SPB stars in the open SMC cluster NGC 371
C. Karoff, T. Arentoft, L. Glowienka, C. Coutures, T. B. Nielsen, G. Dogan, F. Grundahl, H. Kjeldsen
aa r X i v : . [ a s t r o - ph ] F e b Mon. Not. R. Astron. Soc. , 1–8 (2008) Printed 6 December 2018 (MN L A TEX style file v2.2)
SPB stars in the open SMC cluster NGC 371
C. Karoff ⋆ , T. Arentoft , L. Glowienka , C. Coutures , T. B. Nielsen ,G. Dogan , F. Grundahl & H. Kjeldsen Department of Physics and Astronomy, University of Aarhus, DK-8000 Aarhus C, Denmark Institut d’Astrophysique de Paris, CNRS, Universite Pierre et Marie Curie UMR7095, 98bis Boulevard Arago, 75014 Paris, France
ABSTRACT
Pulsation in β Cep and SPB stars are driven by the κ mechanism which dependscritically on the metallicity. It has therefore been suggested that β Cep and SPBstars should be rare in the Magellanic Clouds which have lower metallicities than thesolar neighborhood. To test this prediction we have observed the open SMC clusterNGC 371 for 12 nights in order to search for β Cep and SPB stars. Surprisingly, wefind 29 short-period B-type variables in the upper part of the main sequence, manyof which are probably SPB stars. This result indicates that pulsation is still drivenby the κ mechanism even in low metallicity environments. All the identified variableshave periods longer than the fundamental radial period which means that they cannotbe β Cep stars. Within an amplitude detection limit of 5 mmag no stars in the topof the HR-diagram show variability with periods shorter than the fundamental radialperiod. So if β Cep stars are present in the cluster they oscillate with amplitudes below5 mmag, which is significantly lower than the mean amplitude of β Cep stars in theGalaxy. We see evidence that multimode pulsation is more common in the upper partof the main sequence than in the lower. We have also identified 5 eclipsing binariesand 3 periodic pulsating Be stars in the cluster field.
Key words: star: early-type — stars: oscillations (including pulsations) — galaxies:Magellanic Clouds β Cep and slowly pulsating B stars (SPB) pulsate dueto the κ -mechanism activated by the metal opacity bump(Cox et al. 1992). This give rise to p -mode pulsation in β Cep stars and g -mode pulsation in SPB stars. However thetheoretical standard models do not predict the presence ofpulsation in β Cep and SPB stars in low-metallicity envi-ronments such as the Magellanic Clouds.Analysis of OGLE data have indeed shown that β Cep and SPB stars exist in the Magellanic Clouds(Pigulski & Ko laczkowski 2002; Ko laczkowski et al. 2004,2006) though it is not clear if the rate of β Cep and SPBstars is lower than or equal to the rate in the Galaxy. Wehave therefore observed the open SMC cluster NGC 371 for12 nights in order to search for β Cep and SPB stars. Ob-serving a single cluster instead of the the entire SMC hasa number of advantages, – e.g. the age and metallicity ofthe cluster can be constrained from fitting isochrones to theHR-diagram.The metallicity of the SMC has been measured to be- ⋆ E-mail: karoff@phys.au.dk tween Z = 0.001 and Z = 0.004 (see Maeder et al. 1999,and references therein), while Miglio et al. (2007a,b) havecalculated instability domains of β Cep and SPB stars us-ing improved opacities and metal abundances which do notpredict pulsation in β Cep and SPB stars for
Z < β Cep or SPB starsin the SMC.In order to test this prediction we have chosen to ob-serve the cluster NGC 371 as this cluster has been observedbefore by Kjeldsen & Baade (1994) who saw some signs ofvariability, though no clear evidence of pulsation was seen.The observations presented in this paper have beenmade with a single telescope. This means that clear modeand frequency determination can not be obtained and thefrequencies can not be used for modeling. Instead this workpresents a number of candidate SPB stars for followup ob-servations with at least dual-side photometry and with spec-troscopy.The paper is arranged as follows. Section 2 presentsthe cluster. Section 3 describes how we obtained, reducedand analyzed the data. The identified eclipsing binaries arepresented in Section 4, the Be stars in Section 5 and thepulsating stars in the upper part of the main sequence in c (cid:13) Karoff et al.
