Observational evidence for a correlation between macroturbulent broadening and line-profile variations in OB Supergiants
S. Simón-Díaz, A. Herrero, K. Uytterhoeven, N. Castro, C. Aerts, J. Puls
aa r X i v : . [ a s t r o - ph . S R ] A ug Observational evidence for a correlation between macroturbulent broadening and line-profile variations in OB Supergiants S. Sim´on-D´ıaz
Instituto de Astrof´ısica de Canarias, E-38200 La Laguna, Tenerife, Spain.Departamento de Astrof´ısica, Universidad de La Laguna, E-38205 La Laguna, Tenerife, Spain. [email protected]
A. Herrero
Instituto de Astrof´ısica de Canarias, E-38200 La Laguna, Tenerife, Spain.Departamento de Astrof´ısica, Universidad de La Laguna, E-38205 La Laguna, Tenerife, Spain.
K. Uytterhoeven
Laboratoire AIM, CEA/DSM-CNRS-Universit´e Paris Diderot; CEA, IRFU, SAp, centre de Saclay,F-91191, Gif-sur-Yvette, France.
N. Castro
Instituto de Astrof´ısica de Canarias, E-38200 La Laguna, Tenerife, Spain.Departamento de Astrof´ısica, Universidad de La Laguna, E-38205 La Laguna, Tenerife, Spain.
C. Aerts
Instituut voor Sterrenkunde, Katholieke Universiteit Leuven, Celestijnenlaan 200D, B-3001 Leuven,BelgiumIMAPP, Department of Astrophysics, Radboud University Nijmegen, PO Box 9010, NL-6500 GLNijmegen, the Netherlands and
J. Puls
Universit¨atssternwarte M¨unchen, Scheinerstr. 1, D-81679 M¨unchen, Germany
ABSTRACT
The spectra of O and B supergiants are known to be affected by a significant form of extraline broadening (usually referred to as macroturbulence ) in addition to that produced by stellarrotation. Recent analyses of high resolution spectra have shown that the interpretation of thisline broadening as a consequence of large scale turbulent motions would imply highly supersonicvelocity fields in photospheric regions, making this scenario quite improbable. Stellar oscillationshave been proposed as a likely alternative explanation. As part of a long term observationalproject, we are investigating the macroturbulent broadening in O and B supergiants and itspossible connection with spectroscopic variability phenomena and stellar oscillations. In thisletter, we present the first encouraging results of our project, namely firm observational evidencefor a strong correlation between the extra broadening and photospheric line-profile variations ina sample of 13 supergiants with spectral types ranging from O9.5 to B8.
Subject headings: stars: early-type — stars: atmospheres — stars: oscillations — stars: rotation —supergiants . Introduction The presence of an important extra line broad-ening mechanism (in addition to the rotationalbroadening and usually called macroturbulence )affecting the spectra of O and B supergiants(Sgs) was initially suggested by the deficit ofnarrow lined objects among these types of stars(e.g. Slettebak 1956; Conti & Ebbets 1977;Howarth et al. 1997). The advent of high-qualityspectra permitted confirmation that rotationalbroadening alone is insufficient to fit the line pro-files in many objects, and to investigate the possi-ble disentangling of both broadening contributions(see e.g. Ryans et al. 2002; Sim´on-D´ıaz & Herrero2007). These studies definitely showed that the ef-fect of macroturbulence is dominant in the profilesof metal lines in early B Sgs.Although it was named macroturbulence atsome point, the interpretation of this extra broad-ening as the effect of turbulent motion is quiteimprobable. The effect is present in photosphericlines and affects the whole profile, even wave-lengths close to the continuum. Therefore, what-ever is producing the extra broadening has to bedeeply rooted in the stellar photosphere (and pos-sibly deeper), in layers where we do not expect anysignificant velocity field in these stars.If interpreted as turbulent motion, macroturbu-lence would represent highly supersonic velocitiesin many cases (see Figure 1). This interpretationis incompatible with the previous statement.One physical mechanism suggested as the ori-gin of this extra broadening relates to oscillations.Many OB Sgs are known to show photometric andspectroscopic variability. Based on this, Lucy(1976) postulated that this variability might be apulsation phenomenon, and macroturbulence maybe identified with the surface motions generatedby the superposition of numerous non-radial os-cillations. More recently, Aerts et al. (2009) com-puted time series of line profiles for evolved mas-sive stars broadened by rotation and thousandsof low amplitude non-radial gravity mode oscilla-tions and showed that the resulting profiles could Based on observations made with the Nordic OpticalTelescope, operated on the island of La Palma jointly byDenmark, Finland, Iceland, Norway, and Sweden, in theSpanish Observatorio del Roque de los Muchachos of theInstituto de Astrofisica de Canarias.
