Photometric detection of internal gravity waves in upper main-sequence stars. II. Combined TESS photometry and high-resolution spectroscopy
D. M. Bowman, S. Burssens, S. Simón-Díaz, P. V. F. Edelmann, T. M. Rogers, L. Horst, F. K. Roepke, C. Aerts
AAstronomy & Astrophysics manuscript no. TESS_IGWs_arxiv_2ver c (cid:13)
ESO 2020June 11, 2020
Photometric detection of internal gravity waves in uppermain-sequence stars
II. Combined TESS photometry and high-resolution spectroscopy
D. M. Bowman , S. Burssens , S. Simón-Díaz , , P. V. F. Edelmann , T. M. Rogers , , L. Horst , F. K. Röpke , , andC. Aerts , , Institute of Astronomy, KU Leuven, Celestijnenlaan 200D, B-3001 Leuven, Belgiume-mail: [email protected] Instituto de Astrofísica de Canarias, E-38200 La Laguna, Tenerife, Spain Departamento de Astrofísica, Universidad de La Laguna, E-38205 La Laguna, Tenerife, Spain X Computational Physics (XCP) Division and Center for Theoretical Astrophysics (CTA), Los Alamos National Laboratory, LosAlamos, NM 87545, USA School of Mathematics, Statistics and Physics, Newcastle University, Newcastle-upon-Tyne NE1 7RU, UK Planetary Science Institute, Tucson, AZ 85721, USA Heidelberger Institut für Theoretische Studien, Schloss-Wolfsbrunnenweg 35, 69118 Heidelberg, Germany Zentrum für Astronomie der Universität Heidelberg, Institut für theoretische Astrophysik, Philosophenweg 12, 69120 Heidelberg,Germany Department of Astrophysics, IMAPP, Radboud University Nijmegen, NL-6500 GL Nijmegen, The Netherlands Max Planck Institute for Astronomy, Koenigstuhl 17, D-69117 Heidelberg, GermanyReceived 21 April 2020 / accepted 1 June 2020 ABSTRACT
Context.
Massive stars are predicted to excite internal gravity waves (IGWs) by turbulent core convection and from turbulent pressurefluctuations in their near-surface layers. These IGWs are extremely e ffi cient at transporting angular momentum and chemical specieswithin stellar interiors, but they remain largely unconstrained observationally. Aims.
We aim to characterise the photometric detection of IGWs across a large number of O and early-B stars in the Hertzsprung–Russell diagram, and explain the ubiquitous detection of stochastic variability in the photospheres of massive stars.
Methods.
We combined high-precision time-series photometry from the NASA
Transiting Exoplanet Survey Satellite with high-resolution ground-based spectroscopy of 70 stars with spectral types O and B to probe the relationship between the photometricsignatures of IGWs and parameters such as spectroscopic mass, luminosity, and macroturbulence.
Results.
A relationship is found between the location of a star in the spectroscopic Hertzsprung–Russell diagram and the amplitudesand frequencies of stochastic photometric variability in the light curves of massive stars. Furthermore, the properties of the stochasticvariability are statistically correlated with macroturbulent velocity broadening in the spectral lines of massive stars.
Conclusions.
The common ensemble morphology for the stochastic low-frequency variability detected in space photometry and itsrelationship to macroturbulence is strong evidence for IGWs in massive stars, since these types of waves are unique in providing thedominant tangential velocity field required to explain the observed spectroscopy.
Key words. asteroseismology – stars: early-type – stars: oscillations – stars: evolution – stars: rotation – stars: fundamental parame-ters – stars: massive
1. Introduction
The advent of time-series photometry from space telescopes inthe past decade has revealed a wealth of information for starsborn with a convective core that was not previously attainableby ground-based telescopes. In particular, asteroseismology –the study of stellar structure and evolution by means of forwardmodelling stellar pulsation frequencies – has greatly benefittedfrom the long-term, high-precision, and continuous light curvesassembled by space missions such as CoRoT (Auvergne et al.2009), Kepler / K2 (Borucki et al. 2010; Koch et al. 2010; Howellet al. 2014), and more recently the
Transiting Exoplanet SurveySatellite (TESS; Ricker et al. 2015) mission. This is because var-ious phenomena exist in massive stars that produce variabilitywith periods between minutes and decades, which in turn makes asteroseismology of pulsation mode frequencies in massive starschallenging when using ground-based telescopes (Aerts et al.2010). However, space telescopes have enabled massive star as-teroseismology by revealing a diverse range of di ff erent vari-ability mechanisms across a wide range of masses, evolutionarystages, and metallicity environments (Buysschaert et al. 2015;Ramiaramanantsoa et al. 2018; Bowman et al. 2019b; Pedersenet al. 2019; Handler et al. 2019; Burssens et al. 2020).Despite the challenges in obtaining the necessary data suit-able for asteroseismology, early studies of massive stars revealedthat stellar structure and evolution theory is discrepant withobservations in terms of angular momentum transport (Aertset al. 2019a) and chemical mixing (Aerts 2020). In particu-lar, the convective core masses of massive main-sequence stars Article number, page 1 of 17 a r X i v : . [ a s t r o - ph . S R ] J un & A proofs: manuscript no. TESS_IGWs_arxiv_2ver inferred through asteroseismic modelling are larger than pre-dicted by current theoretical models (Handler et al. 2006; Bri-quet et al. 2007, 2011; Daszy´nska-Daszkiewicz et al. 2013; Aertset al. 2019b). Similarly, asteroseismology of main-sequenceintermediate-mass stars also demonstrates the need for largercore masses than predicted by evolutionary models (Moravvejiet al. 2015, 2016; Schmid & Aerts 2016; Buysschaert et al.2018; Szewczuk & Daszy´nska-Daszkiewicz 2018; Mombarget al. 2019). This core-mass discrepancy for massive stars is alsoevident in the detailed analysis of eclipsing double-lined spec-troscopic binary (SB2) systems, in which a larger amount of ex-tra mixing in the near-core region is needed to explain the loca-tion of binary systems in the Hertzsprung–Russell (HR) diagram(Guinan et al. 2000; Claret & Torres 2016, 2019; Johnston et al.2019; Tkachenko et al. 2020).The evolution of a star born with a convective core involvesthe complex interaction of di ff erent physical processes, whichare largely unconstrained in stellar models and currently con-trolled by free parameters, such as the amount and shape of near-core mixing. The large uncertainties from these unknowns arecompounded by uncertainties associated with interacting multi-ple systems (Sana et al. 2012; Langer 2012), rotation (Ekströmet al. 2012; Chie ffi & Limongi 2013), metallicity (Georgy et al.2013; Groh et al. 2019), and magnetic fields (Keszthelyi et al.2019, 2020). Together these phenomena impact how a massivestar is formed, how it evolves, and ultimately determine its fatebeyond the main sequence. Yet the interior rotation, mixing,and angular momentum transport mechanism for main sequencemassive stars remain largely unconstrained (Aerts et al. 2019a).There are several non-mutually exclusive variability mecha-nisms operating in massive stars. For early-type stars, stochasticlow-frequency variability at the surface is commonly observedin photometry (see e.g. Balona 1992; Buysschaert et al. 2015;Bowman et al. 2019b). On the other hand, spectroscopic vari-ability in the form of spectral line profile variations and vari-able macroturbulence are also typical for massive stars (see e.g.Howarth et al. 1997; Simón-Díaz & Herrero 2014). In their de-tailed study of how macroturbulence is related to stellar parame-ters, Simón-Díaz et al. (2017) make the following two importantconclusions: (i) between early- and late-B main sequence stars,there is diverse behaviour in terms of line-broadening mecha-nisms, which is likely attributed to the diverse pulsational be-haviour in this mass range; and (ii) main-sequence O stars andB supergiants have macroturbulence as their dominant broad-ening mechanism, with most stars sharing a common broaden-ing profile. Previously, macroturbulence has been linked to non-radial pulsations (e.g. Lucy 1976), which are excited by eitherthe opacity mechanism and / or turbulent pressure fluctuations instellar envelopes (Aerts et al. 2009; Grassitelli et al. 2015).The commonly known form of pulsations in massive starsare coherent pulsation modes (i.e. standing waves) triggered byan opacity mechanism operating in the Z-bump associated withiron-peak elements (Dziembowski & Pamyatnykh 1993; Dziem-bowski et al. 1993; Miglio et al. 2007; Szewczuk & Daszy´nska-Daszkiewicz 2017; Godart et al. 2017). Such a heat-engine isable to excite low-radial order pressure (p) modes in stars moremassive than approximately 8 M (cid:12) and high-radial order gravity(g) modes in stars more massive than approximately 3 M (cid:12) . SeeSzewczuk & Daszy´nska-Daszkiewicz (2017) for calculations ofinstability regions of p- and g-mode pulsations in early-type stars Conversely, microturbulence has a length scale much shorter than themean free path of a photon, and is not to be confused with macroturbu-lence. See Gray (2005) for a detailed discussion. including rotation. However, new observations reveal a signifi-cant fraction