Figure 1.
DFOSC image of NGC371, with the variables marked.
Section 6. A summary and concluding remarks are found inSection 7.
NGC 371 ( α , δ = 11 h m . s − ◦ ′ . ′′
0) is ayoung open cluster in the SMC. Wisniewski & Bjorkman(2006) estimated a log(age) of 6.7 based on isochrone fittingto OGLE data assuming a metallicity of Z = 0.002. As thereare no high-resolution spectra available for NGC 371 it hasnot been possible to estimate the metallicity autonomously.The field of the cluster is shown in Fig. 1 with the iden-tified variable stars marked. The observations were obtained with the DFOSC instru-ment at the Danish 1.54-m telescope at ESO, La Silla during12 nights in August and September 2005. 763 frames werecollected in I and 623 in B . The same pointing was main-tained during the observations, i.e., the stars were kept atfixed positions on the CCD (within a few pixels) during theobserving run. Because of the crowding in the field the ob-servations were always made in focus and the exposure timeswere then adjusted according to the seeing so that only thebrightest 5 % of the stars were saturated. This resulted inexposure times of approximately 20 s in I and 60 s in B . InFig. 2 all the data are shown for one of the bright stars inNGC 371, illustrating the time distribution of the data set.The CCD images were calibrated using standard proce-dures, i.e. a master BIAS was subtracted from each framebefore the frames were divided by a master sky-flat in eachfilter. The master BIAS was obtained from a large num-ber of frames from the whole observing run and the mastersky-flats were obtained as the median of a large number ofevening and morning sky-flats. We checked that both the Figure 2.
Light curves of one of the bright stars in NGC 371showing the precision level and data sampling.
BIAS and the flat-fields were indeed stable over the lengthof the observing run.The photometric reductions were done using the soft-ware package MOMF (Kjeldsen & Frandsen 1992). MOMFapplies a very robust algorithm combining PSF and aperturephotometry in semi-crowded fields.Some of the light curves showed residuals of systematictrends caused by, e.g., air mass, cloud cover, and tempera-ture variations (these residuals are sometimes referred to as”red noise”). We therefore applied the algorithm introducedby Tamuz et al. (2005) in order to correct for systematic ef-fects. This clearly improved the rms noise level, in particularin the brightest stars.The resulting light curves had rms noise levels over thewhole observing run ranging from a few mmag in the brightend and up to 50 mmag in the faint end. Some of the lightcurves showed night-to-night drifts, which means that therms noise level in the individual nights was lower than overthe whole observing run. Fig. 2 shows the precision in atypical bright star.
We obtained differential light curves of 6842 stars in the 13.5’ × UBV I photometry was obtained fromZaritsky et al. (2002).We used two complementary algorithms to searchfor variability; the analysis of variance periodograms(Schwarzenberg-Czerny 1996) and simultaneous iterativesine wave fitting (Frandsen et al. 1995) based on the Lombperiodogram (Lomb 1976). We searched for variability bothwith and without using statistical weights (Frandsen et al.1995). The statistical weights were assigned as the standarddeviation of all data points separated by less than 15 min-utes from a given data point. For the analysis of varianceapproach we made visual examination of all light curveswhich had fit qualities better than 0.9 and for the Lombperiodogram approach we made visual examination of alllight curves where the same peak in the periodogram waspresent in both B and I with a S/N higher than 4 in ampli- c (cid:13) , 1–8 PB stars in NGC 371 Figure 3.
Phase light curves of the 5 eclipsing binaries identifiedin NGC 371 tude (Breger et al. 1993). In total we ended up with a listof a little over a hundred stars that were selected for furtheranalysis.In the visual examination, stars were rejected mainlybecause of one of the following three causes: stars wereplaced outside the cluster region, stars were placed closeto a hot pixel or a bad column, or the variability was notthe same in the two filters. This gave us the list of the 37stars presented in this paper. Light curves of all the variablestars will be added to CDS.It is not possible to determine if a star is a true clus-ter member or not, as we only have two dimensional imagesavailable. To determine if a star is likely to be a true clus-ter member in three dimension a spectroscopic analysis isneeded. Therefore we can not give a reliable estimate of thenumber of cluster members and some of the identified vari-able stars could turn out not to be clusters members.
