M. Hill, Peredur M. Williams, Kathryn Gordon, Herbert Pablo, Tahina Ramiaramanantsoa
aa r X i v : . [ a s t r o - ph . S R ] J un MNRAS , 1–11 (2016) Preprint 29 August 2018 Compiled using MNRAS L A TEX style file v3.0
The CHARA Array resolves the long-period Wolf-Rayetbinaries WR 137 and WR 138
Noel D. Richardson ⋆ , Tomer Shenar , Olivier Roy-Loubier ,Gail Schaefer , Anthony F. J. Moffat , Nicole St-Louis , Douglas R. Gies ,Chris Farrington , Grant M. Hill , Peredur M. Williams , Kathryn Gordon ,Herbert Pablo , and Tahina Ramiaramanantsoa Ritter Observatory, Department of Physics and Astronomy, The University of Toledo, Toledo, OH 43606-3390, USA Institut f¨ur Physik und Astronomie, Universit¨at Potsdam, Karl-Liebknecht-Str. 24/25, D-14476 Potsdam, Germany D´epartement de physique and Centre de Recherche en Astrophysique du Qu´ebec (CRAQ), Universit´e de Montr´eal, C.P. 6128,Succ. Centre-Ville, Montr´eal, Qu´ebec, H3C 3J7, Canada The CHARA Array, Mount Wilson Observatory, 91023 Mount Wilson CA, USA Center for High Angular Resolution Astronomy, Department of Physics and Astronomy, Georgia State University, P. O. Box5060, Atlanta, GA 30302-5060, USA W. M. Keck Observatory, 65-1120 Mamalahoa Highway, Kamuela, HI 96743, USA Institute for Astronomy, University of Edinburgh, Royal Observatory, Edinburgh EH9 3HJ, UK
ABSTRACT
We report on interferometric observations with the CHARA Array of two classicalWolf-Rayet stars in suspected binary systems, namely WR 137 and WR 138. In bothcases, we resolve the component stars to be separated by a few milliarcseconds. Thedata were collected in the H -band, and provide a measure of the fractional flux forboth stars in each system. We find that the WR star is the dominant H -band lightsource in both systems ( f WR , = . ± . ; f WR , = . ± . ), which is confirmedthrough both comparisons with estimated fundamental parameters for WR stars andO dwarfs, as well as through spectral modeling of each system. Our spectral modelingalso provides fundamental parameters for the stars and winds in these systems. Theresults on WR 138 provide evidence that it is a binary system which may have gonethrough a previous mass-transfer episode to create the WR star. The separation andposition of the stars in the WR 137 system together with previous results from theIOTA interferometer provides evidence that the binary is seen nearly edge-on. Thepossible edge-on orbit of WR 137 aligns well with the dust production site imagedby the Hubble Space Telescope during a previous periastron passage, showing that thedust production may be concentrated in the orbital plane.
Key words: stars: early-type – binaries: visual – stars: individual (WR 137, WR138) – stars: winds, outflows – stars: mass loss – stars: Wolf-Rayet
Every star is described with a set of fundamental parametersincluding its radius, R , mass, M , luminosity, L , and effectivetemperature, T e ff . For most stars, we must estimate thesefrom accurate photometry, a measure of the distance, and areliable spectrum. This can be calibrated using techniquesinvolving binary stars where we either utilise an eclipsing ⋆ E-mail:[email protected] system to understand fundamental parameters or where avisually resolved orbit can allow us to infer the stellar massesof the binary. Recent progress has been made in the area offundamental parameters through interferometric techniques(e.g., Boyajian et al. 2012) where long-baseline interferome-try can resolve single stars, yielding direct measures of an-gular diameters. When coupled with an SED and distance,measurements of stellar radii and temperatures are possibleto accuracies of a few percent allowing accurate predictionsfor other stars (Boyajian et al. 2014). c (cid:13) N. D. Richardson et al.
