Chemical abundance studies of CP stars in open clusters
aa r X i v : . [ a s t r o - ph . S R ] N ov Chemical abundance studies of CP stars inopen clusters
Fossati L. Argelander-Institut f¨ur Astronomie der Universit¨at Bonn, Auf dem H¨ugel 71,53121, Bonn, Germany, email: [email protected]
Abstract
In stellar astrophysics, the study of the atmospheres of early-typestars plays a very special role. The atmospheres of these stars display a variety ofdifferent phenomena, such as the presence of large magnetic fields, strong surfaceconvection, pulsation, diffusion of chemical elements. In particular, about 10% ofearly F-, A- and late B-type stars present chemical peculiarities, which rise as aresult of diffusion. A detailed study of the evolution of the chemical peculiaritiesas traces of diffusion processes requires the precise knowledge of stellar ages andinitial chemical composition. Open clusters provide these information: 1) it ispossible to assume that all cluster members have approximately the same originalchemical composition and age; 2) the age of stars belonging to open clusters canbe determined with much higher accuracy than for field stars. For this reasonchemically peculiar stars member of open clusters have been targeted to studythe evolution of the chemical peculiarities primarily to provide constraints todiffusion models. We review the abundance studies of chemically peculiar stars inopen clusters performed until now, putting their results into the broader contextof stellar evolution.
1. Introduction
Stellar evolution is the study of how stars change with time and of thephysical processes driving these changes. The study of stellar evolution isof crucial importance in almost every astrophysical area, from planets togalaxies. For example, our understanding of the past and future of theEarth’s climate and habitability is tightly bound to our understanding ofthe Sun and of its evolution as a star.Stars spend most of their lifetime on the main sequence, burning hy-drogen in the core, during which they go through only little structuralchanges. Main sequence stars are therefore well studied and believed to bewell understood. Nevertheless, several physical processes such as diffusion,convection and rotation, driving stellar evolution along the main sequencephase, are far from being well understood and reproduced by the currentstate-of-the-art stellar evolution models (e.g., Langer [25]). The study ofchemically peculiar (CP) stars allows us to better understand some of the
Fossati, L. most basic physical processes driving stellar evolution, such as diffusion.In absence of mixing (e.g., no rotation and convection) the diffusionprocesses (see e.g., Michaud [29] and references therein), which lead tochemical peculiarities in CP stars, can be generally simplified as follows.Each ion inside a star is simultaneously pushed towards the outer regionsby the pressure of the photons coming from the star’s core (i.e., radia-tion pressure), and towards the stellar interior by the star’s gravity. Fol-lowing their basic structure and characteristics, such as their mass andcross-section to radiation, different ions react differently. In a perfectly hy-drostatic stellar envelope, the ions for which radiation pressure is strongerthen gravity will be pushed towards the star’s surface leading to an ob-servable overabundance of the considered ions; on the other hand, the ions,for which gravity is stronger than radiation pressure, will sink inside thestar, leading to an observable underabundance in the star’s envelope (e.g.,Richer et al. [32]).Since diffusion is a time-dependent process involving the evolution ofchemical abundances, the knowledge of the stellar age and initial chemicalcomposition is crucial. Open cluster stars are therefore the best suitabletargets to study diffusion as a function of time and stellar mass, becauseof two very compelling reasons. First, it is possible to assume that all clus-ter members have approximately the same original chemical compositionand age. Therefore, when the analysis of chemical abundances of stars be-longing to the same cluster is performed, it is possible to assume that thestudied objects are different only by their initial mass, rotational velocityand binarity. Second, the age of stars belonging to open clusters can bedetermined with much higher accuracy then for objects in the field. Thisis clearly shown in Fig. 1 of Bagnulo et al. [2]: assuming for a given fieldearly-type star the effective temperature ( T eff ) and luminosity ( log L/L ⊙ )are know with typical uncertainties of 5% and 0.1 dex, respectively, onecan derive the star’s age with an average uncertainty of 25%. In partic-ular, for stars younger than half of their main sequence lifetime the ageis completely unconstrained. This consideration does not include the ad-ditional uncertainty on the metallicity (initial chemical composition). Onthe other hand, for open cluster stars the age is that of the