Table 1.
Stellar parameters for 5 eclipsing binaries in NGC 371.Periods are in daysID α δ B B – I PV1 01 03 04.1 -72 08 04 18.74 0.07 1.348V2 01 03 34.9 -72 08 60 18.64 2.68 1.045V3 01 03 40.5 -72 01 26 18.86 -0.17 0.556V4 01 30 40.0 -72 02 33 19.14 -0.22 0.463V5 01 03 43.5 -72 05 11 18.64 -0.23 0.609
Table 2.
Stellar parameters for 3 Be stars in NGC 371. The id’sin the second column (ID II) are from Wisniewski & Bjorkman(2006).ID ID II α δ B B – IV6 WBBe 43 01 03 34.5 -72 06 42 16.65 0.35V7 WBBe 110 01 03 30.5 -72 03 46 17.99 -0.34V8 WBBe 18 01 03 52.6 -72 05 39 15.73 0.21
We have detected 5 eclipsing binaries (phased light curvesare shown in Fig. 3). V1, V2 and V3 are located on theedge of the cluster, which means that they might not becluster members. V4 and V5 are located safely inside thecluster and these stars could be important in determiningthe precise distance and age of the cluster. This informationwill again be important when trying to model the excitationmechanisms of β Cep and SPB stars based on data from thiscluster.V2 also shows some signs of pulsation with a frequencyof 6.2 c/d, which suggest that this could be a binary sys-tem with one of the members being a pulsating star. Butmore photometry will be needed in order to evaluate thephenomenon properly, – i.e removing the eclipse from thelight curve before analyzing the pulsation.
Three of the detected variable stars match with the starsidentified as Be stars by Wisniewski & Bjorkman (2006).They have identified 118 Be stars and 11 candidate Be starsin NGC371 based on two-color diagrams of B , V , R and Hα photometry. Though the nature of the variability of Be starsis believed to be transient, Be stars might also exhibit g - or p -mode pulsation. It is therefore clear that these three starscan be used to gain understanding of the relation between g - and p -mode pulsation and the variability of Be stars inlow-metallically environments. The amplitude spectra of thethree Be stars are shown in Fig. 4. The light curves of thethree Be stars all show coherent variability, which indicatethat the variability originates from pulsation and not fromtransient events like activity. c (cid:13) , 1–8 Karoff et al.
Figure 4.
Amplitude spectra for 3 Be stars in NGC371. Notethat the scale on the y-axis is different for the three stars.
The frequency analysis of the variable stars in the upperpart of the main sequence followed Arentoft et al. (2007).This means that we made a simultaneous least-squares fitto all the frequencies with a S/N higher than 4 in the am-plitude spectra of both the I and the B filter data. Foreach frequency we manually inspected the amplitude spec-tra, in order to find the frequency value that best describedthe variation in both filters. This step could not be madecompletely objective as daily aliases were present in the am-plitude spectra because the observations were made from asingle site. Therefore some of the detected frequencies mightbe shifted 1–2 c/d from the true oscillation frequencies. Theuncertainties on the amplitudes and phases were calculatedbased on the error matrix in the least-squares fitting proce-dure, though we also used Monte Carlo simulations to checkthe consistency of these uncertainties [this was done withthe use of Period04 (Lenz & Breger 2005)]. This procedurefor calculating the uncertainties assumes that we know thefrequencies, – i.e. that we know which peak reflect the oscil-lation mode. The calculated uncertainties does therefore notreflect the uncertainty originating from mismatch betweenthe true oscillation mode and the daily aliases.All the detected frequencies for the pulsating stars aregiven in Table 3 together with color, amplitude and phaseinformation, three examples of amplitude spectra of pulsat-ing stars are shown in Fig. 5 and phased light curves areshown in Fig. 6. Figure 5.
Amplitude spectra for three pulsating stars in theupper part of the main sequence of NGC 371
Figure 6.