Fig. 1.— Macroturbulent velocities for B Sgs withspectral types B0-B3, measured in recent stud-ies. Two different symbols are used dependingon the type of definition assumed for the macro-turbulent profile. Diamonds: isotropic Gaus-sian (Dufton et al. 2006; Fraser et al. 2010); tri-angles: radial-tangential definition (Lefever et al.2007; Markova et al. 2008). Characteristic valuesfor the sound speed in the photospheres of thesetypes of stars are also indicated as a solid line.mimic the observed ones. In their paper, Aerts etal. (2009) considered the amplitudes of the pulsa-tion modes to be sufficiently low to allow a lin-ear superposition of mode velocities to derive theoverall pulsational velocity eigenvector, i.e., non-linear mode coupling or other nonlinear effectswere ignored. We refer the reader to Sect. 2.1 inAerts et al. (2009) for a detailed description of thesimulations, and a discussion about how the col-lective effect of low amplitude non-radial gravitymodes can produce the inferred supersonic level ofmacroturbulent broadening when the oscillationsare ignored in the interpretation of spectral lines.Stellar oscillations are a plausible explanation forthe extra broadening in O and B Sgs, but so farthere is no direct observational evidence confirm-ing this hypothesis.Two years ago we began an observationalproject aimed at investigating the macroturbulentbroadening in O and B Sgs, and its possible con-nection with spectroscopic variability phenomenaand stellar oscillations. In this letter, we presentthe first encouraging results, showing observa-tional evidence for a strong correlation between2his extra broadening and photospheric line pro-file variations in a sample of 13 Sgs with spectraltypes ranging from O9 to B8.The letter is structured as follows: the sampleof stars and the spectroscopic observations usedfor this study are presented in Sect. 2; we char-acterize the line-broadening and the line-profilevariations of photospheric lines in Sects. 3 and4, respectively; a connection between the extrabroadening and line profile variations is presentedin Sect. 5. Finally, we discuss the possible relationto stellar pulsations in Sect. 6.
2. Observational data set
We obtained time series spectra of a selectedsample of 11 bright late-O and early-B Sgs duringtwo observing runs (2008/11/05-08, 2009/11/09-12) with the FIES cross-dispersed high resolutionechelle spectrograph on the 2.5m Nordic OpticalTelescope at Roque de los Muchachos Observatoryon La Palma (Canary Islands, Spain). We usedFIES in the medium resolution mode (R=46000, δλ =0.03 ˚A/pix). The sample was complementedwith two OB dwarfs and two late B Sgs. The ex-posure times were chosen so as to reach at leastSNR=200 (measured in the range 4500 – 4600 ˚A).The list of observed stars, along with their spec-tral classification, number of spectra obtained foreach target, time span of the corresponding time-series, and observing run in which they were ob-served are indicated in the first columns of Table1. Note that four of these stars were observed inboth campaigns.The spectra were reduced with the FIEStool package in advanced mode. A proper set ofbias, flat and arc frames obtained each nightwas used to this aim. The FIEStool pipelineprovided wavelength-calibrated, blaze-corrected,order-merged spectra of high quality. Next, thebarycentric correction was performed. We usedthe interstellar Na I D doublet at 5890 ˚A to checkthe accuracy in this correction, and found anagreement in the line doublet position for all thetime series spectra of each star better than 0.5km s − .Once the lines of interest for this study wereidentified, a local normalization of selected regions of the spectra was performed. For each line, the lo-cal continuum used for the normalization was cal-culated from a linear fit between points inside twocontinuum regions adjacent to the line. The samecontinuum windows (selected manually in the av-erage spectrum) were used for all the spectra of agiven star.