Empirical relations for massive stars are more compli-cated to determine due to their scarcity and large distances.Even with a large binary fraction (e.g., Aldoretta et al. 2015,Sana et al. 2014), the calibration of stellar parameters hasrelied on eclipsing binaries. Sana et al. (2012) show that mostmassive stars with periods of P . d will either inter-act in the future or have already interacted. Therefore, moststars that are used to calibrate the mass-radius-luminosityrelationships for massive stars (Martins et al. 2005) are notrepresentative of single-star evolution. If we wish to under-stand the evolution of massive stars without binary effects,we need to measure directly masses for evolved stars in long-period, widely-separated binaries. In order to measure themasses of these stars, as well as place lower limits on themass loss of the more-evolved component, we must visuallyresolve the orbits of nearby systems where we can also obtaina double-lined spectroscopic orbit.The Wolf-Rayet (WR) stars represent a class of ob-jects with great potential to determine the empirical lowerlimits to the mass lost by massive stars, which Smith &Owocki (2006) speculate could be up to 65% of the massof a 60 M ⊙ star prior to the supernova explosion. Thus far,only two WR systems have been examined with interfer-ometry. They are γ Velorum (Hanbury Brown et al. 1970,Millour et al. 2007, North et al. 2007) and WR 140 (Monnieret al. 2004, 2011). Both systems have well-defined double-lined spectroscopic orbits (Schmutz et al. 1997, Fahed etal. 2011) and consist of a carbon-rich WC star orbiting anO star. For γ Vel, the spectral types of the component starsare WC8 + O7.5III (Schmutz et al. 1997), while for WR 140,they are WC7pd + O5.5fc (Fahed et al. 2011).The first results on γ Vel (WR 11, HD 68273) werepresented by Hanbury Brown et al. (1970), who found thatthe Narrabri Intensity Interferometer could be used not onlyto resolve single hot stars as it was designed for, but also toresolve binary orbits and determine the angular sizes of theemission-line forming regions for some hot stars. As such,they derived a separation of the binary ( P = . ± . mas, corresponding to a distance of ± pc when com-pared to the double-lined spectroscopic orbit. Further, Han-bury Brown et al. found that the emission line region was0.24 a in relation to the semi-major axis, a . γ Vel was thenstudied with VLTI/AMBER in the near-infrared by Millouret al. (2007). These results proved that the expected visualorbit was fairly well-constrained at the time of the obser-vation, and that the predicted WR and O star fluxes werecompatible with the previous modeling of the system (deMarco et al. 1999, 2000). Further, they found that roughly5% of the flux in the K − band was from the free-free emissionin the wind-wind collision region. Finally, North et al. (2007)observed γ Vel across its entire orbit with the Sydney Uni-versity Stellar Interferometer (SUSI) to obtain a visual orbitand measure the distance to an unprecedented precision of + − pc. The resulting masses were M WR = . ± . M ⊙ and M O = . ± . M ⊙ , and represent the best-measured massfor any WR star.WR 140 (HD 193793) is a prototype for colliding windsystems with a highly elliptical ( e = . ), long period( P = . years) orbit. The system was resolved with theIOTA3 Imaging Interferometer by Monnier et al. (2004),and then Monnier et al. (2011) combined the IOTA3 mea-surements and new CHARA measurements with the radial velocity orbit based on an intensive spectroscopic campaignby Fahed et al. (2011) to measure the visual orbit preciselyand then derive the masses of the component stars to be M WR = . ± . M ⊙ and M O = . ± . M ⊙ , with the systemat a distance of . ± . kpc.Two long-period WR stars are obvious candidates forfollow-up long-baseline interferometry: WR 137 and WR138. WR 137 (HD 192764) consists of a WC7pd + O9 bi-nary with a period of 13.05 years and a small eccentricityof e = . (Lef`evre et al. 2005). It has a known RV orbit(at least for the WR component; the O star’s orbit is verynoisy) with a sin i = AU, which combined with a distanceof 1.82 kpc (Nugis & Lamers 2000), yields a value of ∼ H − band, with a flux ratio measuredrelative between the two stars of f / f =0.81 and a separa-tion of 9.8 mas in 2005, although the flux ratio is ambiguousas to which component star is represented by which. Thismakes this system appealing to observe with follow-up ob-servations so that a third WR+O system can be resolved.Further, the system is a known dust producer (Williams etal. 1985, Marchenko et al. 1999, Williams et al. 2001), somulti-wavelength interferometric observations could revealthe location of dust formation in these particular collidingwind systems that form dust spirals (e.g., Tuthill et al. 2008).WR 138 (HD 193077) is a potential long-period binaryconsisting of a WN5o (Smith et al. 1996) and an O9 star(Annuk 1990), although the classification of the secondaryhas not been well constrained. Annuk (1990) presented theorbital elements for the Wolf-Rayet and O components. Theputative orbit has a period of P = . years (1538 d) and e = . . The period was confirmed to be 1521 ±
35 d with addi-tional data presented by Palate et al. (2013), although theirnewer data were too sparse to better fit the orbital elements.Further, Palate et al. (2013) found that the X-ray emissionwas fully consistent with a colliding winds binary. The Wolf-Rayet radial velocity curve is much better defined than thatof the O star, which shows an apparent, low-amplitude, vari-ation in anti-phase to the WR star. With a measurement of a sin i = . AU, we expect a separation of ∼ ). However, the O-component absorption lines of WR138 are very broad and shallow, leading Massey (1980) toclaim that WR 138 was a single, newly formed WR star withintrinsic broad absorption lines. Interferometry can resolvethe issue by either resolving two stars or showing an unre-solved star. With very broad absorption lines, the secondarywould more likely be a main sequence star.In this paper, we present interferometric observationsof WR 137 and WR 138 with the CHARA Array that showthat both of these systems are resolved with long-baselinenear-infrared interferometry. Section 2 outlines our interfer-ometric observations. In Section 3, we show the binary fitsfor the observations. We discuss these systems in Section 4, as listed in Rosslowe & Crowther (2015) athttp://pacrowther.staff.shef.ac.uk/WRcat/MNRAS , 1–11 (2016) HARA resolves WR 137 and WR 138 µ m)01 × -13 × -13 × -13 × -13 × -13 F L U X ( e r g s s - c m - A - s - ) Figure 1.