cluster andit can be determined with an uncertainty of 8% or less, regardless of thecluster age. In addition, from the analysis of the chemically normal clusterstars one can derive the initial chemical composition. hemical abundance studies of CP stars in open clusters In most cases, one of the major challenges of studying cluster stars isdetermining their membership. Cluster membership probabilities can bedetermined directly, using photometry and velocities (i.e., proper motionand radial velocity), or indirectly, with age sensitive features (e.g., emis-sion lines and lithium abundance) . We will now consider only the directmethods.Within the uncertainties, stars member of the same cluster lie onthe isochrone corresponding to the cluster age and metallicity. Multicolorphotometry allows one to determine on the color-magnitude diagram thedistance between each observed star and the best fitting cluster isochrone,leading therefore to cluster membership probabilities. It is strongly recom-mended to apply this procedure also to the color-color diagram as somestars will appear as cluster members only in one diagram. More on theimportance of using both color-magnitude and color-color diagrams canbe found in Dias et al. [10].Inhomogeneities of photometric catalogs for the same cluster are afurther problem which can be encountered when determining cluster mem-bership probabilities through photometry. As a matter of fact, looking atthe open cluster photometric data available through the WEBDA database(Mermilliod & Paunzen [28]) one can notice large discrepancies betweenthe magnitudes obtained by different authors on the same cluster, thoughusing the same filters. Figure 1 shows, as an example, the difference be-tween the magnitudes obtained by Sujatha et al. [36] and Colegrove [9] ofstars in the field of view of the young open cluster NGC 1857. Both data-sets were obtained from CCD photometry in the Johnson B and V filters.As shown in Figure 1, the difference is as large as 2 mag and its originis unknown. Other open clusters (e.g., Berkeley 80, NGC 1513, NGC 2112,NGC 6451, NGC 2425, NGC 3590, NGC 7419; Paunzen, E., priv. comm.)show similar discrepancies in the available photometry. For this reason, wesuggest to use always all available photometric references for a membershipprobability determination.Proper motions, particularly for the most nearby clusters, are a verypowerful tool to determine cluster membership probabilities. The left panelof Fig. 2 shows the proper motions (from Kharchenko et al. [20]) of stars Note that parallaxes can also be used to determine cluster membership probabilities,but they are currently available only for a few open clusters (van Leeuwen [26]).
Fossati, L.
Figure 1. Left panel: comparison between the color magnitude diagram, result-ing from CCD photometry with the Johnson B and V filters, of the open clus-ter NGC 1857 as obtained by Sujatha et al. [36] and Colegrove [9]. Data takenfrom the WEBDA database. The continuous line shows the cluster best fittingisochrone, from Bressan et al. [6]. Top-right: comparison between the Johnson V band magnitudes obtained by Sujatha et al. [36] and Colegrove [9] as a functionof the color B − V by Sujatha et al. [36]. Bottom-right: as for the top-right panel,but for the color B − V . lying at an angular separation smaller than 3.5 degrees from the center ofthe Praesepe open cluster. It is evident that the cluster stars have a peculiarproper motion compared to that of the field stars, therefore allowing one todetermine rather precise and unambiguous membership probabilities. Onthe other hand, the majority of the open clusters have a proper motionsimilar to that of the field stars (see for example the right panel of Fig. 2).Radial velocities ( υ r ) can also be used as cluster membership indica-tors, but precise average υ r values are known only for a few clusters. Whenusing this indicator, one should always check how many stars were used todetermine the average cluster υ r and how many times each star has beenobserved, in order to reject the binaries. This is particularly important foryoung cluster, were the brightest stars are very massive (earlier than B5)and therefore likely to be in a binary system (Sana et al. [34]).Given the various issues described above, we suggest to always use allavailable information and tools to determine cluster membership proba-bilities, in a similar fashion to what presented by Kharchenko et al. [21],but including more photometric data sources and carefully checking theaverage cluster υ r against binarity. hemical abundance studies of CP stars in open clusters Figure 2. Left panel: proper motions (from Kharchenko et al. [20]) of stars lyingat an angular separation smaller than 3.5 degrees from the center of the Praesepeopen cluster. Right panel: proper motions of stars lying at an angular separationsmaller than 10 degrees from the center of the open cluster NGC 5460. Clustermembers are marked in red.