3. Characterization of the line broadeningin photospheric lines
Several types of broadening mechanisms aretraditionally considered to produce the final ob-served line profiles of photospheric metal linesin massive stars: instrumental, natural, thermal,microturbulent, macroturbulent and rotationalbroadening. In the case of high resolution spectraof late-O and early B Sgs, rotational and macro-turbulent constributions dominate.We used the Fourier transform technique (Gray1976; see also Sim´on-D´ıaz & Herrero, 2007 for arecent application to the spectra of O and B stars)to disentangle the contributions from rotationaland macroturbulent broadening. The results ofthe analysis are presented in Table 1. We per-formed the analysis for each of the time series spec-tra, obtaining v sin i as indicated by the first zeroof the Fourier transform. The range and medianof derived v sin i values are indicated in columns 6and 7 of Table 1. Note that the dispersion in theobtained v sin i values is between 10% and 30%,depending on the star. Whether this dispersionis real or an effect of noise is not clear from thisdata set. As outlined by Sim´on-D´ıaz & Herrero(2007), the correct identification of the first zero iscomplicated in cases of low SNR and for a large ex-tra broadening contribution. This will be exploredin more detail in a future investigation.Next, we applied a goodness-of-fit method (viz.Ryans et al. 2002) to quantify the contribution ofthe extra broadening in each spectrum from thetime series. We considered as free parameters theequivalent width, the radial velocity, v sin i , andthe parameter defining the extra broadening. Weallowed v sin i to vary in the fitting process in or-der to compare with results from the FT anal- The instrumental broadening contribution can be mini-mized by using high resolution spectra. In our FIES@NOTdata set (R=46000) the instrumental Gaussian profile hasa FWHM ∼ − . III
III v sin i , Θ RT , h v i pp , and h v i / in km s − . Uncertainties in h v i pp ,and h v i / are ∼ ∼ − , respectively v sin i (FT) χ fitting LPVsStar SpT & LC Run ∆T h v sin i i h Θ RT i h v i pp h v i / HD 207198 O9 Ib-II FI09 3.14 14 59–64 62 ± ± ± ± ± ± ± ± ± ±
10 115 ±
15 117 ±
13 12.01 62.8HD 204172 B0 Ib FI09 3.14 14 38–68 58 ±
11 58 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
11 61 ± ± ± ± ± + FI08 1.96 17 30–47 46 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ysis. Two possibilities for the characterization ofthe extra broadening were considered: an isotropicGaussian profile, and a Gaussian radial–tangentialone (Θ RT , see Gray 1976). In the latter, we as-sumed that the radial and tangential componentsprovide equal contributions to the final profile.The analysis using an isotropic Gaussian profilefor the macroturbulent broadening resulted in (a)a large dispersion in the v sin i values obtainedfrom the various spectra in each time series; (b)systematically lower v sin i values than those de-termined through the FT method (a factor of 0.5in many cases, sometimes even “zero” rotation)for all stars with dominant extra broadening. Onthe other hand, when assuming a Gaussian radial–tangential profile, we found that (a) the disper-sion in the v sin i and Θ RT values was rather low;(b) differences between the v sin i values obtainedfrom the goodness-of-fit method and using the FTtechnique are ≤ ± The corresponding We also found that the former generally produces a betterfinal fit to the observed profiles, though the differences are mean and standard deviation for v sin i and Θ RT asobtained from the χ fit are indicated in columns8 and 9 in Table 1. Note the good agreement inthe derived values for the four stars observed inboth 2008 and 2009.Similarly to previous studies (e.g. Dufton et al.2006; Lefever et al. 2007; Markova et al. 2008;Fraser et al. 2010), we found that the size of theextra broadening is generally larger than the ro-tational contribution in all the Sgs. This is notthe case for the B0.5 V star HD 37042, where thetotal broadening is mainly produced by the ef-fect of the stellar rotation (note, however, thata certain extra broadening is also needed in thiscase). The other dwarf star, HD 214680, is a spe-cial case, since it has very low v sin i . For sucha low v sin i values, microturbulence produces asignificant contribution to the total broadening,which then is included in the measured extrabroadening. Note that this argument may alsoexplain the extra broadening needed for HD 37042(and the other two late-B Sgs with low v sin i ). subtle. III
III
4. Characterization of line profile varia-tions in photospheric lines
In agreement with earlier studies of spectro-scopic variability in O and B Sgs (e.g. Ebbets1982; Howarth et al. 1993; Fullerton, Gies, & Bolton1996; Prinja et al. 1996, 2004, 2006; Morel et al.2004; Kaufer et al. 2006; Markova et al. 2005,2008), we found clear signatures of line profilevariations (LPVs) for all the Sgs considered inour sample. To investigate these LPVs quan-titatively we computed the first, h v i , and third, h v i , normalized velocity moments (for definitions,see, e.g., Aerts et al. 2010a) from the Si III
III (Zima 2008).These moments are related to the centroid veloc-ity and the skewness of the line profile, respec-tively, and are well suited for deciding whether anobserved line profile is subject to time-dependentline asymmetry (note that their values are zero forpurely symmetric profiles).Four representative examples of the tempo-ral behavior of the first and third velocity mo-ments are presented in Fig. 2. The associated un-certainties (not included in the plot) are ∼ − , and ∼ s − , respec-tively. In the case of the dwarf star HD 37042, the h v i values are fairly constant and close tozero. The maximum dispersion in velocity forthis star is ∼ − , of the order of the ac-curacy associated with the instrumental settingused for the FIES@NOT observations (indicatedas dashed horizontal lines). All the other starsin Figure 2 show h v i variations above this sig-nificance level, with maximum peak-to-peak am-plitudes reaching ∼ − in some of thecases. The minimum variations are found forHD 214680 (the other luminosity class V objectconsidered in this study), HD 191243 (B5 Ib) andHD 190603 (B1.5 Ia + ), with maximum amplitudesslightly larger than the significance level. Curi-ously, the star HD 190603, expected to be an ex-treme object from its spectral classification, is oneof the cases with smaller variations. We indicatein Table 1 (columns 10 and 11) the peak-to-peakamplitude of the first and third moment variationsmeasured for each of the considered targets.
