Chemical composition of A and F dwarf members of the Coma Berenices open cluster
aa r X i v : . [ a s t r o - ph ] D ec Astronomy&Astrophysicsmanuscript no. gebran˙8807 c (cid:13)
ESO 2018November 9, 2018
Chemical composition of A and F dwarf members of the ComaBerenices open cluster ⋆ M. Gebran , R. Monier ⋆⋆ and O. Richard Groupe de Recherche en Astronomie et Astrophysique du Languedoc,UMR 5024, Universit´e Montpellier II, Place Eug`eneBataillon, 34095 Montpellier, France.e-mail: [email protected] e-mail:
[email protected] e-mail: [email protected]
Received ; accepted
ABSTRACT
Aims.
Abundances of 18 chemical elements have been derived for 11 A (normal and Am) and 11 F dwarfs members of the ComaBerenices open cluster in order to set constraints on evolutionary models including transport processes (radiative and turbulent di ff u-sion) calculated with the Montr´eal code. Methods.
A spectral synthesis iterative procedure has been applied to derive the abundances from selected high quality lines in highresolution high signal-to-noise ´echelle spectra obtained with ELODIE at the Observatoire de Haute Provence.
Results.
The chemical pattern found for the A and F dwarfs in Coma Berenices is reminiscent of that found in the Hyades and theUMa moving group. In graphs representing the abundances [X / H] versus the e ff ective temperature, the A stars often display abun-dances much more scattered around their mean values than the F stars do. Large star-to-star variations are detected for A stars in theirabundances of C, O, Na, Sc, Ti, Mn, Fe, Ni, Sr, Y, Zr and Ba which we interpret as evidence of transport processes competing withradiative di ff usion.The abundances of Mn, Ni, Sr and Ba are strongly correlated with that of iron for A and Am stars. In contrast the ratios [C / Fe] and[O / Fe] appear to be anticorrelated with [Fe / H] as found earlier for field A dwarfs. All Am stars in Coma Berenices are deficient inC and O and overabundant in elements heavier than Fe but not all are deficient in calcium and / or scandium. The F stars have solarabundances for almost all elements except for Mg, Si, V and Ba.The derived abundances patterns, [X / H] versus atomic number, for the slow rotator HD108642 (A2m) and the moderately fast rotatorHD106887 (A4m) were compared to the predictions of self consistent evolutionary model codes including radiative and di ff erentamounts of turbulent di ff usion. None of the models reproduces entirely the overall shape of the abundance pattern. Conclusions.
While part of the discrepancies between derived and predicted abundances may be accounted for by non-LTE e ff ects,the inclusion of competing processes such as rotational mixing in the radiative zones of these stars seems necessary to improve theagreement between observed and predicted abundance patterns. Key words. stars: abundances - stars: main sequence - stars: rotation - Di ff usion - Galaxy: open clusters and associations: individual:Coma Berenices
1. Introduction
Abundance determinations of A and F dwarfs in open clustersand moving groups of known properties aim at elucidatingthe mecanisms of mixing at play in the interiors of thesemain-sequence stars. In two previous papers, abundancedeterminations were presented for 11 chemical elements inthe Hyades (age about 787 Myrs, Varenne & Monier 1999)and in the Ursa Major moving group (age about 500 Myr,Monier 2005). Specifically, 19 A and 29 F dwarfs members ofthe Hyades and 12 A and 10 F dwarf bona-fide and probablemembers of the Ursa Major moving group were analysed.The motivation of this study is to report on new abundance