of pulsating massive stars outside of predicted in-stability regions (Burssens et al. 2020), hence the overall pictureof variability is far from complete.Additionally, quasi-periodic variability caused by surfaceand / or wind inhomogeneities combined with rotation (Mo ff atet al. 2008; David-Uraz et al. 2017; Aerts et al. 2018; Simón-Díaz et al. 2018; Krtiˇcka & Feldmeier 2018), and small-scalevariability triggered by thin subsurface convection zones asso-ciated with local opacity enhancements (Cantiello et al. 2009;Cantiello & Braithwaite 2011, 2019; Lecoanet & Quataert 2013)can play a role in some massive stars. Subsurface convection ispredicted to be more e ffi cient towards later evolutionary phases,hence microturbulence is predicted to increase with increasingluminosity and decreasing surface gravity, the latter of whichhaving been confirmed by observations (Cantiello et al. 2009;Tkachenko et al. 2020). Despite the subsurface convection zoneassociated with the iron bump being absent in main-sequence Bstars within the mass range 3 ≤ M ≤ (cid:12) (Cantiello et al.2009; Cantiello & Braithwaite 2019), stochastic variability hasbeen detected in Slowly Pulsating B (SPB) stars observed by theCoRoT and Kepler space missions (Bowman et al. 2019a; Ped-ersen 2020).Moreover, subsurface convection zones and stellar windsdo not directly provide the large-scale tangential velocity fieldneeded to explain macroturbulent broadening in massive stars(Gray 2005; Simón-Díaz et al. 2010; Simón-Díaz & Herrero2014; Simón-Díaz et al. 2017). A combined radial-tangentialbroadening profile is typically adopted to reproduce observedspectroscopic line profiles (see e.g. Simón-Díaz & Herrero2014). Such a profile from combining rotational and pulsationalbroadening components is motivated by the fact that non-radialgravity-mode pulsations produce predominantly horizontal ve-locities in the line-forming region. Rotational and microturbulentbroadening alone cannot accurately reproduce (the variability of)spectral lines in hot stars (Gray 2005; Aerts et al. 2009).Recent 2D and 3D hydrodynamical simulations demonstratethat internal gravity waves (IGWs) generated at the interface ofthe convective core and radiative envelope are also expected formassive stars (Rogers et al. 2013; Edelmann et al. 2019; Horstet al. 2020). These IGWs are e ffi cient at transporting angularmomentum and chemical species within stellar interiors (Rogers2015; Rogers & McElwaine 2017; Edelmann et al. 2019). Thecollective power of an entire spectrum of IGWs excited by tur-bulent core convection is predicted to produce stochastic low-frequency variability and a large tangential velocity field near thestellar surface (Rogers et al. 2013; Edelmann et al. 2019; Horstet al. 2020). Hence an ensemble of IGWs provides the requiredvelocity field to explain macroturbulent velocity broadening inhot stars (Aerts & Rogers 2015). Furthermore, stochastic low-frequency variability was recently detected in hundreds of mas-sive stars between 3 and 50 M (cid:12) by Bowman et al. (2019b) andinferred to be caused by IGWs because of its similar morphologyto that predicted by hydrodynamical simulations.Here we combine high-precision TESS photometry andhigh-resolution ground-based spectroscopy to probe the relation-ship between a star’s variability, location in the HR diagram, andits measured macroturbulent broadening. In Section 2 we discussour sample selection criteria and methodology. In Section 3, wetest how the morphology of stochastic low-frequency variabil-ity depends on the parameters of a star. Finally, we discuss ourresults in Section 4, and conclude in Section 5. Article number, page 2 of 17. M. Bowman et al.: TESS observations of IGWs in massive stars
2. Method
We extend the methodology developed by Bowman et al.(2019a) of analysing time-series photometry to a much largersample of 70 massive stars. Our sample is comprised of early-type stars with spectral types O and B which have high-precisionTESS photometry and fundamental parameters available fromhigh-resolution spectroscopy (Burssens et al. 2020).
As demonstrated by Blomme et al. (2011) and Bowman et al.(2019a), the variability in massive stars spans a broad range infrequency, and is significant above the instrumental white noiselevel at frequencies as high as 100 d − for some stars. Thereforethe TESS 2-min cadence is essential to avoid amplitude suppres-sion of high-frequency variability introduced by long-cadencetime series photometry (Murphy 2014; Bowman 2017). We ex-clude stars for which the 2-min TESS light curves exhibit strongsignatures of contamination or instrumental systematics as iden-tified by Burssens et al. (2020). We also exclude eclipsing binary(EB) systems as these may contain a significant ( (cid:38) ff ective temperature, T e ff ,spectroscopic luminosity, log ( L / L (cid:12) ), projected surface rota-tional velocity, v sin i , and macroturbulent broadening, v macro .These were derived from high-resolution spectra assembled bythe IACOB (Simón-Díaz et al. 2011, 2015) and OWN (Barbáet al. 2010, 2014, 2017) surveys. Within the IACOB spectro-scopic database, spectra of northern OB stars have been obtainedwith the HERMES spectrograph ( R (cid:39)
85 000) on the 1.2-m Mer-cator telescope (Raskin et al. 2011), and the FIES spectrograph( R (cid:39)
46 000) mounted on the 2.6-m NOT telescope (Telting et al.2014), on La Palma. Whereas spectra of southern O stars assem-bled as part of the OWN survey were obtained with the FEROSspectrograph ( R (cid:39)
48 000) mounted on the ESO / MPG 2.2-mtelescope at La Silla (Kaufer et al. 1997, 1999). The extraction ofspectroscopic parameters followed the methodologies outlinedby Simón-Díaz & Herrero (2014); Holgado et al. (2018); Castroet al. (2018), and we refer the reader to Burssens et al. (2020) forfurther details.Our final sample consists of 70 early-type stars in the south-ern ecliptic hemisphere (TESS sectors 1–13). The spectral typesand fundamental parameters determined from spectroscopy ob-tained from Burssens et al. (2020) are provided in Table A.1.Within our sample, typical uncertainties for log (T e ff ) range be-tween 0.03 and 0.05 dex, and for log ( L / L (cid:12) ) range between0.15 and 0.20 dex (see Simón-Díaz et al. 2017; Holgado et al.2018, 2020). We obtain the 2-min TESS light curves from the MikulskiArchive for Space Telescopes (MAST ). We use the pre-searchdata conditioning simple aperture photometry (PDCSAP) time TESS full frame image (FFI) data have a cadence of 30 min whichcauses significant amplitude suppression near integer multiples of theFFI sampling frequency, hence prevents an accurate determination ofthe high-frequency component of IGWs above ν (cid:38)
10 d − . MAST website: https://archive.stsci.edu/ series, and refer the reader to Jenkins et al. (2016) for furtherdetails of the TESS data pipeline. We perform checks to vali-date the chosen aperture mask and contamination, convert thelight curves into units of stellar magnitudes and perform addi-tional detrending in the form of a low-order polynomial for eachsector. From high-precision 2-min TESS light curves, we calcu-late amplitude spectra by means of a discrete Fourier transform(DFT; Deeming 1975; Kurtz 1985).In addition to requiring high-cadence time-series photom-etry, it is also necessary to remove high-amplitude pulsationmodes to detect and characterise the underlying stochastic low-frequency variability in early-type stars (Degroote et al. 2009,2012; Bowman et al. 2019a). Importantly, we do not discrimi-nate on the mechanism by which these coherent pulsation modesare excited: the opacity mechanism (Szewczuk & Daszy´nska-Daszkiewicz 2017) and / or by core convection (Edelmann et al.2019; Horst et al. 2020). It is known that the excitation of pul-sation modes driven by the opacity mechanism is very sensi-tive to the rotation and metallicity of a star, and the opacityof stellar models (Miglio et al. 2007; Szewczuk & Daszy´nska-Daszkiewicz 2017; Burssens et al. 2020). On the other hand,core convection is predicted to excite a broad spectrum of IGWs,including resonant pressure and gravity eigenmodes (Edelmannet al. 2019; Lecoanet et al. 2019; Horst et al. 2020).We use the standard approach in asteroseismology of early-type stars with coherent pulsations and perform iterative pre-whitening to identify significant frequencies in the light curvesand remove them to produce a residual amplitude spectrum(see e.g. Degroote et al. 2009; Pápics et al. 2012; Van Reethet al. 2015; Bowman 2017). In this iterative process, all high-amplitude coherent pulsation modes are subtracted from the ob-served light curve using the cosinusoidal model: ∆ m = A cos(2 πν ( t − t ) + φ ) , (1)where A is the amplitude, ν is the frequency, φ is the phase, t is the time with respect to a zero-point t . Following Bowmanet al. (2019a,b), we employ the standard amplitude significancecriterion in iterative pre-whitening, which defines significant fre-quencies having an amplitude signal-to-noise ratio (S / N) largerthan four (Breger et al. 1993). We note that not all stars have sig-nificant frequencies following this definition (see Blomme et al.2011; Bowman et al. 2019a).