Phased light curves for three pulsating stars in theupper part of the main sequence of NGC 371c (cid:13) , 1–8
PB stars in NGC 371 Table 3.
Stellar parameters for the pulsating stars in NGC 371. Frequencies ( ν ) are in c/d, amplitudes ( A B ) are in mmag, phasedifferences ( φ I − φ B ) are in radians and pulsation constants Q are in d − . The quoted errors on the amplitude ratios and phasedifferences are based on the error matrix in the least-squares fitting procedure.ID α δ B [mag] B – I ν i ν A B A I /A B φ I − φ B Q V9 01 03 32.0 -72 05 21 13.72 -0.58 ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν We have used the photometry from Zaritsky et al. (2002) tomake a B versus B − I color-magnitude-diagram of NGC371 as shown in Fig. 7. The center of the cluster was placedon α , δ = 11 h m . s − ◦ ′ . ′′ β Cep and SPB starswould be located if the metallicity of the cluster had beenlarger. Instability analysis of models of hot main sequencestars of solar metallicity show a group of mostly p-modepulsators where the longest period is the fundamental radialmode and a second group of cooler stars with mostly g-modes and longer periods. The first group can be identifiedwith the classical β Cep stars (early B-type stars) while thesecond group can be identified with the SPB stars (mid B- type stars). Models based on revised solar metal abundancesand different opacities show that the hot boundary of theSPB instability strip overlaps with the β Cep instabilitystrip (Miglio et al. 2007a,b). At the metallicity of NGC 371,however, no pulsations are excited in the models as alreadymentioned. Therefore theory cannot be used as a guide inclassification. We will come back to this in the next section.We have used the Padova evolutionary code describedin Marigo et al. (2007) to calculate an isochrone for thecolor-magnitude-diagram of the cluster. The model used so-lar heavy elements mixture. We adopted a metalicity of Z =0.002 and a log(age) of 6.7 from Wisniewski & Bjorkman(2006). The distance modulus was chosen to 18.7 fromCrowl et al. (2001) and the reddening to E ( B − V ) = 0 . c (cid:13) , 1–8 Karoff et al.
Figure 7.
Color-magnitude-diagram of NGC 371 with the iden-tified SPB candidates marked. The solid line shows a isochronefit to the observations.
In the Galaxy, β Cephei stars have periods which are con-sistent with mostly p-mode pulsations. The longest periodis that of the fundamental radial mode for which the pulsa-tion constant Q ≈ .
033 d − (Stankov & Handler 2005). Toclassify stars in NGC 371 we need an estimate of Q for eachstar:log Q = log P + 1 / M/M ⊙ ) − / L/L ⊙ )+3 log( T /T ⊙ ) . where P is the period in days, M the mass, L the luminos-ity and T the effective temperature. Since we do not havedata to estimate the masses, luminosities and effective tem-peratures for individual stars, we use these values of theisochrone shown in Fig. 7. The result is shown in Fig. 8together with all the identified periods as a function of B magnitude. There the mode with the longest period in eachstar is plotted with large crosses, while the secondary modeswith shorter periods are plotted with small diamonds.In Fig. 8 it is seen that all stars have periods that arelonger than the fundamental radial periods. We thereforeidentify all the stars as candidate SPB stars. Classificationof the stars as bona fide SPB stars would require more timeseries observations to ensure that the periods are correct andexclude other sources of variability, such as close binaries.If there are β Cep stars in the cluster then we canassume two things about them; That they are generallybrighter than the SPB stars and that they have periodsshorter than the fundamental radial period. We do not de-tect any pulsating stars in this domain than fulfill the de-tection criteria given in Section 3.1 and therefore we do notdetect any candidate β Cep stars in NGC 371.In order to constrain an upper detection limit of β Cepstars in the cluster we measured the highest peak in theamplitude spectrum between 8 and 20 c/d for all the starsbrighter than 15.8 in B . A frequency range from 8 to 20c/d is in agreement with the frequency range in which the β Cep stars are expected for this cluster (see Fig. 8). Forthe brightest stars we could also have included periods upto 0.4 day, but this gave problems with the 1 /f noise in theamplitude spectra of some of the stars.In Fig. 9 we have plotted the amplitude of the high- Figure 8.