Send o ff print requests to : M. Gebran ⋆ Based on observations at the Observatoire de Haute-Provence(France). ⋆⋆ Present a ffi liation: Laboratoire Universitaire d’Astrophysique deNice, UMR 6525, Universit´e de Nice - Sophia Antipolis, Parc Valrose,06108 Nice Cedex 2, France determinations of 18 chemical elements in 11 A and 11 F dwarfsin the Coma Berenices open cluster (age about 447 Myrs). Thespectroscopy presented here is part of an ongoing observationalprogram of A and F dwarfs in galactic open clusters of variousages whose aims are twofold. First, we wish to improve ourknowledge of the chemical composition of A dwarfs (normal Astars and Am stars) which is still poor in particular for normalA stars. Second, we intend to use the derived abundances toset constraints on self-consistent evolutionary models of theseobjects including various particle transport processes. Indeedstars in open clusters originate from the same interstellar ma-terial ( ie they have the same age and the same initial chemicalcomposition) and as such are very useful to test the predictionsof evolutionary models.Few studies have addressed the chemical composition of theA and F dwarfs in the Coma Berenices open cluster. Highquality high resolution spectroscopy is feasible for the brightestmembers because of the proximity (d ≃ − M. Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster focused on F and G dwarfs whose low apparent rotationalvelocities facilitate the abundance determination. For concision,we have collected previous abundance determinations of the A,F and G dwarfs in Coma Berenices in Table 1 where the biblio-graphical references, the number of stars analysed, their spectraltypes and the investigated chemical elements are collected.Lithium abundances have been determined for several F (and G)dwarfs in Coma Berenices by Boesgaard (1987), Je ff ries (1999)and Soderblom et al. (1990). Carbon and iron abundanceswere derived for 14 F dwarfs by Friel & Boesgaard (1992).Carbon, oxygen, silicon, calcium, iron, barium, magnesium,scandium, chromium, nickel, lithium, aluminium, sulfur andeuropium abundances have been determined for only a fewnormal A and Am stars (fewer than 7 stars) by Savanov (1996),Hui-Bon-Hoa et al. (1997), Hui-Bon-Hoa & Alecian (1998)and Burkhart & Coupry (2000). The abundance determinationsfor A stars focuse mainly on the chemically peculiar Am stars.Savanov (1996) found significant star-to-star di ff erences inabundances among the Am stars of Coma Ber for a givenchemical element. In contrast, little attention has been paid tothe ”normal” A stars of the cluster. The chemical compositionof normal A dwarfs (field and cluster stars) remains generallypoorly known as too few objects have been analysed so far,mainly because of their high rotational velocities. Significantabundance di ff erences have been found among the few field”normal” A stars analysed so far (Holweger et al. 1986;Lambert et al. 1986; Lemke 1998,1990; Hill & Landstreet 1993;Hill 1995; Rentzsch-Holm 1997; Varenne 1999). Varenne &Monier (1999) also found significant star-to-star variations ofthe abundances of O, Na, Ni, Y and Ba for the normal A and theAm stars in the Hyades whereas the F dwarfs display much lessscatter. Similarly, Monier (2005) found star-to-star variations in[Fe / H], [Ni / H] and [Si / H] much larger for the A dwarfs than forthe F dwarfs in the Ursa Major group.The main thrust of this paper is to report on the abundancesof 18 chemical elements (C, O, Na, Mg, Si, Ca, Sc, Ti, V,Cr, Mn, Fe, Co, Ni, Sr, Y, Zr and Ba) for F and A dwarfs inComa Berenices. These abundances have been compared topublished predictions of recent self consistent models includingtransport processes at ages close to that of Coma Berenices(Turcotte et al. 1998, Richer et al. 2000, Richard et al. 2002)or to new models calculated with the Montr´eal code (Richardet al. 2002). The abundances were derived by synthesizingcarefully selected lines in high quality high resolution spectraof 11 A and Am stars and 11 F stars of the cluster. The selectionof the sample of stars observed, the observations and the datareduction are described in Section 2. The determination of thefundamental parameters, the construction of the linelist and thespectrum synthesis are discussed in Section 3. The behaviourof the abundances of individual chemical elements in the Aand F dwarfs of Coma Berenices versus e ff ective temperatureand the abundance of iron are described in Section 4. Theastrophysical implications of our findings and the detailedcomparison of the found abundance patterns for 2 Am stars withrecent self-consistent evolutionary models including transportprocesses are presented in Section 5. Conclusions are given inSection 6.