After iteratively prewhitening any significant pulsation modefrequencies and / or frequencies that may represent harmonics ofthe rotation frequency of each star, we use the method devel-oped by Bowman et al. (2019a) and Bowman et al. (2019b) tocharacterise the stochastic low-frequency variability in its resid-ual amplitude spectrum. We utilise a Bayesian Markov chainMonte Carlo (MCMC) framework with the Python code emcee (Foreman-Mackey et al. 2013) to fit the residual amplitude spec-trum of a star with the function: α ( ν ) = α + (cid:16) νν char (cid:17) γ + C w , (2)where α represents the amplitude at a frequency of zero, γ is the logarithmic amplitude gradient, ν char is the characteris-tic frequency (i.e. the inverse of the characteristic timescale, τ , of stochastic variability present in the light curve such that Article number, page 3 of 17 & A proofs: manuscript no. TESS_IGWs_arxiv_2ver ν char = (2 πτ ) − ), and C w is a frequency-independent (i.e. white)noise term (Blomme et al. 2011; Bowman et al. 2019b).Similarly to Bowman et al. (2019b), we use non-informative(flat) priors and 128 parameter chains, and we fit the (residual)amplitude spectrum up to the TESS Nyquist frequency (i.e. 0 . ≤ ν ≤ . − ) using the model given in Eq. (2) for each starin our sample. At each iteration, a parameter chain is used toconstruct a model that is subject to a log-likelihood evaluation:ln L ∝ − (cid:88) i (cid:32) y i − M ( Θ i ) σ i (cid:33) , (3)where ln L is the log-likelihood, y i are the data, σ i are their un-certainties, and M Θ is the model with parameters Θ . After burn-ing the first few hundred iterations, convergence is confirmedfor the subsequent ∼ γ , and at what frequencythe profile turns over, ν char , are important parameters to distin-guish the source of the stochastic variability (Bowman et al.2019a). Hydrodynamical simulations predict that an ensembleof IGWs generated by core convection produces an amplitudespectrum with 0 . ≤ γ ≤ − . Fur-thermore, the exact values of γ and ν char depend on the massand radius of the host star (Rogers et al. 2013; Edelmann et al.2019; Horst et al. 2020). On the other hand, other sources oflow-frequency variability such as small-scale waves generatedby subsurface convection produce a steeper frequency spectrum( γ ≥ .
25; see Couston et al. 2018).At very low frequencies, such as below 0.1 d − , the prob-ing power of time series photometry is limited by the length ofobservations. Furthermore, instrumental systematics present inthe light curve may dominate in the amplitude spectrum below0.1 d − since they correspond to variability with periods of or-der the length of the time series. In the case of TESS data, thiscorresponds to approximately 12 d (i.e. half of a single sector),which is why Bowman et al. (2019b) only characterised variabil-ity above 0.1 d − . In all massive stars studied by Bowman et al.(2019b) and those in our current TESS sample, the measured ν char parameter is significant at frequencies higher than 0.2 d − .
3. Results
The amplitude spectrum of each massive star in our sample wasfit using Eq. (2). For all 70 stars in our sample, we provide theresultant fit parameters in Table A.2 and their 1 σ statistical un-certainties as determined from the converged parameter chainsfrom our MCMC framework (Bowman et al. 2019a). The fittedlogarithmic (residual) amplitude spectra for all stars are providedas figures in Appendix B, so as to demonstrate the broad range offrequencies and amplitudes within our sample of massive stars.We provide three examples of massive stars with fittedstochastic low-frequency variability in Fig. 1 that were pre-viously observed by CoRoT. These stars demonstrate that thebroad frequency excess occurs in both the TESS data of mas-sive stars but also in the completely independent CoRoT ob-servations from approximately a decade ago. In Fig. 1, orangelines denote the original TESS amplitude spectra (before itera-tive pre-whitening), and black lines denote residual amplitude Fig. 1.
Fitted amplitude spectra calculated using TESS light curves forthe O4 V((f)) star HD 46223 (top panel), the O5 V((f)) star HD 46150(middle panel), the B2.5 V star HD 48977 (bottom panel), which werepreviously observed by the CoRoT mission and concluded to exhibitIGWs (Bowman et al. 2019a). We note that of these three examples,only HD 48977 underwent iterative pre-whitening to produce a residualamplitude spectrum because it exhibits significant p-mode frequenciesabove 10 d − . spectra after iterative pre-whitening has removed S / N ≥ Article number, page 4 of 17. M. Bowman et al.: TESS observations of IGWs in massive stars (e.g. HD 48977; Thoul et al. 2013). The stochastic variabilityof the O dwarfs is significant over the 1-sector TESS whitenoise level of order a few µ mag at frequencies higher than ∼
30 d − ( ∼ µ Hz). Whereas in the B dwarfs, the stochasticlow-frequency variability in the residual amplitude spectrum af-ter coherent pulsation modes have been removed is significantup to ∼
10 d − ( ∼ µ Hz). Such a frequency dependence of thestochastic variability on the spectral type is common throughoutour sample, as demonstrated by the amplitude spectra providedin Appendix B.Moreover, we emphasise the ubiquitous detection of stochas-tic photometric variability in our sample of 70 O and early-Bstars. In more massive O stars, this is the dominant form ofvariability. However, in a few early-B stars (e.g. HD 34816 andHD 46328) a series of harmonics are present in their amplitudespectra, which are indicative of rotational modulation and / or bi-narity. Also, some early-B stars in our sample show clear signa-tures of low-frequency coherent gravity-mode pulsations (e.g.HD 35912 and HD 57539), or high-frequency pressure-modepulsations (e.g. HD 37209 and HD 37481). We refer the readerto Burssens et al. (2020) for the detailed frequency analysis ofthe coherent pulsators in the sample. To more accurately investigate the observed photometric vari-ability, we place our stars in the spectroscopic HR (sHR) dia-gram in Fig. 2. This common approach when studying massivestars involves using the e ff ective stellar luminosity on the ordi-nate axis defined as: L : = T ff / g (Langer & Kudritzki 2014).This has the significant advantage of avoiding issues pertainingto uncertain distances and reddening propagating into bolomet-ric luminosity calculations. Each star is represented by a circlein the spectroscopic HR diagrams in the bottom-left and bottom-right panels of Fig. 2, which have been colour-coded by the fitparameters α and ν char , respectively, and have a symbol sizethat is proportional to the fit parameter α . To illustrate the dis-tribution of our sample in terms of spectroscopic mass and evo-lutionary stage, we plot the non-rotating evolutionary tracks forinitial masses between 4 and 80 M (cid:12) calculated by Burssens et al.(2020) as grey lines in Fig. 2, and a indicative ZAMS line as thedashed-grey line.In the top row of Fig. 2, we also plot the pairwise relation-ship between the individual fit parameters α , ν char and γ as filledcircles, which have been colour-coded by the spectroscopic lu-minosity of each star. For the most massive stars within oursample ( M >
20 M (cid:12) ), it is clear that the more luminous starshave larger α values, such that they have larger amplitudes intheir stochastic photometric variability. Furthermore, as best ev-idenced by the spectroscopic HR diagrams in the bottom rowof Fig. 2, more massive and more evolved stars not only havelarger amplitudes in their stochastic photometric variability, butsmaller ν char values, such that their dominant photometric vari-ability is constrained to longer periods. This is a characteristicsignature of IGWs probing stellar evolution: more evolved starshave larger radii, hence IGWs have longer periods.Inferring relationships among fit parameters and the distribu-tion in the HR diagram for stars with masses between approxi-mately 5 (cid:46) M (cid:46)
20 M (cid:12) is less clear. This is expected because thestars in this mass regime have diverse causes for their variabil-ity, such as rotational modulation and coherent pulsation modesexcited by the opacity mechanism ( β Cephei and Slowly Pulsat-ing B stars; Aerts et al. 2010) superimposed on their stochasticlow-frequency variability. In the case of short-length time series and a large number of independent pulsation modes, this makesthe extraction of the morphology of the stochastic low-frequencyvariability subject to somewhat larger uncertainties (Bowmanet al. 2019a). Such a photometric result is supported by the di-verse range in spectroscopic variability and broadening mecha-nisms found in main-sequence B stars (see e.g. Simón-Díaz et al.2017).In their study of the initial detection of stochastic variabil-ity in three O stars observed by CoRoT, which included theO dwarfs HD 46223 and HD 46150 in this work, Blomme et al.(2011) discussed an apparent dichotomy in the photometric vari-ability of massive stars. More specifically, early-O stars typicallyhave stochastic low-frequency variability, and late-O and early-B stars typically have coherent pulsation modes. The transitionbetween these takes place at spectral type of around O8, as alsoillustrated by the star HD 46149 (Degroote et al. 2010). Thisspectral type approximately corresponds to a mass of approxi-mately 20 M (cid:12) . A similar dichotomy of stars later than O8 havingvariability caused by coherent pulsation modes has also found inspectroscopic variability studies of early-type stars (e.g. Simón-Díaz et al. 2017).Our much larger sample of massive stars compared to theprevious studies by Blomme et al. (2011) and Bowman et al.(2019a), clearly supports the importance of stochastic photo-metric variability in massive stars. Furthermore, our results arethe first evidence that such photometric variability is increas-ingly important for more massive stars, and is related to themass and evolutionary stage of the star, as evidenced by Fig. 2.The parameter space in the spectroscopic HR diagram occupiedby the O stars is also where macroturbulence is the dominantbroadening mechanism (Simón-Díaz et al. 2017). We explorethe connection between macroturbulence and stochastic photo-metric variability in Section 3.2.