NIR spectra of WR 137 (black) and WR 138 (red)from Shara et al. (2012) are shown, with an H -band filter re-sponse function overplotted as a dashed line. For WR 137, thestrong emission line of C iv λ . µ m can drastically alter the in-strumental response of the CHARA Array, and were accountedfor through comparison of the Fourier spectrum of the fringe en-velopes of the calibrator and target stars. In contrast, the weakerlines in the spectrum of WR 138 account for very little differencein the instrumental response. and present goals of future studies and conclude this paperin Section 5. We collected long baseline near-infrared interferometry ofWR 137 and WR 138 using the CHARA Array (ten Brum-melaar et al. 2005) and CLIMB beam combiner (ten Brum-melaar et al. 2013) in the H − band during the nights of 2013August 13–14. The CHARA Array is a (cid:5) -shaped interfer-ometric array of six 1-m telescopes with baselines rangingfrom 34 to 331 meters in length. Our observations usedlonger baselines consisting of the S1, E2, and W1 telescopeswhich yield maximum baselines between 251 and 278 meters.The observations are summarized in Table 1.To measure the instrument response and calibrate ourdata, we observed calibrator stars with small angular di-ameters both before and after each observation of a targetstar. Namely, we observed the calibrator stars listed in Ta-ble 2. In order to calibrate correctly our data, we fit thesestars’ spectral energy distributions with Kurucz model at-mospheres corresponding to the spectral types previouslypublished. The resulting angular diameters, and publishedspectral types are also given in Table 2, in good agreementwith Lafrasse et al. (2010), and are all unresolved by theCHARA Array in the H -band, where the resolution limitfor a single star is ≈ . Wolf-Rayet stars have strong emission lines throughout the opticaland NIR spectrum (Fig. 1). These emission lines reduce theeffective bandpass over which the fringe amplitude is mea-sured causing the true visibility to be smaller than if a fixedbandpass was assumed. We measured the effective bandpassthrough comparisons of the width of the power spectra ofboth calibrator stars and the targets, giving a reasonable ap-proximation to the instrumental response. This effect causedthe visibilities of WR 137 to change by − ( u , v ) plane (Fig. 2), meaning that any fits to the data shouldprovide meaningful measurements of the separation, positionangle, and flux ratio for the two stars. We also note that fur-ther measurements of these systems were not possible duringthe observing window as a target-of-opportunity (Nova Del2013) became a priority (Schaefer et al. 2014). However, theobservations obtained allow for measurements of the binarynature of these stars. Interferometers measure the fringe contrast or visibility ofa source. The complex visibility of a binary system variesperiodically (e.g., Boden et al. 2000): V = V + f V exp [ − π i ( u ∆ α + v ∆ δ )](1 + f ) where ∆ α and ∆ δ represent the binary separation in rightascension and declination, u and v represent the spatial fre-quencies of the Array projected on the sky, f represents theflux ratio of the binary, and V and V are the uniform diskvisibilities of the primary and secondary components. In ourcase, we assume the stars are unresolved ( V = V = . ), asa main sequence O star or the emitting region for tau =1 optical depth would be smaller than 0.5 mas at the es-timated distance of these systems. The real and imaginaryparts of the complex visibilities are used to compute thesquared visibility amplitude and closure phase to comparewith the observations.We began our analysis using an adaptive grid searchprocedure where we searched for binary solutions througha large grid of separations in right ascension and decli-nation. At each step in the grid, we optimized the posi-tion ( ∆ RA, ∆ DEC) and flux ratio of the binary using theLevenberg-Marquardt least squares minimization routinempfit (Markwardt 2009), and computed the χ statistic foreach solution. We performed the adaptive grid search over arange of ±
16 mas in ∆ RA and ∆ DEC using 0.1 mas steps andkept the solution with the lowest χ as the global best-fit bi-nary model. If the components in the binary are separatedby more than the coherence length ( ∼ −
278 m baselines in the H -band), thenthey will begin to show two separated fringe packets in theinterferometric scans (e.g., Farrington et al. 2010, 2014). We http://cow.physics.wisc.edu/ ∼ craigm/idl/idl.htmlMNRAS , 1–11 (2016) N. D. Richardson et al.
Table 1.
CHARA Interferometric Observing LogUT Date Baselines Baseline Length(s) [m] N N Calibrator HD number(s)2013 Aug 14 S1/E2/W1 278, 251, 278 2 2 191703, 1928042013 Aug 15 S1/E2/W1 278, 251, 278 3 0 191703, 192804, 192536 -200 -100 0 100 200 u (10 cycles radian -1 )-200-1000100200 v ( cyc l e s r ad i an - ) -200 -100 0 100 200 u (10 cycles radian -1 )-200-1000100200 v ( cyc l e s r ad i an - ) Figure 2.
The on-sky ( u , v ) coverage for WR 137 (left) and WR 138 (right). While fewer observations were made on WR 138, the ( u , v ) plane is still well-covered. Table 2.
Calibrator StarsHD number Spectral Type Reference Angular Diameter [mas] Nights used191703 F0V Honeycutt & McCuskey (1966) . ± .
13, 14 Aug 2013192804 F8V Ljunggren & Oja (1961) . ± .
13, 14 Aug 2013192536 A7III Barbier (1963) . ± .