2. Program stars, observations and data reduction
Our observing sample consists of all A and F stars members ofthe Coma Berenices cluster brighter than V = ff ects of scat-tered light on the line profiles in the blue region obtained withELODIE. ELODIE is a fiber-fed cross-dispersed ´echelle spec-trograph attached to 193 cm at OHP (Baranne et al. 1996). Itrecords in a single exposure a spectrum extending from 3850 Åto 6811 Å at a resolving power of about 42000 on a relativelysmall CCD (1024X1024). The observing dates, exposure timesand Signal to Noise ratios achieved for each star are collected inTable 3, the first part being dedicated to ELODIE spectra whilethe second describes the AURELIE spectra obtained in three 70Å wide regions centered around 4505 Å, 5080 Å, 5530 Å and6160 Å.The fundamental data of these stars are collected in Table2. The Trumpler and Henry Draper identifications appear incolumns 1 and 2, the spectral types retrieved from SIMBAD incolumn 3, the apparent magnitudes in column 4. Columns 5 to 8display the e ff ective temperatures and surface gravities adoptedfor the analysis and the rotational velocities and microturbulentvelocities derived in our analysis (see section 3.2.1). Commentsabout binarity and pulsation appear in the last column. Note that2 stars, HD 107655 and HD 106999, may not be members ofthe cluster (Bounatiro 1993). Although we derived abundancesfor these stars too, their data do not appear in the figures. Thereare no very rapid rotators in this cluster, the apparent rotationalvelocities range from 9 km · s − to 102 km · s − .Inspection of the Catalogue of Double and Multiple stars(CCDM, Dommanget & Nys 1995) reveals that only one star is abinary system. HD 106887 is the primary star in a double systemand has a much fainter companion (V = λ λ λ λ λ ≤ > = ff set removal, division by the mean flat field, wavelengthcalibration and continuum normalization).At the end of an observing night, the ELODIE spectraare provided to the observer fully reduced by the INTER-TACOS (INTERpreter for the Treatment, the Analysis and the . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 3 Table 1.
Previous abundance determinations for the Coma Ber A, F and G dwarfs.
Reference Stars studied Chemical ElementsSavanov (1996) 13 A-Am and F-Fm C,O,Si,Ca,Fe,BaHui-Bon-Hoa et al. (1997) 2 A-Am Mg,Ca,Sc,Cr,Fe,NiHui-Bon-Hoa & Alecian (1998) 4 A-Am Mg,Ca,Sc,Cr,Fe,NiBurkhart & Coupry (2000) 7 A-Am Li,Al,Si,S,Fe,Ni,EuBoesgaard (1987) 22 A and F LiFriel & Boesgaard (1992) 14 F Fe,CJe ff ries (1999) 15 F,G,K LiCayrel et al. (1988) 4 G FeSoderblom et al. (1990) 28 G Li Table 2.
Data on the programme stars. Spectral type are taken from SIMBAD and WEBDA online database. T e ff and log g arethose determined by UVBYBETA code. v e sin i and ξ t are determined as explained in sect. 3.2.1. References (a) and (b) areBounatiro (1993) and Rodriguez et al. (1994) respectively. TR HD Type m v T e ff log g v e sin i ξ t Remarques(K) km · s − km · s −
19 HD106103 F5V 8.09 6707 4.45 19.7 1.4HD106293 F5V 8.09 6545 4.34 47 1.436 HD106691 F5IV 8.08 6713 4.43 37 1.649 HD106946 F2V 7.87 6892 4.30 62 1.986 HD107611 F6V 8.50 6491 4.57 22 1.4HD109530 F2V 7.30 6497 3.85 67 2.1101 HD107877 F6 8.35 6598 4.54 29.5 1.25114 HD108154 F5 8.56 6497 4.54 19 1.15118 HD108226 F5 8.34 6530 4.45 19 1.05162 HD108976 F6 V 8.54 6413 4.49 20 1.1HD109069 F0 V 7.55 6864 4.06 89 2.2107 HD107966 A3V / A3IV 5.18 8541 3.82 51 2.9130 HD108382 A4V / A3IV 4.96 8317 3.92 75.5 3.247 HD106887 A4m 5.71 8291 4.20 82 3.886 HD107655 A0V 6.18 9675 4.10 45 2.0 < Coma (a)62 HD107168 A8m / kA5hA5mF0 6.24 8283 4.20 14.3 4.0183 HD109307 A4Vm / A3IV-V 6.26 8396 4.10 14.5 3.3144 HD108642 A2m / kA2hA7mA7 6.54 8079 4.06 9.2 3.768 HD107276 Am / Ka5mA7 6.63 8000 4.00 102 2.8139 HD108486 AmkA3hA5mA7 6.67 8148 4.11 37 3.052 HD106999 Am 7.46 8148 4.09 44 3.0 < Coma (a)82 HD107513 Am / kA7hF0mF0 7.38 7279 4.02 62 3.0 δ Scuti (b)