In addition to the location in the spectroscopic HR diagram, weare able to probe the relationship between stochastic photomet-ric variability measured in TESS photometry and macroturbu-lence measured in spectroscopy for a large number of massivestars. Among main-sequence B stars, it has been noted by severalstudies that non-radial g-mode pulsations are a plausible phys-ical mechanism to explain macroturbulence given their domi-nant horizontal velocities (Aerts et al. 2009; Aerts & Rogers2015; Simón-Díaz et al. 2010; Simón-Díaz & Herrero 2014;Simón-Díaz et al. 2017). Using spectroscopic observations anddetailed simulations of broadening and variability in spectral lineprofiles of main-sequence B stars, Aerts et al. (2009) demon-strated that macroturbulence is well explained by non-radial co-herent g-mode pulsations. Aerts & Rogers (2015) extended thisstudy to demonstrate that the collective power of an ensembleof IGWs excited by core convection is also a plausible mech-anism for macroturbulent broadening. Gravity waves (coherentand damped) have a ratio in their horizontal to vertical velocitiesthat ranges approximately between 10 and 100 in observations(De Cat & Aerts 2002; Aerts & De Cat 2003), with similar val-ues predicted near the line-forming region by hydrodynamicalsimulations of IGWs excited by core convection (Rogers et al.2013; Aerts & Rogers 2015; Horst et al. 2020).Given the scarce number of stars more massive than some20 M (cid:12) with coherent g-mode pulsations, Grassitelli et al. (2015)demonstrated how the predicted amplitude of turbulent pres-sure fluctuations originating from the iron subsurface convec-tion zone correlate with observed macroturbulence in massive
Article number, page 5 of 17 & A proofs: manuscript no. TESS_IGWs_arxiv_2ver
Fig. 2.
Top row: pairwise relationship between α , ν char and γ (cf. Eq. 2) best-fit parameters for our sample of OB stars. Bottom row: location ofstars in the spectroscopic HR diagram as filled circles that are colour-coded by the best-fit parameters α (left) and ν char (right), and have a symbolsize proportional to the fit parameter α . Evolutionary tracks (in units of M (cid:12) ) from Burssens et al. (2020) are shown as solid grey lines and thedashed grey line represents the ZAMS. A typical spectroscopic error bar for our sample is shown in the top-left corner. O stars. Therefore, IGWs launched by turbulent pressure inthe envelopes of massive O stars has also been inferred to bea plausible mechanism to explain macroturbulence (Grassitelliet al. 2015). However, surface mechanisms, such as subsurfaceconvection, cause small-scale and time-independent microtur-bulence in massive stars (Cantiello et al. 2009), as they can-not reproduce the large-scale required tangential velocity fieldthroughout stellar interiors including the line-forming region(Aerts et al. 2009; Aerts & Rogers 2015).In Fig. 3 we test the correlation of stochastic photometricvariability and macroturbulent broadening measured using high-resolution spectroscopy using the 59 out of 70 stars in our sam-ple with reliable estimates of macroturbulence. We find a clearcorrelation between the amplitude of the stochastic photomet-ric variability ( α ) and v macro within our sample. We providethe Spearman’s rank correlation coe ffi cient, R , and the corre-sponding p -value (obtained from a t -test) in Fig. 3. A linearregression reveals a strong correlation ( R = . p < . R = . p < .
01) between the measured ν char in thestochastic photometric variability and the spectroscopic macro-turbulent broadening. This is expected for IGWs as they aresensitive to the mass and radius of a star. Finally, our resultsdemonstrate that macroturbulence has no significant correlation( R = − . p = .
24) with the steepness of the observed fre-quency spectrum.The observed relationship between v macro and the fit param-eters of the stochastic photometric variability shown in Fig. 3builds on the previous theoretical work by Aerts & Rogers(2015) and Grassitelli et al. (2015), and spectroscopy by Simón-Díaz et al. (2017). It demonstrates the importance of IGWs inmassive stars. Furthermore, our study provides photometric evi-dence that IGWs are increasingly more important for stars withlarger spectroscopic masses and luminosities. Non-radial pulsa-tions (coherent modes and / or travelling waves — i.e. IGWs) areunique in their ability to explain the required large-scale tan-gential velocity field near the stellar surface, given such strongcorrelations with macroturbulence measured spectroscopically. Article number, page 6 of 17. M. Bowman et al.: TESS observations of IGWs in massive stars
Fig. 3.
Relationship between α , ν char and γ (cf. Eq. 2) and spectroscopic measured of macroturbulence ( v macro ), colour-coded by spectroscopicluminosity. The Spearman’s rank correlation coe ffi cient, R , and the corresponding p -value (obtained from a t -test) are also provided. Fig. 4.
Morphology of the stochastic low-frequency variability (cf.Eq. 2) determined from TESS light curves for our sample of OB stars,which are colour-coded by spectroscopic luminosity.