Magnitude-period relation for the candidate SPBstars stars together with the fundamental radial period from theisochrone. The mode with the longest period in each star is plot-ted with large crosses, while the secondary modes with shorterperiods is plotted with small diamonds. est peaks as a function of magnitude together with themean V oscillation amplitude of galactic β Cep stars fromStankov & Handler (2005). It is seen that our detection limitis significantly lower than the mean V oscillation amplitudeof galactic β Cep. If β Cep stars are present in the clusterthey have oscillation amplitudes significantly lower than thegalactic β Cep stars.Though we do not detect any candidate β Cep starssome of the stars do show variability between 8 and 20 c/d,but not with a S/N higher than 4 in both filters. In Fig. 9 wehave marked the peaks which have S/N higher than 3.5 in B and in Fig. 10 we have plotted the amplitude spectrum ofone of them. The detection criteria of S/N higher than 4 inboth filters is generally conservative and for longer regularlysampled data sets it is too high as the observations in thetwo filters are independent. But for the present data set itseems as a good indication of pulsation. More photometricdata is needed in order to classify these stars as candidate β Cep stars. On the other hand if these stars are β Cep starsthey would be very interesting targets as they would haveshorter periods and lower oscillation amplitudes comparedto the galactic β Cep stars.Though periods below 0.1 day are not common in β Cepstars they have been seen in the very young cluster NGC6231 (Balona & Shobbrook 1983; Arentoft et al. 2001).
We have also calculated and plotted the amplitude ratiosand phase difference between the B and I filters (Fig. 11) inorder to identify any systematic effects that could be com-pared with theoretical models as in Arentoft et al. (2007)or used for mode identification (Heynderickx et al. 1994).Though we do not find any systematic effects in phasedifference versus amplitude ratio within the uncertainties(Fig. 11) we do see that the main part of the observed fre-quencies cluster around an amplitude ratio between 0.5 and1.0 which is expected for oscillations in β Cep stars withdegrees between 0 and 2 (Heynderickx et al. 1994).The
I/B amplitude ratio can also be used to excludeother causes of the variability than pulsation (e.g. ellipsoidal c (cid:13) , 1–8 PB stars in NGC 371 Figure 9.
Amplitudes of the highest peak as a function of mag-nitude for the stars brighter than the identified candidate SPBstars. The diamonds mark the peaks with a S/N higher than 3.5.The horizontal line at 17 mmag shows the mean oscillation am-plitude of β Cep stars in the Galaxy in V . The arrow marks thestar plotted in Fig. 10. variables) as the I/B amplitude ratios are general belowunity (see, e.g., Arentoft et al. 2007, and references therein).
The last thing to note about the frequencies is that all themultimode pulsators are found among the pulsating starsin the upper part of the main sequence (except for one).One reason for this could be that these stars are brighter,which in general gives a higher S/N, but some of the fre-quencies that we detect in the stars in the lower part of themain sequence have such a high amplitude that multimodepulsation should have been discovered if it was present (seeTable 3).
We have identified 5 eclipsing binaries, 3 periodic Be starsand 29 candidate SPB stars in NGC371.The results indicates that excitation of oscillations inSPB stars is more common in low metallicity environmentssuch as the SMC than predicted by standard stellar mod-els. As a possible extension to the standard stellar modelsto account for this discrepancy could be to include localiron enhancement by diffusion and radiative accelerations(Miglio et al. 2007c).If there are β Cep stars in the cluster they would gener-ally be brighter than the SPB stars and have periods shorterthan the fundamental radial period and we do not detectpulsation in any stars in this domain with an upper limit onthe oscillation amplitudes of 5 mmag.Though it is not possible to calculate a reliable num-ber for the fraction of SPB stars in NGC 371 the absolutenumber of pulsating stars seems to be high compared tothe fraction obtained for the Galaxy by Stankov & Handler(2005).We see evidence that multimode pulsation is more com-mon in upper part of the main sequence than in the lower.We have also identified periodic pulsation in 3 stars that
Figure 10.