COrrelation of Spectra) pipeline developed by D. Queloz and L.Weber (Baranne et al. 1996). The reduction is actually includedin the spectrograph data flow and provides fully flat-fielded andwavelength calibrated spectra directly at the end of each expo-sure. However, we have chosen to perform our own reduction ofthe ELODIE spectra. Indeed, ELODIE was primarily designedto provide accurate radial velocity measurements. One sourceof concern when deriving abundances is to properly correct thespectrum for scattered light, especially in the blue region. Thebackground in a stellar ELODIE exposure can be estimated bymeasuring the flux in the inter-orders. In INTER-TACOS, thisbackground is removed using a two dimensional polynomial fitwith a typical 5 % error which peaks in the middle of the orders(see fig. 11 in Baranne et al. 1996). Erspamer & North (2002)have devised a reduction procedure using a set of IRAF func-tions which they apply to the raw image in order to providean improved correction for scattered light compared to INTER-TACOS. We have also chosen to perform our own reduction of the ELODIE spectra using IRAF (Image Reduction and AnalysisFacility, Tody 1993) routines. Although it follows Erspamer &North’s (2002) procedure, our reduction slightly di ff ers fromtheir method. The steps are as follows:1- Averaging the several o ff sets and flat-fields taken throughoutthe night, using zerocombine and flatcombine .2- Removal of the mean o ff set from all images and bad pixelscorrection using ccdproc .3- Finding and centering the orders using the flat-field imageusing apfind and apcenter .4- Removal of the scattered light ( apscatter ). Scattered lightfills in the line profiles, making the lines shallower andthus leading to underabundances if not taken into account.The scattered light is estimated (as explained in sect 3.2 ofErspamer & North 2002) and substracted from the originalimage. An example of the e ff ect of this improved removal ofthe scattered light is shown in Figure 2, where the correctedprofiles of the FeII lines at 4520.224 Å and 4522.634 Å are M. Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster compared to the INTER-TACOS profiles. In this case, ignor-ing the scattered light would lead to underestimation of theiron abundance deduced from these 2 lines by 0.07 and 0.08dex respectively. The e ff ect is more pronounced, about 0.14dex, for the MgII triplet at 4481 Å.5- Extraction of the images using apsum . The averaged flat-field is used to determine the shape of the orders which isused later as a reference for the extraction of the images.The extraction method uses Horne’s (1986) algorithm.6- Calibration of the thorium image using apsum , ecidentify and ecreidentify .7- Division by the extracted flat-field using sarith .8- Wavelength calibration of the spectra using the thorium spec-tra using dispcor .9- Normalization to the continuum using continuum . To ensurewe correctly located regions free of lines (when available) ineach order, we have computed synthetic spectra using thecode SYNSPEC48 (Hubeny & Lanz 1992) assuming a solarmetallicity for the various temperatures and surface gravitiesof our stars. The spectrum was then rectified to this localcontinuum.10- Merging the 67 normalized orders using scombine . The lastthree orders do not overlap and thus could not be merged.For the abundance analysis, we have discarded lines locatedin the overlapping region of two successive orders.11- Radial velocity determination using fxcor : the final mergedspectrum is cross-correlated with di ff erents masks of spectraltypes A0V, A5V, A9V and F5V to derive the radial velocity.The merged spectrum is then corrected for the radial velocityfound.Figure 3 compares the spectral order 21, centered around λ ff ect ofincreasing stellar rotation on line profiles is conspicuous. Thelast plot of Figure 3 displays a typical agreement between theobserved spectrum of HD 107655 (A0V) and the synthetic spec-trum (computed as explained in sect. 3) that provides the bestfit.
3. Abundance analysis
The abundances of 18 chemical elements have been derived byiteratively adjusting synthetic spectra to the observed normalizedspectra and minimizing the chisquare of the models to the ob-servations. Spectrum synthesis is the most appropriate methodas our stars have apparent rotational velocities ranging from a9 to 102 km · s − . Specifically, synthetic spectra were computedassuming LTE using Takeda’s (1995) iterative procedure anddouble-checked using Hubeny & Lanz’s (1992) SYNSPEC48code. This version of SYNSPEC calculates lines for elementsheavier than Zn up to Z = The e ff ective temperatures ( T e ff ) and surface gravi-ties (log g ) of the stars have been determined using theNapiwotzki et al. (1993) UVBYBETA calibration of theStr¨omgren photomery indices uvby in terms of T e ff and log g .The found e ff ective temperatures and surface gravities arecollected in Table 2. The errors on T e ff and log g are estimated tobe ±