4. Discussion
In this work, we provide quantitative results for measuring thestochastic low-frequency variability in photometry for the largestnumber of massive stars to date. As demonstrated by Fig. 1 (andthe figures in appendix B), the frequency range of the measuredvariability is very broad. To demonstrate the common morphol-ogy in the observed stochastic photometric variability in oursample of 70 massive stars, we plot all the fitted profiles usingEq. (2) in Fig. 4, which are colour coded by each star’s spectro-scopic luminosity. Clearly a common morphology exists for allearly-type stars, but as mentioned in Section 3.1, the relationshipbetween the observed stochastic photometric variability and thespectroscopic parameters of a star is most evident for the mostmassive stars in our sample ( M ≥
20 M (cid:12) ). The variance withinthe morphologies for the main-sequence B stars is quite diverseas expected.The important features in ascertaining the physical mecha-nism for the stochastic low-frequency variability in photometryof massive stars are the steepness of the measured amplitudespectrum and the frequency range for the dominant variability(Bowman et al. 2019a,b). Such features are well characterisedby the fit parameters γ and ν char , respectively. The remaining fitparameters given in Eq. (2) are more dependent on the photomet- ric data. To preserve the homogeneity of the TESS photometry,we do not combine it with light curves from di ff erent telescopes.For example, the amplitude of the stochastic variability at zero-frequency, α , is a function of the wavelength range of the obser-vations, since the light curves from which amplitude spectra arecalculated are not bolometric but in fact wavelength dependent— specifically d F λ / F λ and not d L / L . Also, the white noise am-plitude, C w , is dependent on the number, length and photometricprecision of the data points in the light curve. Previous studies by Blomme et al. (2011); Aerts & Rogers(2015); Bowman et al. (2019a) were limited to a few O stars ob-served by the CoRoT mission. Similarly, Bowman et al. (2019b)studied the stochastic photometric variability in 167 OB starsobserved by the K2 and TESS missions. However, parametersfrom high-resolution spectroscopy were not yet available, andthe location of these stars in the colour-magnitude diagram usingGaia photometry (Gaia Collaboration et al. 2016, 2018) basedon distance (Bailer-Jones et al. 2018), reddening and extinctionestimates (McCall 2004; Green et al. 2018), meant that massesand evolutionary stages could not be inferred for their sample.Nevertheless, the measured morphology of the stochastic low-frequency variability in this large sample of galactic and extra-galactic OB stars studied by Bowman et al. (2019b) yielded γ ≤ . γ , was found to be insensitive to the metallicity of the star, sincethe sample of O and B stars studied by Bowman et al. (2019b)included 114 ecliptic stars (i.e. Z ≥ Z (cid:12) ) observed by the K2 mis-sion and 53 metal-poor stars (i.e. Z (cid:39) . Z (cid:12) ) within the LargeMagellanic Cloud (LMC) galaxy. The properties of the predictedvariability caused by (sub)surface convection are determined bythe e ffi ciency of convection in the iron opacity bump, and thusthe metallicity of a star (Cantiello et al. 2009; Grassitelli et al.2015; Lecoanet et al. 2019), but also the presence of a magneticfield (see e.g. Sundqvist et al. 2013). A study of how magneticfields systematically a ff ect the presence of stochastic photomet-ric variability and macroturbulence in massive stars requires de-tections of these two phenomena for a large sample of magneticstars, which are currently not available.However, in the case of IGWs excited by core convection,only the radius and (convective core) mass of the star set the Article number, page 7 of 17 & A proofs: manuscript no. TESS_IGWs_arxiv_2ver dominant frequency range ( ν (cid:46) ν char ) of the IGW amplitudespectrum with a similar steepness ( γ ). Asteroseismology of co-herent g-mode pulsations has recently allowed the convectivecore masses of 24 Slowly Pulsating B stars observed by the Ke-pler mission, which cover the mass range [3 ,
9] M (cid:12) , to be deter-mined (Pedersen 2020). The simultaneous detection of IGWs inmany of these main sequence B stars also allows the e ffi ciency ofdriving waves by core convection to be tested, but this is beyondthe scope of the current work.Our TESS sample comprises a large number of OB starswith masses above 5 M (cid:12) across the southern ecliptic hemisphereobserved by TESS. The measured values of the steepness ( γ )and dominant frequency range ( ν (cid:46) ν char ) of stochastic low-frequency photometric variability are in full agreement with pre-vious observational findings based on massive stars observed bythe CoRoT and K2 space missions (Bowman et al. 2019a,b).Therefore, our new TESS results provide further observationalevidence that stochastic variability in massive stars is caused byIGWs, either from turbulent core convection and / or from the tur-bulent pressure fluctuations in subsurface convection zones intheir outer envelopes. However, no variability mechanism otherthan IGWs excited by turbulent convection in the deep interiorof stars (i.e. from the convective core during the main sequenceand / or shell-burning for post-main sequence stars) is able toexplain the similar morphology that extends to relatively largefrequencies, the similar γ values, and the large ν char values formetal-poor and metal-rich stars, across such a wide range ofmasses and evolutionary stages of stars. Recent 3D numerical simulations using a physical stellar struc-ture model as the input reference state predict that core convec-tion produces an IGW amplitude spectrum compatible with afrequency exponent of 0 . ≤ γ ≤ γ values for IGWs excited by core con-vection from current 3D hydrodynamical simulations and ob-served γ values for our sample of OB stars is striking.We plot the histogram of the measured γ values for our 70stars in Fig. 5, which is colour-coded using each star’s spectro-scopic luminosity similarly to Bowman et al. (2019a). Our anal-ysis reveals that all massive stars observed by TESS have γ < ≤ γ ≤ .
5. Furthermore, the ob-served stochastic variability is significant up to a relatively highfrequency regime of tens of d − in many of the stars. Such abroad frequency range can be explained by an entire spectrum ofIGWs, which includes a large range of spatial scales (Edelmannet al. 2019; Horst et al. 2020).
5. Conclusions
In this work we have assembled a sample of 70 massive stars thathave spectroscopic parameters determined from high-resolutionground-based spectroscopy by the IACOB project (Simón-Díazet al. 2017; Holgado et al. 2018, and references therein), andhigh-precision time-series photometry from the TESS mission(Burssens et al. 2020). We applied the methodology devised byBowman et al. (2019a) and further developed by Bowman et al.
Fig. 5.
Histogram of the steepness parameter γ (cf. Eq. 2) of the stochas-tic low-frequency variability for our sample of OB stars colour-coded byspectroscopic luminosity. (2019b) to our sample of stars to measure the morphologicalproperties of their stochastic low-frequency variability. This firstinvolved removing any significant frequencies associated withrotational modulation and / or coherent pulsation modes via iter-ative pre-whitening, and subsequently fitting the residual ampli-tude spectra using a Bayesian MCMC framework to determinethe amplitude ( α ), dominant frequency range ( ν (cid:46) ν char ), steep-ness ( γ ), and white-noise term ( C w ). Our sample of stars andtheir determined fit parameters are provided in Tables A.1 andA.2, respectively.We place our sample in the spectroscopic HR diagram us-ing the accurate parameters from ground-based spectroscopyand demonstrate that the morphology of the amplitude spectrumof stochastic photometric variability is sensitive to the spectro-scopic luminosity and evolutionary stage of a star, as shown inFig. 2. We demonstrate that stochastic photometric variability isincreasingly more important in more massive stars, and that themorphology of the variability directly probes the properties of astar. We also find a clear correlation among the amplitude andcharacteristic frequency of the stochastic photometric variabilityand the measured macroturbulent broadening in our sample ofmassive stars. Macroturbulence and spectral line profile variabil-ity have been previously associated with non-radial g-mode pul-sations for main-sequence B stars (e.g. Aerts et al. 2009; Simón-Díaz et al. 2010; Aerts & Rogers 2015) and turbulent pressurefluctuations exciting IGWs for O stars (Grassitelli et al. 2015).Here we show that the photometric amplitudes of the stochasticvariability strongly correlate with the spectroscopic macroturbu-lence, as shown in Fig. 3. Thus, we conclude that our observa-tional study supports the predictions from theoretical and nu-merical work that IGWs, excited by core convection and / or tur-bulent pressure fluctuations, are indeed a plausible mechanismfor macroturbulent broadening in massive stars (Aerts & Rogers2015; Grassitelli et al. 2015; Simón-Díaz et al. 2017).Moreover, we find that the measured fit parameters ν char and γ agree with predictions of the amplitude spectrum of IGWs ex-cited by core convection in hydrodynamical simulations (Rogerset al. 2013; Edelmann et al. 2019; Horst et al. 2020). The dis-tribution in the steepness of the observed amplitude spectra, γ , is shown in Fig. 5. Clearly, the excitation, propagation, anddetectability of IGWs is an important issue for massive starsfrom theoretical, hydrodynamical, and observational perspec-tives (Lecoanet & Quataert 2013; Shiode et al. 2013; Rogers Article number, page 8 of 17. M. Bowman et al.: TESS observations of IGWs in massive stars et al. 2013; Aerts & Rogers 2015; Aerts et al. 2018, 2019b; Gras-sitelli et al. 2015; Augustson & Mathis 2019; Bowman et al.2019a,b; Edelmann et al. 2019; Horst et al. 2020). Our resultsdemonstrate a requirement to include the mixing and angularmomentum transport caused by IGWs in the next generationof stellar structure and evolution models, especially for massivestars on the main sequence. In turn this may alleviate the largediscrepancies between predicted interior rotation rates from cur-rent angular momentum transport theory and observations forstars born with a convective core (Aerts et al. 2019a).Finally, our results are useful for guiding future asteroseis-mic studies of massive stars, such as the most massive O starsand blue supergiants (e.g. Saio et al. 2006; Kraus et al. 2015;Bowman et al. 2019b), which may not be pulsating in coherent p-and / or g-modes but do exhibit photometric variability because ofIGWs. In the future, we will expand our study to include all mas-sive stars in the IACOB database, with long-term light curves as-sembled as part of the nominal and extended TESS mission. Tomore accurately explore the parameter space beyond the mainsequence, we will also extend our methodology to include bluesupergiants in both the Galaxy and the LMC, such that we caninvestigate the role of metallicity on the driving mechanism(s) ofIGWs in massive stars. Acknowledgements.