2. Elemental abundances of open cluster chemicallypeculiar stars
The standard classification of CP stars is that given by Preston [31] whosubdivided CP stars into four groups: metallic line stars (CP1, otherwisecalled Am stars), magnetic Ap stars (CP2), HgMn stars (CP3), and He-weak stars (CP4). Magnetic fields have been detected and confirmed onlyfor the CP2 and CP4 stars, while the non-magnetic CP stars (CP1 andCP3) usually belong to binary systems and rotate slower than stars witha normal (i.e., close to Solar) chemical composition (Abt [1]).In the following we will review the results obtained from homogeneouschemical abundance studies of CP stars in open clusters, subdividing themon the basis of the chemical peculiarity: metallic line stars (Sect. 2.1.),HgMn stars (Sect. 2.2.), and magnetic CP stars (Sect. 2.3.). Section 2.1.includes also results obtained for open cluster chemically normal stars, inthe context of diffusion.
Metallic line stars (CP1) have masses between about 1.2 and 3.0
M/M ⊙ (early F- and A-type stars) and are characterised mostly by an underabun- Fossati, L. dance of Sc (up to 2 orders of magnitude), and overabundances of Fe-peakand rare-earth elements (up to 4 orders of magnitude - see e.g., Fossati etal. [11]). In addition, they show large microturbulence velocity values, usu-ally around 4 km s − , compared to ∼ km s − typical of chemically normalstars (Landstreet et al. [23]).The first to systematically analyse and compare the abundance pat-tern of CP1 and chemically normal stars in open clusters were Hui-Bon-Hoa et al. [17], Hui-Bon-Hoa & Alecian [18], and Hui-Bon-Hoa [19]. Burkhart& Coupry [7] looked for systematic differences in the lithium abundancebetween open cluster CP1 and chemically normal stars, concluding thatlithium is depleted by 3 times in CP1 stars.Varenne & Monier [39] were instead the first to look for correlationsbetween abundances and stellar parameters. For the chemically normalstars member of the Hyades open cluster, they found an anti-correlationbetween Fe abundance and projected rotational velocity ( υ sin i ) for starswith υ sin i> km s − . A later inspection of the spectra revealed this find-ing was due to a wrong continuum normalisation (Monier R., priv. comm.).This was further confirmed by Gebran et al. [16], who also concluded thatCP1 chemical peculiarities result to be present up to υ sin i ∼ km s − .The absence of correlations between element abundance and υ sin i for bothchemically normal and CP1 stars was also reported by Gebran et al. [15].Fossati et al. [12] analysed a sample of CP1 stars belonging to thePraesepe open cluster. In contrast to the previous results, they obtainedthat the abundances of the elements peculiar in CP1 stars display a signif-icant correlation with υ sin i , with peculiarities decreasing with increasing υ sin i . This result is in agreement with diffusion model calculations byTalon et al. [38]. The reason for the discrepancy between the analysis ofthe different clusters is unknown, though the most realistic explanation isdifferences in the adopted analysis method.Direct comparisons between diffusion models and observed abundancepatterns of open cluster CP1 stars have been presented by Gebran etal. [15], [16], and Vick et al. [40]. The major common conclusion is that dif-fusion models are able to reproduce either the light elements (i.e., atomicnumber .