125 K and ± Fig. 1.
Hertzsprung-Russel diagram of the observed stars inComa Berenices (visual magnitude versus T e ff ). The two starsflagged with question marks are probably not members of thecluster. FeII (0.08 dex)FeII (0.07 dex)
Fig. 2. E ff ect of a proper removal of scattered light. The cor-rected profiles (IRAF reduction, thick lines) are deeper than theINTER-TACOS ones (dashed lines). The abundances deducedfrom the corrected profiles are about 0.08 dex larger.contain 64 layers with a regular increase in log τ Ross = . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 5 λ (A)-0.400.40.81.21.622.42.8 N o r m a li ze d f l ux HD107168 (A8m) V e sini=14.3 km/s HD108486 (Am) V e sini=37 km/s HD106887 (A4m) V e sini=82 km/s HD107276 (Am) V e sini=102 km/sFeI FeI FeIFeIMgII TiII FeII FeII FeII FeIIFeIITiII TiII + CaIIFeI
A stars, Coma Ber λ (A)00.511.522.53 N o r m a li ze d f l ux HD106946 (F2V) V e sini=62 km/s HD106293 (F5V) V e sini=47 km/s HD106691 (F5IV) V e sini=37 km/s HD106103 (F5V) V e sini=19.7 km/sFeI FeI FeI FeI FeI FeIFeIFeIMnI FeI FeI FeIMgII TiII FeIITiI CrIZrII MnI TiIIMnI FeII TiI FeII TiI TiIIFeII FeIILaIITiI F stars, Coma Ber λ (A)0.60.811.2 N o r m a li ze d f l ux Observed spectrumSynthetic spectrum
HD107655 (AOV) V e sini=45 km/s FeI MgII TiIICaIIFeII FeII FeIIFeI TiII FeII FeII FeII FeII
Fig. 3.
Selected observed spectra of A (top) and F (middle) dwarf stars members of Coma Berenices open cluster. Spectra arearbitrarily shifted vertically by 0.5, 1 and 1.5 unit of normalized flux. The smearing out of the spectra by rotation is noticeable.The bottom figure displays the final synthetic spectrum (dashed thick line) superimposed on the observed one (thin line) for the starHD107655 (A0V). Identifications for the most intense lines in each region are provided.
M. Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster
Table 3.
Observing log of the programme stars of Coma Berenices. The first table describes the ELODIE observations and thesecond one the AURELIE observations
HD spectral M V exposure S / N Datetype mag time (s)106103 F5V 8.09 4500 144 04 / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / M V grating 5: λ c = λ c = λ c = λ c = / /
04 75 104 120 (grating 1)107513 Am 7.38 2000 03 / /
04 120 120 120106103 F5V 8.09 4500 03 / /
04 60107655 A0V 6.18 800 03 / /
04 135 85 90106887 A4m 5.71 600 03 / /
04 60107131 A6IV-V 6.44 800 03 / /
04 110107276 Am 6.63 1000 03 / /
04 90 120 75107966 A3V 5.18 500 03 / /
04 45 105 40108382 A4V 4.96 500 03 / /
04 60 74108486 Am 6.67 1000 03 / /
04 180 90108642 A2m 6.54 1000 03 / /
04 90 120108651 A0p 6.65 1000 03 / /
04 105 150109307 A4Vm 6.26 800 03 / /
04 120 120 (grating 1) 80106293 F5V 8.09 4500 03 / /
04 130106691 F5IV 8.08 4500 03 / /
04 120 150 120109069 F0V 7.55 3000 03 / /
04 100 150 110107877 F6 8.35 4500 03 / /
04 180 165107611 F6V 8.50 4500 03 / /
04 90108154 F5 8.56 4500 03 / /
04 180 165108226 F5 8.34 4500 03 / /
04 90 180108976 F6V 8.54 4500 03 / /
04 150 140109530 F2V 7.30 3000 03 / /
04 100 120 110 ratio of the mixing length to the pressure scale height ( α = LH P )and also for the microturbulent velocity (constant with depth). The linelist was constructed from Kurucz’s gfall.dat list, fromwhich we selected lines between 3000 and 7000 Å. However thedata in this list have been modified and complemented in di ff er-ent manners. For the lines which we expected to contribute sig-nificantly to the absorption, we have carefully checked the wave-lengths, lower excitation potential, oscillator strength and damp-ing constants (radiative, Stark and Van der Waals) in gfall.datagainst more accurate and / or more recent critically evalu-ated laboratory determinations when available. Specifically, twoatomic databases were searched for improved values of these http: // kurucz.harvard.edu / LINELISTS / GFALL / parameters and their uncertainties: the VALD database and theNIST database. We then modified their values in the originallinelist accordingly. Damping constants not available in