The authors thank the TESS science team for the excellentdata and the referee for the supportive and constructive comments. The TESSdata presented in this paper were obtained from the Mikulski Archive for SpaceTelescopes (MAST) at the Space Telescope Science Institute (STScI), which isoperated by the Association of Universities for Research in Astronomy, Inc.,under NASA contract NAS5-26555. Support to MAST for these data is pro-vided by the NASA O ffi ce of Space Science via grant NAG5-7584 and by othergrants and contracts. Funding for the TESS mission is provided by the NASAExplorer Program. This research has made use of the SIMBAD database, oper-ated at CDS, Strasbourg, France; the SAO / NASA Astrophysics Data System; andthe VizieR catalog access tool, CDS, Strasbourg, France. Some of the observa-tions used in this work were obtained with the HERMES spectrograph attachedto the Mercator Telescope, operated on the island of La Palma by the FlemishCommunity, at the Spanish Observatorio del Roque de los Muchachos of theInstituto de Astrofísica de Canarias, and further observations obtained with theFEROS spectrograph attached to the 2.2-m MPG / ESO telescope at the La Sillaobservatory. The HERMES spectrograph is supported by the Fund for Scien-tific Research of Flanders (FWO), Belgium, the Research Council of KU Leu-ven, Belgium, the Fonds National Recherches Scientific (FNRS), Belgium, theRoyal Observatory of Belgium, the Observatoire de Genéve, Switzerland, andthe Thüringer Landessternwarte Tautenburg, Germany. The research leading tothese results has received funding from the European Research Council (ERC)under the European Union’s Horizon 2020 research and innovation programme(grant agreement No. 670519: MAMSIE). SS-D acknowledges support from theSpanish Government Ministerio de Ciencia e Innovación through grants PGC-2018-091 3741-B-C22 and SEV 2015-0548, and from the Canarian Agency forResearch, Innovation and Information Society (ACIISI), of the Canary IslandsGovernment, and the European Regional Development Fund (ERDF), undergrant with reference ProID2017010115. The work of PVFE was supported bythe US Department of Energy through the Los Alamos National Laboratory. LosAlamos National Laboratory is operated by Triad National Security, LLC, forthe National Nuclear Security Administration of U.S. Department of Energy(Contract No. 89233218CNA000001). Support for this research was providedby STFC grant ST / S000542 / References
Aerts, C. 2020, Reviews of Modern Physics, under review, arXiv e-prints,arXiv:1912.12300Aerts, C., Bowman, D. M., Símon-Díaz, S., et al. 2018, MNRAS, 476, 1234Aerts, C., Christensen-Dalsgaard, J., & Kurtz, D. W. 2010, Asteroseismology(Springer)Aerts, C. & De Cat, P. 2003, Space Sci. Rev., 105, 453Aerts, C., Mathis, S., & Rogers, T. M. 2019a, ARA&A, 57, 35Aerts, C., Pedersen, M. G., Vermeyen, E., et al. 2019b, A&A, 624, A75Aerts, C., Puls, J., Godart, M., & Dupret, M.-A. 2009, A&A, 508, 409Aerts, C. & Rogers, T. M. 2015, ApJ, 806, L33 Augustson, K. C. & Mathis, S. 2019, ApJ, 874, 83Auvergne, M., Bodin, P., Boisnard, L., et al. 2009, A&A, 506, 411Bailer-Jones, C. A. L., Rybizki, J., Fouesneau, M., Mantelet, G., & Andrae, R.2018, AJ, 156, 58Balona, L. A. 1992, MNRAS, 254, 403Barbá, R., Gamen, R., Arias, J. I., et al. 2014, in Revista Mexicana de Astronomiay Astrofisica Conference Series, Vol. 44, Revista Mexicana de Astronomia yAstrofisica Conference Series, 148–148Barbá, R. H., Gamen, R., Arias, J. I., et al. 2010, in Revista Mexicana de As-tronomia y Astrofisica Conference Series, Vol. 38, Revista Mexicana de As-tronomia y Astrofisica Conference Series, 30–32Barbá, R. H., Gamen, R., Arias, J. I., & Morrell, N. I. 2017, in IAU Symposium,Vol. 329, The Lives and Death-Throes of Massive Stars, ed. J. J. Eldridge,J. C. Bray, L. A. S. McClelland, & L. Xiao, 89–96Blomme, R., Mahy, L., Catala, C., et al. 2011, A&A, 533, A4Borucki, W. J., Koch, D., Basri, G., et al. 2010, Science, 327, 977Bowman, D. M. 2017, Amplitude Modulation of Pulsation Modes in Delta ScutiStars (Springer International Publishing)Bowman, D. M., Aerts, C., Johnston, C., et al. 2019a, A&A, 621, A135Bowman, D. M., Burssens, S., Pedersen, M. G., et al. 2019b, Nature Astronomy,3, 760Breger, M., Stich, J., Garrido, R., et al. 1993, A&A, 271, 482Briquet, M., Aerts, C., Baglin, A., et al. 2011, A&A, 527, A112Briquet, M., Morel, T., Thoul, A., et al. 2007, MNRAS, 381, 1482Burssens, S., Simón-Díaz, S., Bowman, D. M., et al. 2020, A&A, in press, arXive-prints, arXiv:2005.09658Buysschaert, B., Aerts, C., Bloemen, S., et al. 2015, MNRAS, 453, 89Buysschaert, B., Aerts, C., Bowman, D. M., et al. 2018, A&A, 616, A148Cantiello, M. & Braithwaite, J. 2011, A&A, 534, A140Cantiello, M. & Braithwaite, J. 2019, ApJ, 883, 106Cantiello, M., Langer, N., Brott, I., et al. 2009, A&A, 499, 279Castro, N., Oey, M. S., Fossati, L., & Langer, N. 2018, ApJ, 868, 57Chie ffi , A. & Limongi, M. 2013, ApJ, 764, 21Claret, A. & Torres, G. 2016, A&A, 592, A15Claret, A. & Torres, G. 2019, ApJ, 876, 134Couston, L.-A., Lecoanet, D., Favier, B., & Le Bars, M. 2018, Journal of FluidMechanics, 854, R3Daszy´nska-Daszkiewicz, J., Szewczuk, W., & Walczak, P. 2013, MNRAS, 431,3396David-Uraz, A., Owocki, S. P., Wade, G. A., Sundqvist, J. O., & Kee, N. D. 2017,MNRAS, 470, 3672De Cat, P. & Aerts, C. 2002, A&A, 393, 965Deeming, T. J. 1975, Ap&SS, 36, 137Degroote, P., Aerts, C., Michel, E., et al. 2012, A&A, 542, A88Degroote, P., Briquet, M., Auvergne, M., et al. 2010, A&A, 519, A38Degroote, P., Briquet, M., Catala, C., et al. 2009, A&A, 506, 111Dziembowski, W. A., Moskalik, P., & Pamyatnykh, A. A. 1993, MNRAS, 265,588Dziembowski, W. A. & Pamyatnykh, A. A. 1993, MNRAS, 262, 204Edelmann, P. V. F., Ratnasingam, R. P., Pedersen, M. G., et al. 2019, ApJ, 876, 4Ekström, S., Georgy, C., Eggenberger, P., et al. 2012, A&A, 537, A146Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125,306Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2018, A&A, 616, A1Gaia Collaboration, Prusti, T., de Bruijne, J. H. J., et al. 2016, A&A, 595, A1Gelman, A. & Rubin, D. B. 1992, Statistical Science, 7, 457Georgy, C., Ekström, S., Eggenberger, P., et al. 2013, A&A, 558, A103Godart, M., Simón-Díaz, S., Herrero, A., et al. 2017, A&A, 597, A23Grassitelli, L., Fossati, L., Simón-Diáz, S., et al. 2015, ApJ, 808, L31Gray, D. F. 2005, The Observation and Analysis of Stellar Photospheres, 3rdedn. (Cambridge University Press)Green, G. M., Schlafly, E. F., Finkbeiner, D., et al. 2018, MNRAS, 478, 651Groh, J. H., Ekström, S., Georgy, C., et al. 2019, A&A, 627, A24Guinan, E. F., Ribas, I., Fitzpatrick, E. L., et al. 2000, ApJ, 544, 409Handler, G., Jerzykiewicz, M., Rodríguez, E., et al. 2006, MNRAS, 365, 327Handler, G., Pigulski, A., Daszy´nska-Daszkiewicz, J., et al. 2019, ApJ, 873, L4Holgado, G., Simón-Díaz, S., Barbá, R. H., et al. 2018, A&A, 613, A65Holgado, G., Simón-Díaz, S., Haemmerlé, L., et al. 2020, arXiv e-prints,arXiv:2005.05446Horst, L., Edelmann, P. V. F., Andrassy, R., et al. 2020, arXiv e-prints,arXiv:2006.03011Howarth, I. D., Siebert, K. W., Hussain, G. A. J., & Prinja, R. K. 1997, MNRAS,284, 265Howell, S. B., Sobeck, C., Haas, M., et al. 2014, PASP, 126, 398Jenkins, J. M., Twicken, J. D., McCauli ff , S., et al. 2016, in Proc. SPIE, Vol.9913, Software and Cyberinfrastructure for Astronomy IV, 99133EJohnston, C., Pavlovski, K., & Tkachenko, A. 2019, A&A, 628, A25Kaufer, A., Stahl, O., Tubbesing, S., et al. 1999, The Messenger, 95, 8Kaufer, A., Wolf, B., Andersen, J., & Pasquini, L. 1997, The Messenger, 89, 1Keszthelyi, Z., Meynet, G., Georgy, C., et al. 2019, MNRAS, 485, 5843 Article number, page 9 of 17 & A proofs: manuscript no. TESS_IGWs_arxiv_2ver