15) or the heavier elements (i.e., Fe-peak elements). In addition,there is always a significant discrepancy between the observed sodiumabundance in CP1 stars compared to that predicted by diffusion mod-els, with the observed abundance being systematically higher. Modellingsuggests that sodium should not be affected by diffusion, whereas obser- hemical abundance studies of CP stars in open clusters T eff , with an abundance increase for temperatures up to T eff ∼ Mercury-manganese stars (CP3) lie between about 2.5 and 5
M/M ⊙ (lateB-type stars) and are characterised mostly by a large overabundance of Mnand Hg (up to several orders of magnitude). In addition, CP3 stars presenttime-variable surface abundance spots (e.g., Kochukhov et al. [22]).There are no systematic abundance studies of cluster CP3 stars. Themost relevant work is that presented by Wolf & Lambert [41] who dis-covered three CP3 stars in the Orion nebula, placing therefore an obser-vational upper limit of 1.7 Myr on the timescale needed to produce CP3chemical peculiarities. Magnetic CP stars can be found all along the upper main sequence andare characterised by the presence of sometimes complex and strong sur-face magnetic fields and various flavors of abundance anomalies (see e.g.,Ryabchikova [33]).Only very recently systematic abundance studies of cluster magneticstars have been published, but mostly devoted to the analysis of singlestars (e.g., Bailey et al. [3], [4], Bailey & Landstreet [5]). Nevertheless, at
Fossati, L. this conference, John Landstreet showed the first results obtained from theanalysis of several magnetic CP stars in open clusters of different age. Theresults indicate that the chemical peculiarities decrease with increasing age(see Landstreet [24] and Bailey et al. in prep. for more details).
3. Conclusions
The ESA GAIA satellite will soon provide us extremely precise distancesfor most of the stars we observe in the sky. For this reason it is legitimateto ask ourselves the following questions. Do we still need to study opencluster stars after GAIA? Bagnulo et al. [2] concluded that one can derivethe star’s main sequence lifetime with an average uncertainty of 25%: isthis still true after GAIA?To answer these questions we repeat the experiment performed byBagnulo et al. [2] in their Fig. 1 and assuming the expected very best casescenario T eff and log L/L ⊙ uncertainties after GAIA. Although probablyunrealistic, particularly for CP stars, some (e.g., Liu et al. [27]) suggestthat GAIA spectrophotometry would allow one to determine T eff witha precision of 1%. Regarding the luminosity, we can assume that afterGAIA the uncertainty on log L/L ⊙ will be due only to the uncertaintyon the bolometric correction and therefore we consider an uncertainty on log L/L ⊙ of 0.05 dex.Figure 3, to be compared to Fig. 1 of Bagnulo et al. [2], was producedassuming these uncertainties. Even with the very best case scenario, it willnot be possible to determine the age of young field stars with the same pre-cision of that given by open clusters. More importantly only open clusterswill provide the crucial information on the initial chemical composition,impossible to derive for field CP stars. Acknowledgements.
The author thanks Vincenzo Andretta for makingFigure 3. This work made use of the WEBDA database.
References
1. Abt, H. A. 2000, ApJ, 544, 9332. Bagnulo, S., Landstreet, J.D., Mason, E., et al. 2006, A&A, 450, 7773. Bailey, J. D., Landstreet, J. D., Bagnulo, S., et al. 2011, A&A, 535,A25 hemical abundance studies of CP stars in open clusters Figure 3. Position of a star in the HR diagram (left panel), and the star’s positiontransformed into a diagram of age as a function of stellar mass (right panel),assuming one knows T eff ±
1% and luminosity ± Z = 0.02 (Schaller et al. [35]); several fractionalages (fraction of main sequence life completed) are labelled. Figure made byVincenzo Andretta.
4. Bailey, J. D., Grunhut, J., Shultz, M., et al. 2012, MNRAS, 423, 3285. Bailey, J. D. & Landstreet, J. D. 2013, MNRAS, 432, 16876. Bressan, A., Marigo, P., Girardi, L., et al. 2012, MNRAS, 427, 1277. Burkhart, C. & Coupry, M. F. 2000, A&A, 354, 2168. Charbonneau, P. & Michaud, G. 1991, ApJ, 370, 6939. Colegrove, P. T. 1999, Open cluster ages and an apparent periodicityin the history of star formation in the solar neighborhood, PhD thesis,University of Nevada, Reno10. Dias, W. S., Monteiro, H., Caetano, T. C., et al. 2012, A&A, 539, A12511. Fossati, L., Bagnulo, S., Monier, R., et al. 2007, A&A, 476, 91112. Fossati, L., Bagnulo, S. & Landstreet, J. D., et al. 2008, A&A, 483,89113. Fossati, L., Folsom, C. P., Bagnulo, S., et al. 2011, MNRAS, 413, 113214. Gebran, M. & Monier, R. 2008, A&A, 483, 5670