14 Aug 2013 selected a maximum search range of 16 mas, correspondingto twice the coherence length, to represent the separation atwhich the two fringe packets are completely separated andno longer overlap. Beyond this range, binary components ofsimilar brightnesses would be detected through a visual in-spection of the fringes. In order to search out to these widerseparations, we added bandwidth smearing to the binarymodel assuming a rectangular bandpass profile (Kraus et al.2005).This method utilized both measurements of visibilityand closure phase, which helps to remove the 180 ◦ ambiguityin the position angle. The results of these fits are summa-rized in Table 3, along with the phasing of the binary systemsfrom Annuk (1990; WR 138) and Lef`evre et al. (2005; WR137), and the fits are shown in the plots of the data in Figs.3 and 4. In general, the model works well for both systems,clearly showing a binary in both cases. Note that in the caseof a single star, the visibility function follows a simple Besselfunction (e.g., Boyajian et al. 2012), and would also follow amonotonic decrease in the case of a single star with a largewind (e.g., P Cygni; Richardson et al. 2013). As the obser-vations appear as a scatter plot in the upper left panels ofFigs. 3 and 4, a resolved binary is the simplest explanationfor the systems.We examined the resulting χ maps from the grid-searching routine, which show an ambiguity in the positionof the secondary star reflected across the origin. However,in these mirrored solutions the flux ratio of the binary isalso flipped, so in essence, the solutions are identical. Thesolution with the minimum χ in the maps is reported in Table 3. There are other possible solutions, however, nonefall within ∆ χ = 3.53 (the 1 σ confidence interval for threefit parameters). These alternative solutions could be furtherruled out by acquiring additional observations in the futureand fitting the orbital motion directly to the visibilities andclosure phases across multiple epochs.A visibility near unity would be indicative of a com-pletely unresolved source. If we are partially resolving theWolf-Rayet wind or the O star in these systems, the peak inthe visibility curves would be lower than 1. For WR 137 andWR 138, the measured visibilities that are closest to unityare all within the errors from the binary model that assumesthe component stars are unresolved. Moreover, the numberof data points is not dense enough to include the angulardiameters of the component stars as free parameters in thefit. According to our interferometric measurements for both WR137 and WR 138, one component slightly dominates overthe other in the H-band. While calibrations of H -band mag-nitudes with spectral types (Martins et al. 2005, Rosslowe& Crowther 2015) imply that the WR component domi-nates in the H -band in both systems, the typical magnitudescatter portrayed by WR stars hinders us to conclude thiswith certainty. To ensure that the relative flux contributionsare correctly assigned to the correct components of WR 137 MNRAS , 1–11 (2016)
HARA resolves WR 137 and WR 138 Table 3.
Binary FitsStar Separation (mas) PA ( ◦ ) Date Phase f WR f O ReferenceWR 137 9.8 ± ± ± ± ± ± ± ± ± ± ± ±
220 230 240 250 260 270 280BASELINE (m)0.00.20.40.60.81.0 O BSE R VE D V V O BSE R VE D V
275 276 277 278 279 280LONGEST BASELINE (m)-15-10-505 O BSE R VE D C P ( o ) -15 -10 -5 0 5MODEL CP ( o )-15-10-505 O BSE R VE D C P ( o ) Figure 3.
CHARA observations of WR 137. The upper left plot shows the squared visibilities compared with the baseline, where thesolid points represent the measurements and the open circles connected by dotted lines represent the best model (Table 3) for thebinary fit. The upper right compares the model and the measurements, with a 1:1 relation overplotted. The lower left panel shows ourfour measurements of closure phase (solid points) compared with the longest baseline from the three telescope configuration, with themodel points connected in the same manner as the upper left panel. The lower right panel shows the comparison of the model and themeasurements. and WR 138, we performed a spectral analysis of both sys-tems using the Potsdam Wolf-Rayet (PoWR ; Hamann &Gr¨aefner 2004) non-LTE model atmosphere code, suitablefor any hot stars with winds. Comparing synthetic spectrawith observations further provides us with the fundamentalparameters of the components of each system.A thorough description of the code is beyond the scopeof this paper; we only repeat fundamental assumptions here.The prespecified velocity field v ( r ) takes the form of the β -law(Castor et al. 1975) with β = and the terminal velocity v ∞ as a free parameter. Clumping is treated via the microclump-ing approach (Hillier 1984). The clumping factor D is treatedas a free parameter for WR models, and is fixed to D = in the winds of O-star models (e.g., Feldmeier et al. 1997).The depth-dependent microturbulence parameter ξ ( r ) growslinearly with v ( r ) from the photospheric value ξ ph to . v ∞ inthe wind, where ξ ph = km s − for O models and ξ ph = kms − for WR models (e.g., Shenar et al. 2015). Macroturbu-lence is accounted for by convolving the synthetic spectrawith appropriate radial-tangential profiles with v mac = kms − (e.g., Gray 1975, Bouret et al. 2012). A more detailed de-scription of the assumptions and methods used in the code isgiven by Gr¨aefener et al. (2002), Hamann & Gr¨afener (2004),and Sander et al. (2015). MNRAS , 1–11 (2016)
N. D. Richardson et al.
220 230 240 250 260 270 280BASELINE (m)0.00.20.40.60.81.0 O BSE R VE D V V O BSE R VE D V
275 276 277 278 279 280LONGEST BASELINE (m)-1001020304050 O BSE R VE D C P ( o ) -10 0 10 20 30 40 50MODEL CP ( o )-1001020304050 O BSE R VE D C P ( o ) Figure 4.
CHARA observations of WR 138, with the same format as Fig. 3.