Keszthelyi, Z., Meynet, G., Shultz, M. E., et al. 2020, MNRAS, 493, 518Koch, D. G., Borucki, W. J., Basri, G., et al. 2010, ApJ, 713, L79Kraus, M., Haucke, M., Cidale, L. S., et al. 2015, A&A, 581, A75Krtiˇcka, J. & Feldmeier, A. 2018, A&A, 617, A121Kurtz, D. W. 1985, MNRAS, 213, 773Langer, N. 2012, ARA&A, 50, 107Langer, N. & Kudritzki, R. P. 2014, A&A, 564, A52Lecoanet, D., Cantiello, M., Quataert, E., et al. 2019, ApJ, 886, L15Lecoanet, D. & Quataert, E. 2013, MNRAS, 430, 2363Lucy, L. B. 1976, ApJ, 206, 499McCall, M. L. 2004, AJ, 128, 2144Miglio, A., Montalbán, J., & Dupret, M.-A. 2007, MNRAS, 375, L21Mo ff at, A. F. J., Marchenko, S. V., Zhilyaev, B. E., et al. 2008, ApJ, 679, L45Mombarg, J. S. G., Van Reeth, T., Pedersen, M. G., et al. 2019, MNRAS, 485,3248Moravveji, E., Aerts, C., Pápics, P. I., Triana, S. A., & Vandoren, B. 2015, A&A,580, A27Moravveji, E., Townsend, R. H. D., Aerts, C., & Mathis, S. 2016, ApJ, 823, 130Murphy, S. J. 2014, PhD thesis, Jeremiah Horrocks Institute, University of Cen-tral Lancashire, UKPápics, P. I., Briquet, M., Baglin, A., et al. 2012, A&A, 542, A55Pedersen, M. G. 2020, PhD thesis, Institute of Astronomy, KU Leuven, BelgiumPedersen, M. G., Chowdhury, S., Johnston, C., et al. 2019, ApJ, 872, L9Ramiaramanantsoa, T., Ratnasingam, R., Shenar, T., et al. 2018, MNRAS, 480,972Raskin, G., van Winckel, H., Hensberge, H., et al. 2011, A&A, 526, A69Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2015, Journal of AstronomicalTelescopes, Instruments, and Systems, 1, 014003Rogers, T. M. 2015, ApJ, 815, L30Rogers, T. M., Lin, D. N. C., McElwaine, J. N., & Lau, H. H. B. 2013, ApJ, 772,21Rogers, T. M. & McElwaine, J. N. 2017, ApJ, 848, L1Saio, H., Kuschnig, R., Gautschy, A., et al. 2006, ApJ, 650, 1111Sana, H., de Mink, S. E., de Koter, A., et al. 2012, Science, 337, 444Schmid, V. S. & Aerts, C. 2016, A&A, 592, A116Shiode, J. H., Quataert, E., Cantiello, M., & Bildsten, L. 2013, MNRAS, 430,1736Simón-Díaz, S., Aerts, C., Urbaneja, M. A., et al. 2018, A&A, 612, A40Simón-Díaz, S., Castro, N., Garcia, M., Herrero, A., & Markova, N. 2011, Bul-letin de la Societe Royale des Sciences de Liege, 80, 514Simón-Díaz, S., Godart, M., Castro, N., et al. 2017, A&A, 597, A22Simón-Díaz, S. & Herrero, A. 2014, A&A, 562, A135Simón-Díaz, S., Herrero, A., Uytterhoeven, K., et al. 2010, ApJ, 720, L174Simón-Díaz, S., Negueruela, I., Maíz Apellániz, J., et al. 2015, in Highlightsof Spanish Astrophysics VIII, ed. A. J. Cenarro, F. Figueras, C. Hernández-Monteagudo, J. Trujillo Bueno, & L. Valdivielso, 576–581Sundqvist, J. O., Petit, V., Owocki, S. P., et al. 2013, MNRAS, 433, 2497Szewczuk, W. & Daszy´nska-Daszkiewicz, J. 2017, MNRAS, 469, 13Szewczuk, W. & Daszy´nska-Daszkiewicz, J. 2018, MNRAS, 478, 2243Telting, J. H., Avila, G., Buchhave, L., et al. 2014, Astronomische Nachrichten,335, 41Thoul, A., Degroote, P., Catala, C., et al. 2013, A&A, 551, A12Tkachenko, A., Pavlovski, K., Johnston, C., et al. 2020, A&A, 637, A60Van Reeth, T., Tkachenko, A., Aerts, C., et al. 2015, A&A, 574, A17 Appendix A: Extended data tables
Article number, page 10 of 17. M. Bowman et al.: TESS observations of IGWs in massive stars
Table A.1.
Parameters of OB stars studied in this work including common name,TIC number, spectral type, e ff ective temperature (log (T e ff )), spectroscopic lu-minosity (log ( L / L (cid:12) )), where L : = T ff / g , projected surface rotational ve-locity ( v sin i ), and macroturbulent broadening ( v macro ), which are taken fromBurssens et al. (2020). The typical uncertainty for log (T e ff ) is estimated to be0.03 dex, and for log ( L / L (cid:12) ) estimated to be 0.15 dex (see Simón-Díaz et al.2017; Holgado et al. 2018). Name TIC Sp. Type log (T e ff ) log ( L / L (cid:12) ) v sin i v macro (km s − ) (km s − ) O dwarf stars:
HD 96715 306491594 O4 V((f))z 4.66 4.10 59 86HD 46223 234881667 O4 V((f)) 4.62 4.16 60 91HD 155913 216662610 O4.5 Vn((f)) 4.63 3.88 278 − HD 46150 234840662 O5 V((f)) 4.61 4.03 71 94HD 90273 464295672 ON7 V 4.59 3.95 55 55HD 110360 433738620 ON7 V 4.59 3.60 96 86HD 47839 220322383 O7 V 4.58 3.70 43 65HD 53975 148506724 O7.5 Vz 4.56 3.74 181 − HD 41997 294114621 O7.5 Vn((f)) 4.55 3.85 262 − HD 46573 234947719 O7 V((f))z 4.56 3.93 77 81HD 48279 234009943 O8 V 4.55 3.76 131 74HD 46056 234834992 O8 Vn 4.55 3.58 370 − HD 38666 100589904 O9.5 V 4.53 3.59 111 56
O subgiant stars:
HD 74920 430625455 O7.5 IVn((f)) 4.54 3.90 291 − HD 135591 455675248 O8 IV((f)) 4.54 3.99 60 60HD 326331 339568114 O8 IVn((f)) 4.54 3.82 332 − HD 37041 427395049 O9.5 IVp 4.54 3.28 134 − HD 123056 330281456 O9.5 IV(n) 4.50 3.70 193 − O giant stars:
HD 97253 467065657 O5 III(f) 4.59 4.16 70 105HD 93843 465012898 O5 III(fc) 4.57 4.15 58 120HD 156738 195288472 O6.5 III(f) 4.58 3.87 65 103HD 36861 436103278 O8 III((f)) 4.55 4.06 53 75HD 150574 234648113 ON9 III(n) 4.52 3.87 252 − HD 152247 339570292 O9.2 III 4.51 3.94 82 96HD 55879 178489528 O9.7 III 4.49 3.85 26 60HD 154643 43284243 O9.7 III 4.49 3.85 101 78
O bright giant and supergiant stars:
CPD-47 2963 30653985 O5 Ifc 4.57 4.16 67 110HD 156154 152659955 O7.5 Ib(f) 4.53 4.22 62 102HD 112244 406050497 O8.5 Iab(f)p 4.50 4.15 124 80HD 151804 337793038 O8 Iaf 4.45 4.33 72 73HD 303492 459532732 O8.5 Iaf 4.45 4.29 87 55HD 57061 106347931 O9 II 4.51 4.01 57 93HD 152249 339567904 OC9 Iab 4.49 4.15 71 70HD 152424 247267245 OC9.2 Ia 4.48 4.14 59 66HD 154368 41792209 O9.5 Iab 4.48 4.28 65 78HD 152003 338640317 O9.7 Iab Nwk 4.48 4.12 65 83HD 152147 246953610 O9.7 Ib Nwk 4.48 4.04 91 64
B dwarf stars:
HD 36960 427373484 B0.5 V 4.46 3.31 23 37HD 37042 427395058 B0.7 V 4.47 3.06 33 13HD 43112 434384707 B1 V 4.41 2.95 7 12HD 35912 464839773 B2 V 4.26 2.44 11 21HD 48977 202148345 B2.5 V 4.25 2.61 26 8
B subgiant stars:
HD 34816 442871031 B0.5 IV 4.46 3.22 25 − HD 46328 47763235 B0.5 IV 4.40 3.28 7 20HD 50707 78897024 B1 IV 4.38 3.31 29 46HD 37481 332913301 B1.5 IV 4.34 2.78 74 21HD 37209 388935529 B2 IV 4.38 2.76 50 15HD 26912 283793973 B3 IV 4.20 2.61 53 30
Article number, page 11 of 17 & A proofs: manuscript no. TESS_IGWs_arxiv_2ver
Table A.1. continued.