linelistsare calculated in SYNSPEC48 using approximations. We havealso excluded lines in the overlaping regions of two succes-sive orders, as we felt that the observed line profiles may notbe reliable there. The final linelist contains 270 transitions for18 elements which we believe are reliable enough to derive theabundances. Most of the lines studied here are weak lines whichare formed deep in the atmosphere where LTE should prevail.They are well suited for abundance determinations. The finallinelist appears in Table 8 where, for each element, the wave-length, adopted oscillator strength, its accuracy (when available)and original bibliographical reference are given. We have alsoincluded data for hyperfine splitting for the selected transitions http: // ams.astro.univie.ac.at / vald / , Kupka et al. 1999 http: // physics.nist.gov / cgi-bin / AtData / lines-form. Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 7 when relevant, in particular for Mn II (these were retrieved fromthe linelist gfhyperall.dat ). However the moderate spectral res-olution of the spectra and smearing out of spectra by stellar ro-tation clearly prevent us from detecting signatures of hyperfinesplitting and isotopic shifts in our spectra. Two codes have been used to derive abundances: Takeda’s code(Takeda 1995) and SYNSPEC (Hubeny & Lanz 1992) to checkthe abundances produced by Takeda’s iterative procedure. Webriefly review the assumptions of these codes in section 3.2.1and section 3.2.2.
Takeda’s procedure iteratively minimizes the dispersion σ be-tween the normalized synthetic spectrum and the observed onedefined as: σ = P Ni = y i − η i − C ) N where N is the number of wavelength points, y i the logarith-mic of the observed spectra ( f λ i ) and η i the logarithmic of thesynthetic spectra ( F λ i ). C is an o ff set constant reflecting a pos-sible di ff erence of units between F λ and f λ (it should be veryclose to zero when working with normalized fluxes). The syn-thetic flux η i in a line is a priori a function of several physicalvariables ( x , ..... x K ) which represent the unknowns one might belooking for: abundances of individual elements, projected rota-tional velocity and microturbulent velocity, oscillator strengths,damping constants. The dispersion σ is thus a function of K + σ thus requires that itspartial derivatives with respect to these K + + + x k s become su ffi ciently small, typically less than 10 − .Takeda’s code consists of two routines: the first is a modifiedversion of Kurucz’s Width9 code (Kurucz 1992a) for computingthe opacity data which needs a Kurucz’s model atmosphere asinput. Once the opacities are computed, the second routine com-putes the emergent flux and minimizes the dispersion betweenthe synthetic and the observed spectra.In this study, we kept the number of free variables at 3. At eachiteration, the code outputs 3 parameters: the abundance(s) of thestudied element(s) (log ǫ ), the microturbulent velocity ( ξ t ) con-stant whith depth and the projected rotational velocity ( v e sin i ),the oscillator strengths and damping constants being kept fixed.We usually synthesized unblended lines of one chemical elementonly, although the procedure allows us to simultaneously synthe-size lines of di ff erent elements.Before tackling the abundance of each individual chemical el-ement, we derived the rotational and microturbulent velocitiesfor each star as follows. Allowing small variations around solarabundances of Mg and Fe, we iteratively fitted the unblendedline profile of the MgII triplet at 4480 Å and a set of neigh-boring unblended weak and moderately strong FeII lines such as4491.405 Å and 4508.288 Å leaving ξ t and v e sin i as free param-eters. Even in the fastest rotators, the continuum is well seen inthis spectral region where there are only a few regularly spaced http: // kurucz.harvard.edu / LINELISTS / GFHYPERALL / λ (Α) F l ux Iteration-1Iteration-2iteration-3iteration-4iteration-5iteration-6Observed spectrum
FeII
HD107966 (A3V)
Fig. 4.