To analyze the systems, we use a multitude of spec-tra ranging from the UV to the IR. For both objects, allhigh-resolution, flux calibrated spectra taken with the In-ternational Ultraviolet Explorer (IUE) in the spectral range − were retrieved from the MAST archive. Sincethe changes with orbital phase in both systems are extremelysmall, we co-added these spectra to obtain a higher signal-to-noise (S/N ≈ ) for each system. We also retrieved IUEspectra covering the range − for both systems fora larger coverage of the spectral energy distribution (SED),available in the MAST archive. For WR 137, we utilized ahigh resolution, echelle spectrum from the Keck II telescopeand the ESI spectrograph. These data were taken near intime to our interferometry, and covered ∼ ˚A–1 µ m, witha typical S/N across the spectrum of 500 or better. For WR138, we use complementary, low resolution ( ∆ λ ≈ ) op-tical spectra to cover the spectral ranges − ˚Aand − ˚A, taken between 1991 June 24 – July 1 withthe 2.2m-telescope in Calar Alto, Spain (for further details,see Hamann et al. 1995) and low-resolution optical spectraobtained at the Observatoire du Mont M´egantic (Qu´ebec)near in time to the interferometric observations, spanning ∼ − ˚A. We also used the NIR spectra obtained forspectral typing of infrared WR stars by Shara et al. (2012).All spectra were rectified by fitting a low-order polynomialto the apparent continuum.For both systems, we take UBV magnitudes from Neckel et al. (1980), R and I magnitude from Monet et al. (2003),and JHK magnitudes from Kharchenko (2001). The syn-thetic spectra are convolved with Gaussians of appropriatewidths to mimic the spectral resolution. For WR 137, wealso recovered MSX photometry from Egan et al. (2003) andWISE photometry from Cutri et al. (2012). Fortuitously,these missions observed WR137 close to the 1997 dust-formation maximum and that expected in 2010 (Williamset al. 2001) respectively. The fluxes are marked A, C, D andE (MSX) and W1 and W2 (WISE) in Fig. 6, where they liewell above the stellar model SED and were not used in thefittingWe represent each binary system as a composite of twomodels corresponding to their component spectra. As a firststep, we calibrate the fundamental parameters of the compo-nents against their spectral types. These include the effectivetemperatures T ∗ and transformed radii R t (which are relatedto the mass-loss rates ˙ M , see Schmutz et al. 1989) for theWR stars, and T ∗ , ˙ M and gravities g ∗ for the O components.We note that T ∗ and g ∗ refer to the inner boundary of ourmodels, where the mean Rosseland optical depth τ Ross = .For the O companions, the values of these parameters vir-tually coincide with the photospheric values at τ Ross = / .With the parameters fixed, the light ratio can be de-rived by examining a multitude of features in the availablespectra, primarily in the UV and optical. An example isshown in Fig. 5, where the best-fitting models for both sys- MNRAS , 1–11 (2016)
HARA resolves WR 137 and WR 138 ❲❘ ✶✸✼(cid:0)❈✁ ✰ ❖✽✳✺ ❱❍❡■ ✂■✄ ❍❡■✵☎✹✵☎✻✵☎✆✝☎✵✝☎✷✝☎✹✝☎✻ ✹✞✆✵ ✹✹✝✵ ✹✹✹✵ ✹✹✟✵❧ ✴ ❆♦◆✠r♠❛✡✐③☛❞❢✡✉① ❍❡■ ✂■✄ ❍❡■❲❘ ✶✸☞(cid:0)✌✺ ✰ ❖✾ ❱✵☎✹✵☎✻✵☎✆✝☎✵✝☎✷ ✹✞✆✵ ✹✹✝✵ ✹✹✹✵ ✹✹✟✵❧ ✴ ❆♦ Figure 5.
The synthetic composite spectrum (red dotted line) iscompared to observations (blue ragged lines) of WR 137 and 138in the vicinity of the He i λ line. The relative offsets of theWR (black solid line) and O (green dashed line) correspond tothe relative light contributions in this spectral range. Table 4.
Inferred stellar parameters for WR 137 and 138WR 137 WR 138Component WR O WR Odistance d [pc] Spectral type WC7pd O9 V WN5o O9 V T ∗ [kK] + − + − + − + − log g ∗ [cm s − ] - . + . − . - . + . − . log L [ L ⊙ ] . + . − . . + . − . . + . − . . + . − . log R t [ R ⊙ ] . + . − . - . + . − . - v ∞ [km s − ] + − + − + − + − R ∗ [ R ⊙ ] . + − . + − . + − . + − D M − . + . − . − . + . − . − . + . − . − . + . − . v sin i [km s − ] - + − - + − M V [mag] − . + . − . − . + . − . − . + . − . − . + . − . M H [mag] − . + . − . − . + . − . − . + . − . − . + . − . E B − V [mag] . + . − . . + . − . A V [mag] . + . − . . + . − . tems are compared to optical observations in the vicinityof the He i λ line. This line maintains an almost identi-cal equivalent width in the temperature domain − kK,which is the relevant temperature domain for both O com-panions concerned here. Its strength is therefore directly re-lated to the dilution factor, i.e., the light ratio in the visual.In contrast, the He i λ line, also shown in Fig. 5, is sensi-tive to T ∗ and g ∗ . The derived light ratios are consistent witha multitude of features in the available spectra. We furtherconfirmed our results by comparing with equivalent widthcalibrations for single O stars (e.g.,Conti & Alschuler 1971).The stellar parameters are then further refined until nosignificant improvement can be achieved. Wind parametersare derived based on the strengths and shapes of emissionlines. For the O stars, v ∞ could be constrained from theC iv and Si iv resonance lines. Their mass-loss rates are veryroughly constrained