Name TIC Sp. Type log (T e ff ) log ( L / L (cid:12) ) v sin i v macro (km s − ) (km s − )HD 37711 59215060 B3 IV 4.21 2.61 68 51HD 57539 10176636 B3 IV 4.13 2.54 162 13HD 41753 151464886 B3 IV 4.23 2.61 25 40HD 224990 313934087 B5 IV 4.13 2.33 20 40 B giant stars:
HD 48434 234052684 B0 III 4.48 3.93 48 82HD 61068 349043273 B2 III 4.39 3.08 12 23HD 35468 365572007 B2 III 4.29 2.99 53 27
B bright giant and supergiant stars:
HD 44743 34590771 B1 II-III 4.37 3.20 24 40HD 54764 95513457 B1 II 4.30 3.97 123 87HD 52089 63198307 B2 II 4.34 3.60 26 50HD 51309 146908355 B3 II 4.20 3.64 27 43HD 46769 281148636 B5 II 4.11 3.16 70 23HD 27563 37777866 B7 II 4.16 2.61 34 26HD 53244 148109427 B8 II 4.14 2.74 36 21HD 37128 427451176 B0 Ia 4.47 4.05 55 85HD 38771 66651575 B0.5 Ia 4.47 4.06 53 83HD 53138 80466973 B3 Iab 4.23 4.13 37 56HD 39985 102281507 B9 Ib 4.11 2.34 26 28
Peculiar stars:
HD 37061 427393920 O9.5 V 4.49 3.07 210 − HD 37742 11360636 O9.2 Ib var Nwk 4.47 4.15 122 97HD 57682 187458882 O9.2 IV 4.54 3.62 12 38HD 54879 177860391 O9.7 V 4.52 3.16 7 10
Table A.2.
Optimised parameters for the morphology of low-frequency variabil-ity (cf. Eq. (2)) using a Bayesian MCMC fitting method.
Name TIC α ν char γ C W ( µ mag) (d − ) ( µ mag) O dwarf stars:
HD 96715 306491594 99 . ± .
039 5 . ± . . ± . . ± . . ± .
047 3 . ± . . ± . . ± . . ± .
033 6 . ± . . ± . . ± . . ± .
044 3 . ± . . ± . . ± . . ± .
051 2 . ± . . ± . . ± . . ± .
061 2 . ± . . ± . . ± . . ± .
046 3 . ± . . ± . . ± . . ± .
055 2 . ± . . ± . . ± . . ± .
040 2 . ± . . ± . . ± . . ± .
058 2 . ± . . ± . . ± . . ± .
052 2 . ± . . ± . . ± . . ± .
026 9 . ± . . ± . . ± . . ± .
048 4 . ± . . ± . . ± . O subgiant stars:
HD 74920 430625455 367 . ± .
036 3 . ± . . ± . . ± . . ± .
061 1 . ± . . ± . . ± . . ± .
056 2 . ± . . ± . . ± . . ± .
085 0 . ± . . ± . . ± . . ± .
033 4 . ± . . ± . . ± . O giant stars:
HD 97253 467065657 837 . ± .
048 2 . ± . . ± . . ± . . ± .
060 2 . ± . . ± . . ± . . ± .
032 4 . ± . . ± . . ± . . ± .
064 2 . ± . . ± . . ± . . ± .
049 2 . ± . . ± . . ± . . ± .
043 2 . ± . . ± . . ± . . ± .
063 1 . ± . . ± . . ± . . ± .
061 1 . ± . . ± . . ± . Article number, page 12 of 17. M. Bowman et al.: TESS observations of IGWs in massive stars
Table A.2. continued.
Name TIC α ν char γ C W ( µ mag) (d − ) ( µ mag) O bright giant and supergiant stars:
CPD-47 2963 30653985 705 . ± .
043 2 . ± . . ± . . ± . . ± .
056 1 . ± . . ± . . ± . . ± .
090 0 . ± . . ± . . ± . . ± .
055 0 . ± . . ± . . ± . . ± .
057 0 . ± . . ± . . ± . . ± .
078 1 . ± . . ± . . ± . . ± .
044 1 . ± . . ± . . ± . . ± .
059 0 . ± . . ± . . ± . . ± .
037 1 . ± . . ± . . ± . . ± .
090 0 . ± . . ± . . ± . . ± .
060 1 . ± . . ± . . ± . B dwarf stars:
HD 36960 427373484 30 . ± .
047 1 . ± . . ± . . ± . . ± .
066 1 . ± . . ± . . ± . . ± .
117 0 . ± . . ± . . ± . . ± .
048 2 . ± . . ± . . ± . . ± .
066 1 . ± . . ± . . ± . B subgiant stars:
HD 34816 442871031 61 . ± .
114 0 . ± . . ± . . ± . . ± .
112 1 . ± . . ± . . ± . . ± .
035 4 . ± . . ± . . ± . . ± .
061 1 . ± . . ± . . ± . . ± .
097 0 . ± . . ± . . ± . . ± .
061 1 . ± . . ± . . ± . . ± .
058 0 . ± . . ± . . ± . . ± .
044 2 . ± . . ± . . ± . . ± .
048 1 . ± . . ± . . ± . . ± .
065 1 . ± . . ± . . ± . B giant stars:
HD 48434 234052684 1647 . ± .
063 1 . ± . . ± . . ± . . ± .
090 1 . ± . . ± . . ± . . ± .
078 0 . ± . . ± . . ± . B bright giant and supergiant stars:
HD 44743 34590771 133 . ± .
080 1 . ± . . ± . . ± . . ± .
070 0 . ± . . ± . . ± . . ± .
069 1 . ± . . ± . . ± . . ± .
087 0 . ± . . ± . . ± . . ± .
056 1 . ± . . ± . . ± . . ± .
059 0 . ± . . ± . . ± . . ± .
065 0 . ± . . ± . . ± . . ± .
069 0 . ± . . ± . . ± . . ± .
083 0 . ± . . ± . . ± . . ± .
108 0 . ± . . ± . . ± . . ± .
081 1 . ± . . ± . . ± . Peculiar stars:
HD 37061 427393920 34 . ± .
068 1 . ± . . ± . . ± . . ± .
070 1 . ± . . ± . . ± . . ± .
095 1 . ± . . ± . . ± . . ± .
052 1 . ± . . ± . . ± . Appendix B: Fitted amplitude spectra figures
Article number, page 13 of 17 & A proofs: manuscript no. TESS_IGWs_arxiv_2ver
Fig. B.1.
Fitted logarithmic amplitude spectra of stars given in Table A.2. Line styles and colours are the same as in Fig. 1.Article number, page 14 of 17. M. Bowman et al.: TESS observations of IGWs in massive stars
Fig. B.2.
Fitted logarithmic amplitude spectra of stars given in Table A.2. Line styles and colours are the same as in Fig. 1.Article number, page 15 of 17 & A proofs: manuscript no. TESS_IGWs_arxiv_2ver
Fig. B.3.
Fitted logarithmic amplitude spectra of stars given in Table A.2. Line styles and colours are the same as in Fig. 1.Article number, page 16 of 17. M. Bowman et al.: TESS observations of IGWs in massive stars