Amplitude spectrum in B of one of the stars showingsigns of β Cep variability between 8 and 20 c/d with a S/N below4. were previously identified as Be stars. These stars can usedto study the relation between the Be phenomenon and β Cep and SPB stars in low-metallicity environments.The results presented here strongly confirm and increaseour interest in NGC 371 as an excellent laboratory for β Cepand SPB stars.
ACKNOWLEDGMENTS
We thank the referee Luis Balona for suggesting that we ex-amined the fundamental radial periods of the stars, whichhelped to improve the paper significantly. We also thankJørgen Christensen-Dalsgaard for useful suggestions. TheDanish Natural Science Research Council and the Instru-ment Center for Danish Astrophysics (IDA) are acknowl-edged for financial support. C.K. and F.G. acknowledges fi-nancial support from IDA. C.K, T.A., G.D. and F.G. also ac-knowledges support from the Danish AsteroSeismology Cen-tre.
REFERENCES
Arentoft, T., Sterken, C., Knudsen, M. R., Freyhammer,L. M., Duerbeck, H. W., Pompei, E., Delahodde, C. E.,& Clasen, J. W. 2001, A&A, 380, 599Arentoft, T., et al. 2007, A&A, 465, 965Balona, L. A., & Shobbrook, R. R. 1983, MNRAS, 205, 309Breger, M., et al. 1993, A&A, 271, 482Cox, A. N., Morgan, S. M., Rogers, F. J., & Iglesias, C. A.1992, ApJ, 393, 272Crowl, H. H., Sarajedini, A., Piatti, A. E., Geisler, D., Bica,E., Clari´a, J. J., & Santos, J. F. C., Jr. 2001, AJ, 122, 220Frandsen, S., Jones, A., Kjeldsen, H., Viskum, M., Hjorth,J., Andersen, N. H., & Thomsen, B. 1995, A&A, 301, 123Heynderickx, D., Waelkens, C., & Smeyers, P. 1994, A&ASuppl., 105, 447Kjeldsen, H., & Baade, D. 1994, Pulsation; Rotation; andMass Loss in Early-Type Stars, 162, 29Kjeldsen, H., & Frandsen, S. 1992, PASP, 104, 413Ko laczkowski, Z., et al. 2004, IAU Colloq. 193: VariableStars in the Local Group, 310, 225 c (cid:13) , 1–8 Karoff et al.
Figure 11.
Phase difference I − B versus amplitude ratio I/B for the candidate SPB stars
Ko laczkowski, Z., et al. 2006, Memorie della Societa Astro-nomica Italiana, 77, 336Larsen, S. S., Clausen, J. V., & Storm, J. 2000, A&A, 364,455Lenz, P., & Breger, M. 2005, Communications in Astero-seismology, 146, 53Lomb, N. R. 1976, Ap&SS, 39, 447Maeder, A., Grebel, E. K., & Mermilliod, J.-C. 1999, A&A,346, 459Miglio, A., Montalb´an, J., & Dupret, M.-A. 2007a, MN-RAS, 375, L21Miglio, A., Montalb´an, J., & Dupret, M.-A. 2007b, Com-munications in Asteroseismology, 151, 48Miglio, A., Bourge, P.-O., Montalb´an, J., & Dupret, M.-A.2007c, Communications in Asteroseismology, 150, 209Marigo, P., Girardi, L., Bressan, A., Groenewegen,M. A. T., Silva, L., & Granato, G. L. 2007, ArXiv e-prints,711, arXiv:0711.4922Pietrinferni, A., Cassisi, S., Salaris, M., & Castelli, F. 2004,ApJ, 612, 168Pamyatnykh, A. A. 1999, Acta Astronomica, 49, 119Pigulski, A., & Ko laczkowski, Z. 2002, A&A, 388, 88Schwarzenberg-Czerny, A. 1996, ApJ, 460, L107Stankov, A., & Handler, G. 2005, ApJS, 158, 193Tamuz, O., Mazeh, T., & Zucker, S. 2005, MNRAS, 356,146Zaritsky, D., Harris, J., Thompson, I. B., Grebel, E. K., &Massey, P. 2002, AJ, 123, 855Wisniewski, J. P., & Bjorkman, K. S. 2006, ApJ, 652, 458 c (cid:13)000