based on the existence of P Cygni or emission features clearly associated with the O componentsor lack thereof. The projected rotational velocity v sin i isderived primarily from He i and He ii lines (see e.g. Fig. 5).The components’ luminosities L and the reddening E B − V arederived by comparing the composite synthetic spectrum tothe spectral energy distribution (SED) of each system. Weassume the reddening law given by Seaton (1979) with R V = . . We adopt distances from Rosslowe & Crowther (2015),which result in consistent luminosities for both parameters.The SEDs and spectra along with the best-fitting mod-els are shown in Figs. 6 and 7. The derived parameters arelisted in Table 4. Uncertainties are estimated based on thesensitivity of the fit quality to changes in the correspond-ing parameter. The Table also gives stellar radii R ∗ (calcu-lated via the Stefan-Boltzmann law), V − and H − band mag-nitudes, and extinctions A V . Our results turn out to be veryconsistent with what is expected for stars of the given spec-tral types. The derived light ratios indeed imply that theWR component is the dominant source in the H -band inboth systems. For WR 137, the WR component contributes(59 ± H -band flux as derived by the interferome-try, and is seen to contribute a fractional flux of 0.59 in thePoWR model. Similarly, for WR 138, we measured the WRcomponent to contribute (67 ± H -band flux, withthe PoWR results show the fraction to be 0.66. These mea-surements show the strong agreement in the two methods,and strengthen the results of both investigations. The interferometric measurements of WR 138 represent thefirst time a WN star has been resolved in a spectroscopicbinary. We note that the fundamental parameters derivedthrough our spectroscopic modeling of the binary systemagree with the expectations for a late O dwarf (Martins etal. 2005, Martins & Plez 2006) as the companion. The ab-sorption lines in the system are very broad and shallow (seeFigs. 5 and 7), and have caused some controversy in the lit-erature. The star was first called a binary based upon theabsorption lines of hydrogen and neutral helium in the blueby Hiltner (1945). Massey (1980) studied a time-series ofmoderate-resolution photographic spectra to find that thesemi-amplitude of any binary would likely be . km s − for periods less than 6 months, for which his radial velocitystudy is most sensitive.Lamontagne et al. (1982) examined a longer time-seriesof spectra of WR 138, including the measurements of Massey(1980) and Bracher (1966). They found a long-period orbit,with a period of ∼ d, and attributed short-term variabil-ity of the WR emission lines as arising from a short-period,2.3 d, orbit with a neutron star. The short-period orbit wasnot confirmed when Annuk (1990) later examined the sys-tem and concluded that the system was a spectroscopic bi-nary, with a period of ∼ d. The O star is fairly exoticcompared to most O dwarfs, as the measured value of v sin i is roughly km s − (Hamann et al. 2006, Massey 1980,Annuk 1990), while our spectral modeling indicates a lowervalue of ∼ km s − . This is incredibly high in comparisonto the population of O stars both in the Galaxy (Howarthet al. 1997), or even in the massive, star-forming region of30 Dor (Ram´ırez-Agudelo et al. 2013), where the measured MNRAS , 1–11 (2016)
N. D. Richardson et al.
Figure 6.
The best fitting synthetic SED (upper panel) and normalized spectrum (lower panels) compared to observations of WR 137(blue lines and squares). The composite model is the superposition of the WR (black solid line) and O (green dashed line) models. Therelative offsets of the model continua shown in the lower panels depict the relative light contribution in each spectral range. The fewarrows shown in the upper panel illustrate the IR excess variability in the system from Williams et al. (2001). Note that the CHARAobservations were collected in the H -band ( log λ ≈ . ). MNRAS , 1–11 (2016) HARA resolves WR 137 and WR 138 Figure 7.
As Fig. 6, but for WR 138. Note that the CHARA observations were collected in the H -band ( log λ ≈ . )MNRAS , 1–11 (2016) N. D. Richardson et al. v sin i would place the companion in the top − % of thepopulation.The period derived by Annuk (1990) and Palate etal. (2013) is quite long (4.2 yr), and should be longer thantypical orbits associated with binary interactions (e.g., Sanaet al. 2012). However, the period limit for a zero-age main-sequence O star to experience spin-up from a companionfrom the results of Sana et al. (2012) is actually similar tothe orbital period of WR 138. Therefore, the extreme rapidrotation of this companion may actually be the remnant ofa past interaction where the current WR star lost mass viaRoche lobe overflow which then deposited angular momen-tum and mass onto the O star. If the semi-major axis isconserved in this process, then if the original system were + M ⊙ stars that interacted to become + M ⊙ starsthrough a combination of mass loss and accretion, the pe-riod would have originally been shorter ( < d) and nearthe adopted upper limits