Iterative adjustment of synthetic profiles of the FeII line at4508.288 Å for the A3V star HD107966 with Takeda’s program.Convergence was achieved properly at the fifth iteration.lines. The weakest FeII lines are sensitive to rotational veloc-ity and not to microturbulent velocity, the moderately strongFeII lines respond to microturbulent velocity changes. The MgIItriplet, commonly used to derive stellar rotational velocities, ap-pears to be sensitive to both ξ t and v e sin i . Each line of Fe andMg yielded a set of values for log ǫ , ξ t and v e sin i , which werein good agreement. For the F stars, the found ξ t were checkedagainst Nissen’s (1981) prediction which invoked a polynomialfit for ξ t as a function of T e ff and log g for the F stars. The agree-ment is generally good, the maximum di ff erence we found being ∆ ξ t = .
32 km · s − . Morever, the derived v e sin i in this analysiscorrelate well with those derived from AURELIE spectra of theregion around 4500 Å published in Monier & Richard (2004).For the slowest rotating Am stars, curves of growths of the FeII lines were also performed. They led to values of ξ t consistentwith those derived from the fit of the Mg II triplet and Fe II linesaround 4500 Å. Once ξ t and v e sin i were fixed, we then pro-ceeded to determine the abundance for each selected unblendedline of each chemical element. In practice, complete conver-gence was reached for each star after up to 10 iterations (fivein the most favorable cases).A final test on Procyon was also performed, assuming a so-lar composition (Ste ff en 1985). The iterative fitting of the MgII and Fe II lines around 4500 Å yielded an apparent rota-tional velocity of 6 km · s − and a microturbulent velocity of 2.2km · s − in good agreement with Ste ff en’s (Ste ff en 1985) values( v e sin i = · s − and ξ t = · s − respectively).Figure 4 shows an example of the iterative fitting of the FeII lineat 4508,288 Å for the A3V star HD 107966 (6 iterations). Forthis star, the convergence was already achieved at the fifth itera-tion. We have also used Version 48 of SYNSPEC(Hubeny & Lanz 1992) to check the abundances producedby Takeda’s iterative procedure. SYNSPEC48 allows us tocalculate line profiles of elements up to Z =
99. In its LTE mode,SYNSPEC needs a model atmosphere and a linelist plus anauxiliary file containing non standard flags. For a given star, thesame ATLAS9 models, the same linelist and the v e sin i and ξ t derived using Takeda’s (1995) procedure were used to calculate M. Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster
10 20 30 40Z-2-1.5-1-0.500.511.5 [ X / H ] ATLAS12 solarATLAS12 metATLAS9 solar
CI OI MgII SiII CaII ScIITiII CrIIMnIFeII NiI YII ZrIINaI
HD107168 (A8m)
Fig. 5.
Influence of the underlying model atmosphere on the de-rived abundances for HD107168 (A8m): black dots are abun-dances derived using a solar ATLAS9 model, hatched blacksquares using a solar ATLAS12 model and black triangles areabundances derived with a ATLAS12 model computed for thespecific chemical composition of this star.synthetic spectra. Only the abundance of the studied unblendedline were left as free parameters. The derived abundances withSYNSPEC were found to always agree with those derived fromTakeda’s procedure within the error bars. The abundances listedin Tables 4 and 5 are those derived from Takeda’s procedure.We also checked the influence of the underlying atmo-spheric structure on the derived abundances using ATLAS12(Kurucz 2005). Abundances can be adjusted individually inATLAS12 which employs the Opacity Sampling techniquefor line opacity. The e ff ect should be noticeable for the Amstars whose abundances depart most from the solar values. ForAm stars we found that the inclusion of an ALTAS12 modelcalculated for the specific chemical composition found withATLAS9 and Takeda’s procedure yields new mean abundanceswhich di ff er by up to 0.08 dex from those derived with ATLAS9.This is less than the error bar on mean abundances. Figure 5compares the abundances of several elements for the A8mstar HD107168 derived using ATLAS9 and ATLAS12 modelatmospheres. For the other stars, whose chemical compositionsdepart less from solar, the e ff ect will be even smaller. Since weintend to model all stars in a uniform manner and because ofthe much larger computing time needed to run an ATLAS12model atmosphere, we decided to use only ATLAS9 modelatmospheres.
4. Results and uncertainties
Abundances for the 22 stars are collected in Table 4 for A starsand in Table 5 for F stars. These abundances are relative to thesun ([ XH ] = log( XH ) ⋆ − log( XH ) ⊙ ) . They are weighted means ofthe abundances derived from each transition. The calculationof their uncertainties and the weighted mean abundances isexplained in Appendix A. Solar abundances are those from Grevesse & Sauval (1998).