5. The “macroturbulence”-LPV connec-tion
We investigate the possible connection betweenmacroturbulent broadening and LPVs in our sam-ple of stars. Such a connection should appearin case macroturbulent broadening is producedby any time-dependent physical phenomena (e.g.non-radial oscillations, see Aerts et al. 2009). InFig. 3 (upper pannels), we plot the average size of5he macroturbulent broadening, h Θ RT i , versus thepeak-to-peak amplitude of h v i and h v i variationsfor each of the stars studied. Results from the fourstars observed in both campaigns are conectedwith solid lines. A clear positive correlation ispresent in both cases. To our knowledge, this isthe first clear observational evidence for a connec-tion between extra broadening and LPVs in B andlate O Sgs. Particularly remarkable is the Θ RT - h v i correlation: the larger the extra broadening,the more asymmetric line profiles can be found inthe time series. Note that this does not mean thatlines with a significant macroturbulence contribu-tion are always asymmetric since h v i oscillatesbetween positive and negative values over time.We also present in Fig. 3 (bottom pannels) sim-ilar plots with data from Table 1 in Aerts et al.(2009), based on simulations of line profiles broad-ened by rotation and by hundreds of low amplitudenon-radial gravity mode pulsations. These simu-lations consider various combinations of the incli-nation angle ( i ), the projected rotational velocity( v sin i ), and the intrinsic amplitude of the modesdenoted by a (see Aerts et al. 2009, for a defini-tion). We present the maximum values of v mac obtained for each set of simulated profiles. Similartrends for v mac as in Fig. 3 are found using aver-age or minimum values for the macroturbulenceparameter, but with a different (smaller) scale inthe y-axis.The simulations lead to clear trends which arecompatible with spectroscopic observations.
6. Is macroturbulent broadening in OB-Sgs caused by pulsations?
Non-radial oscillations have been often sug-gested as the origin of LPVs and photosperic linesin OB Sgs, as well as the driver of large scale windstructures (see references in Sect. 4); however, afirm confirmation (by means of a rigorous seis-mic analysis) has not been achieved yet. From atheoretical point of view, g-mode oscillations werenot initially expected in B Sgs because the radia-tive damping in the core was suspected as beingtoo strong. Saio et al. (2006) detected simultane-ous p- and g-modes in HD 163899 (B2 Ib/II) usingdata from the MOST satellite. These authors alsocomputed new models showing that g-modes canbe excited in massive post-main sequence stars, as Fig. 3.— (Top) Empirical relations betweenthe average size of the macroturbulent broad-ening ( h Θ RT i ) and the peak-to-peak amplitudeof the first and third moments of the line-profile.(Bottom) Similar relations are found fromthe simulations by Aerts et al. (2009).the g-modes are reflected at the convective zoneassociated with the H-burning shell. Lefever et al.