for an interacting binary as definedby Sana et al. (2012).These interferometric observations of WR 137 representthe second reported binary measurements of this star. Ra-jogopal (2010) found a binary flux ratio equal to our mea-surements. However, the projected separations between ourmeasurements and those of Rajagopal (2010) differ by abouta factor of two, and they differ by ∼ ◦ , but were takenat opposite quadratures. Two measurements of separationand position angle from two different instruments are notenough to warrant an attempt of a visual orbit. Further, wenote that we are unsure if Rajagopal (2010) accounted forthe contamination of emission lines in the interferometry,which can alter the measured separation and position an-gle. However, it seems that the eventual orbital solution willfavor a system with a nearly edge-on geometry.A directly edge-on system will have masses that are per-haps lower than expected in the WR 137 system, but thereare supporting reasons to trust the inclination to be closer toedge-on. Marchenko et al. (1999) imaged the system in the H - and K -bands with the HST and the NICMOS camera.These images were taken during and after a periastron pas-sage when dust is formed. These images revealed structuresin the K -band, but not in the H -band, that were thoughtto have emerged from the wind-wind collision region. It isinteresting that the position angle of the structure imagedby HST is ≈ ◦ . If the positions measured by the IOTAand CHARA interferometers are indeed pointing towards anedge-on orbit, then the dust formation may occur near theorbital plane. Further observations can elucidate the detailsof the orbit and the formation region of the dust. We alsonote that the opening angle of the shock cone is ≈ ◦ ac-cording to the approximations given by Gayley (2009) andour spectral modeling, so the small opening angle in the HST /NICMOS imaging is consistent with this explanation.Clearly, the next step in these systems is two-fold. Weneed to resolve the orbits at multiple epochs in order to mea-sure the visible orbit of the systems. This will lead to con-straints on the inclinations of the system and the orbit. Con-temporaneous to this, we need to re-determine the double-lined spectroscopic orbits of these systems as both orbits arepoorly constrained for the O star components. The O starin the WR 138 system is extremely difficult to measure dueto its large value for v sin i , so high resolution spectroscopywith high S/N is needed. The resolution and interferometric followup of WR 137 and WR 138 will double the number ofthe masses measured for Wolf Rayet stars in longer periodorbits, with WR 138 being the first WN star with a visualorbit. Future interferometry of WR 137 would benefit frommeasurements made in both the H − and K − band near intime so that the the binary parameters can first be resolvedin the H − band. These derived parameters can then be usedto pinpoint the exact place where the dust formation hap-pens in the close environs of the system. Such observationscan be used to further understand the K -band HST imagingof the system that was reported by Marchenko et al. (1999).WR 137 provides an exciting example of a WC + O systemwhere we can determine empirically the location of the dustformation through long-baseline infrared interferometry andthen compare the close environs of the system’s dust produc-tion to the larger scale imaging reported by Marchenko etal. (1999).
ACKNOWLEDGEMENTS
We wish to thank Michael Shara, Jacqueline Faherty, andGraham Kanarek for allowing us to use their NIR spectraof these two stars. We thank the CHARA staff for support-ing these observations and the Mount Wilson Institute forits continued support of the CHARA Array. This material isbased upon work supported by the National Science Founda-tion under Grants AST-1211929 and AST-1411654. Some ofthe spectroscopic data were obtained at the W. M. Keck Ob-servatory on the summit of Mauna Kea. The authors wishto recognize and acknowledge the very significant culturalrole and reverence that the summit of Mauna Kea has al-ways had within the indigenous Hawaiian community. Weare most fortunate to have the opportunity to conduct obser-vations from this mountain. The
IUE data presented in thispaper were obtained from the Mikulski Archive for SpaceTelescopes (MAST). STScI is operated by the Associationof Universities for Research in Astronomy, Inc., under NASAcontract NAS5-26555. Support for MAST for non-HST datais provided by the NASA Office of Space Science via grantNNX09AF08G and by other grants and contracts.NDR acknowledges postdoctoral support by the Uni-versity of Toledo and by the Helen Luedtke Brooks En-dowed Professorship, and is thankful for his former CRAQ(Qu´ebec) fellowship which supported him at the beginningof this research. TS is grateful for financial support from theLeibniz Graduate School for Quantitative Spectroscopy inAstrophysics, a joint project of the Leibniz Institute for As-trophysics Potsdam (AIP) and the institute of Physics andAstronomy of the University of Potsdam. AFJM and NSLare grateful for financial aid from NSERC (Canada) andFQRNT (Quebec). PMW is grateful to the Institute for As-tronomy for continued hospitality and access to the facilitiesof the Royal Observatory Edinburgh.
REFERENCES
Aldoretta, E. J., Caballero-Nieves, S. M., Gies, D. R., et al. 2015,AJ, 149, 26Annuk, K. 1990, AcA, 40, 267Barbier, M. 1963, POHP, 6, 36 MNRAS , 1–11 (2016)
HARA resolves WR 137 and WR 138 Boden, A. F. 2000, in Principles of Long Baseline Stellar Inter-ferometry, ed. P. R. Lawson (Pasadena, CA: JPL), 9Bouret, J. C., Hillier, D. J., Lanz, T., & Fullerton, A. W. 2012,A&A, 544, 67Boyajian, T. S., McAlister, H. A., van Belle, G., et al. 2012, ApJ,746, 101Boyajian, T. S., van Belle, G., & von Braun, K. 2014, AJ, 147, 47Bracher, K. 1966, Ph.D. thesis, Indiana UniversityCastor, J. I., Abbott, D. C., & Klein, R. I. 1975, ApJ, 195, 157Conti, P. S., & Alschuler, W. R. 1971, ApJ, 170, 325Cutri, R. M., et al. 2012, VizieR Online Data Catalogs: II/311De Marco, O., & Schmutz, W. 1999, A&A, 345, 163De Marco, O., Schmutz, W., Crowther, P. A., et al. 2000, A&A,358, 187Egan, M. P., Price, S. D., Kraemer, K. E., et al. 2003, VizieROnline Data Catalogs: V/114Fahed, R., Moffat, A. F. J., Zorec, J., et al. 2011, MNRAS, 418,2Farrington, C. D., ten Brummelaar, T. A., Mason, B. D., et al.2010, AJ, 139, 2308Farrington, C. D., ten Brummelaar, T. A., Mason, B. D., et al.2014, AJ, 148, 48Feldmeier, A., Puls, J., & Pauldrach, A. W. A. 1997, A&A, 322,878Gayley, K. G. 2009, ApJ, 703, 89Gr¨afener, G., Koesterke, L., & Hamann, W.-R. 2002, A&A, 387,244Gray, D. F. 1975, ApJ, 202, 148Hamann, W.-R., & Gr¨aefener, G. 2004, A&A, 427, 697Hamann, W.-R., Gr¨aefener, G., & Liermann, A. 2006, A&A, 457,1015Hamann, W.-R., Koesterke, L., & Wessolowski, U. 1995, Astron.& Astrophys. Suppl., 113, 459.Hanbury Brown R., Davis J., Herbison-Evans D., Allen L. R.1970, MNRAS, 148, 103Hillier, D. J. 1984, ApJ, 280, 744Hiltner, W. A. 1945, ApJ, 101, 356Honeycutt, R. K., & McCuskey, S. W. 1966, PASP, 78, 289Howarth, I. D., Siebert, K. W., Hussain, G. A. J., Prinja, R. K.1997, MNRAS, 284, 265Kharchenko, N. V. 2001, KFNT, 17, 409Kraus, S., Schloerb, F. P., Traub, W. A., et al. 2005, AJ, 130, 246Lafrasse S., Mella G., Bonneau D., et al. 2010, Proc. SPIE, 7734,140Lamontagne, R., Koenigsberger, G., Seggewiss, W., & Moffat, A.F. J. 1982, AJ, 253, 230Lef`evre, L., Marchenko, S. V., L´epine, S., et al. 2005, MNRAS,360, 141Ljunggren, B., & Oja, T. 1961, Uppsala Astron. Obs. Ann., 4, 1Marchenko, S. V., Moffat, A. F. J., & Grosdidier, Y. 1999, ApJ,522, 433Markwardt, C. B. 2009, in Astronomical Society of the PacificConference Series, Vol. 411, Astronomical Data Analysis Softwareand Systems XVIII, ed. D. A. Bohlender, D. Durand, & P. Dowler,251Martins, F., & Plez, B. 2006, A&A, 457, 637Martins, F., Schaerer, D., & Hillier, D. J. 2005, A&A, 436, 1049Massey, P. 1980, ApJ, 236, 526Millour F., Driebe T., Chesneau O., et al. 2009, A&A, 506, L49Millour F., Petrov, R. G., Chesneau O., et al. 2007, A&A, 464,107Monet, D. G., Levine, S. E., Canzian, B., et al. 2003, AJ, 125,984Monnier, J. D., Traub W. A., Schloerb F. P., et al. 2004, ApJ,602, L57Monnier, J. D., Zhao, M., Pedretti, E., et al. 2011, ApJ, 742, L1Neckel, T., Klare, G., & Sarcander, M. 1980, BICDS, 19, 61 North, J. R., Tuthill P. G., Tango W. J., & Davis J. 2007, MN-RAS, 377, 415Nugis, T. & Lamers, H. J. G. L. M. 2000, A&A, 360, 227Palate, M., Rauw, G., De Becker, M., Naz´e, Y., & Eenens, P.2013, A&A, 560, 27Pauls, T. A., Young, J. S., Cotton, W. D., & Monnier, J. D. 2005,PASP, 117, 1255Rajagopal, J. 2010, in Revista Mexicana de Astronomia y As-trofisica Serie de Conferencias, Vol. 38, 54Ram´ırez-Agudelo, O. H., Sim´on-D´ıaz, S., Sana, H., et al. 2013,A&A, 560, 29Richardson, N. D., Schaefer, G. H., Gies, D. R., et al. 2013, ApJ,769, 118Rosslowe, C. K., & Crowther, P. A. 2015, MNRAS, 447, 2322Sana, H., de Mink, S., de Koter, A., et al. 2012, Science, 337, 444Sana, H., Le Bouquin, J.-B., Lacour, S., et al. 2014, ApJS, 215,15Sander, A., Shenar, T., Hainich, R., et al. 2015, A&A, 577, 13Schaefer, G. H., ten Brummelaar, T., Gies, D. R., et al. 2014,Nature, 515, 234Schmutz, W., Hamann, W.-R., & Wessolowski, U. 1989, A&A,210, 236Schmutz, W., Schweickhardt, J., Stahl, O., et al. 1997, A&A, 328,219Seaton, M. J. 1979, MNRAS, 187, 73Shenar, T., Oskinova, L., Hamann, W.-R., et al. 2015, ApJ, 809,135Shara, M. M., Faherty, J. K., Zurek, D., et al. 2012, AJ, 143, 149Smith, L. F., Shara, M. M., & Moffat, A. F. J. 1996, MNRAS,281, 163Smith, N., & Owocki, S. 2006, ApJ, 645, L45ten Brummelaar, T. A., McAlister, H. A., Ridgway, S. T., et al.2005, ApJ, 628, 453ten Brummelaar, T. A., Sturmann, J., Ridgway, S. T., et al. 2013,JAI, 2, 1340004Tuthill, P., Monnier, J. D., Lawrance, N., et al. 2008, ApJ, 675,698Williams, P. M., Kidger, M. R., van der Hucht, K. A., et al. 2001,MNRAS, 324, 156Williams, P. M., Longmore, A. J., van der Hucht, K. A., et al.1985, MNRAS, 215, 23Williams, P. M., Marchenko, S. V., Marston, A. P., et al. 2009,MNRAS, 395, 1749This paper has been typeset from a TEX/L A TEX file prepared bythe author.MNRAS000