(2007) presented observational evidence of g-modeinstabilities in a sample of photometrically vari-able B Sgs from the location of the stars in the(log g , log T eff )-diagram.These results, along with our observational con-firmation of a tight connection between macrotur-bulence and parameters describing observed LPVsrender stellar oscillations the most probable phys-ical origin of macroturbulent broadening in B Sgs;however, it is too premature to consider them asthe only physical phenomenon to explain the un-known broadening.Recent studies of high resolution spectra of OBstars indicate that the appearance of a signifi-cant macroturbulent broadening contribution isnot only limited to B Sgs but occurs also in Ostars of all luminosity classes (Sim´on-D´ıaz et al.2010; Markova et al. 2010) as well as in main-sequence B-type pulsators (e.g. Morel et al. 2006).Moreover, with the recent discovery of a strange-6ode oscillation in the B6Ia supergiant HD 46005(Aerts et al. 2010b) and of stochastic oscillationsin the O8V star HD 46149 (Degroote et al. 2010)and the β Cephei star V1449 Aql (Belkacem et al.2009) from high-precision space photometry, ob-servational evidence of atmosphere phenomenadue to oscillations, which must ◦ REFERENCES
Aerts, C., Puls, J., Godart, M., & Dupret, M.-A.2009, A&A, 508, 409Aerts, C., Christensen-Dalsgaard, J., & Kurtz,D. W. 2010a, Asteroseismology (Springer)Aerts, C., Lefever, K., Baglin, A., Degroote,P., Oreiro, R., Vuˇckovi´c, M., Smolders, K.,Acke, B., Verhoelst, T., Desmet, M., Godart,M., Noels, A., Dupret, M.-A., Auvergne, M.,Baudin, F., Catala, C., Michel, E., Samadi, R.2010b, A&A, 513, L11Belkacem, K., et al. 2009, Science, 324, 1540Conti, P. S., & Ebbets, D. 1977, ApJ, 213, 438 Degroote, P., Briquet, M., Auvergne, M., Sim´on-D´ıaz, S., Aerts, C., Noels, A., Rainer, M.,Hareter, M., Poretti, E., Mahy, L., Oreiro, R.,Vuckovic, M., Smolders, K., Baglin, A., Baudin,F., Catala, C., Michel, E., Samadi, R. 2010,A&A, in press (arXiv:1006.3139)Dufton, P. L., Ryans, R. S. I., Sim´on-D´ıaz, S.,Trundle, C., & Lennon, D. J. 2006, A&A, 451,603Ebbets D. 1982, ApJS, 48, 399Fraser, M., Dufton, P. L., Hunter, I., & Ryans,R. S. I. 2010, MNRAS, 404, 1306Fullerton A. W., Gies D. R., & Bolton C. T. 1996,ApJS, 103, 475Gray, D. F. 1976, The Observations and Analy-sis of Stellar Photospheres (1st ed.; New York:Wiley)Howarth I. D., et al. 1993, ApJ, 417, 338Howarth, I. D., Siebert, K. W., Hussain, G. A. J.,& Prinja, R. K. 1997, MNRAS, 284, 265Kaufer A., Stahl O., Prinja R. K., & Witherick D.2006, A&A, 447, 325Lefever, K., Puls, J., & Aerts, C. 2007, A&A, 463,1093Lucy, L. B. 1976, ApJ, 206, 499Markova N., Puls J., Scuderi S., & Markov H.2005, A&A, 440, 1133Markova N., Prinja R. K., Markov H., Kolka I.,Morrison N., Percy J., & Adelman S. 2008,A&A, 487, 211Markova, N., et al. 2010, in preparationMorel, T., Marchenko, S. V., Pati, A. K., Kup-puswamy, K., Carini, M. T., Wood, E., & Zim-merman, R. 2004, MNRAS, 351, 552Morel, T., Butler, K., Aerts, C., Neiner, C., &Briquet, M. 2006, A&A, 457, 651Prinja R. K., Fullerton A. W., & Crowther P. A.1996, A&A, 311, 2647rinja, R. K., Rivinius, T., Stahl, O., Kaufer, A.,Foing, B. H., Cami, J., & Orlando, S. 2004,A&A, 418, 727Prinja R. K., Markova N., Scuderi S., & MarkovH. 2006, A&A, 457, 987Ryans, R. S. I., Dufton, P. L., Rolleston, W. R. J.,Lennon, D. J., Keenan, F. P., Smoker, J. V., &Lambert, D. L. 2002, MNRAS, 336, 577Saio, H., et al. 2006, ApJ, 650, 1111Sim´on-D´ıaz, S., & Herrero, A. 2007, A&A, 468,1063Sim´on-D´ıaz, S., et al. 2010, in preparationSlettebak, A. 1956, ApJ, 124, 173Zima W. 2008, CoAst, 155, 17