Graphs where abundances are displayed against atomic number(abundance pattern plots) are particularly appropriate to com-pare the behaviour of the A, Am and F stars for di ff erent chem-ical elements. The abundance patterns for the A stars, Am starsand F stars are displayed respectively in Figures 6a, 6b and 6c.The seven Am stars of the clusters display a characteristic jig-saw pattern with much larger excursions from the solar compo-sitions than the normal A stars do. This trend had already beenfound by Hui-Bon-Hoa & Alecian (1998) who derived the abun-dances of Mg, Ca, Sc, Cr, Fe and Ni in four Am stars of oursample (HD107168, HD107966, HD108486 and HD109307).Our abundances for these elements agree well with theirs forHD 107168, for which all fundamental parameters are simi-lar in both studies. However, slight to moderately large dif-ferences occur for HD108486, HD109307 and HD107966, forwhich we adopted larger microturbulent velocities (in particu-lar for HD107966), all other parameters being consistent in bothstudies. The 2 normal A stars of our sample, HD107966 andHD108382, have very similar abundances, nearly solar, in al-most all elements and thus exhibit almost the same abundancepattern (Figure 6a). Inspection of Figure 6b reveals that gener-ally, for almost all chemical elements, Am stars display star-to-star variations that can be larger than the typical uncertainty. AllAm stars are deficient in C and O, but not all are deficient in Caand / or Sc and most have pronounced overabundances of iron-peak and heavy elements. The two normal A stars have almostsolar abundances in most elements. The F stars definitely exhibitless scatter than the Am stars. They tend to be mildly overabun-dant, in particular in Mg, Si, V and Ba. They have nearly solarabundances for C, O, Na, Ti, Fe, Ni, Sr and Y. In this section, we use the found abundances for each elementto address two issues. First, do the abundances depend on stel-lar parameters such as the e ff ective temperature and v e sin i ?Any such correlation could be very valuable for theorists inves-tigating the various hydrodynamical mechanisms a ff ecting pho-tospheric abundances. Second, how do the abundances of eachelement vary with respect to each other? Given that iron is as-trophysically one of the most important elements (since it pro-vides a rough estimate of the ”metallicity”) and given that theabundances derived for this element are probably the most re-liable, we have examined whether the abundances of individualelements correlate with those of iron. Hill (1995) and Lemke(1989, 1990) also looked for similar correlations and we will re-fer to their findings later.We can roughly separate the elements we studied into twogroups. For most elements (Fe II , Ti II, OI, Cr II, Mg II, MnI, C I, Si II, Ca II and Ni I), we synthesized several lines ofquality A to D and we can be confident in the abundances wederived, in particular for iron, titanium and chromium. For ele-ments with many lines, errors in individual oscillator strengthsshould tend to cancel out. For other elements, we have very fewlines (Sr II), so that errors on oscillator strengths may inducescatter in the abundances. For V II, Co I, Y II and Zr II, sev-eral lines are available but their accuracy is not necessarily wellknown. The abundances of these elements should be viewed withmore caution However, our main goal here is to investigate howthe abundances of individual elements vary with that of iron andalso to study star-to-star variations of fundamental parameters. . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 9 Table 4.
Abundances relative to hydrogen and to the solar value, [ XH ] = log( XH ) ⋆ − log( XH ) ⊙ for the A stars. The solar values are thoseof Grevesse & Sauval (1998). The HD numbers in italics are those for which the uncertainties have been calculated as explained inAppendix A. For the others, the quantities labeled as σ are standard deviations. HD SpT CI σ C OI σ O NaI σ Na MgII σ Mg SiII σ Si HD107966
A3V / A3IV -0.07 0.04 0.00 0.05 0.52 0.13 0.13 0.09 0.14 0.09HD108382 A4V / A3IV -0.71 0.15 -0.02 0.06 0.24 0.18 0.13 0.12 0.18 0.19HD106887 A4m -0.56 0.15 -0.43 0.06 0.54 0.08 0.17 0.13 0.23 0.20
HD107655
A0V -0.75 0.19 -0.58 0.06 0.55 0.21 -0.12 0.18 0.06 0.07
HD107168
A8m / kA5hA5mF0 -0.65 0.08 -0.40 0.04 1.06 0.03 0.49 0.18 0.77 0.12HD109307 A4Vm / A3IV-V -0.36 0.01 -0.31 0.12 0.65 0.45 -0.07 0.29 0.23 0.18HD108642 A2m / kA2hA7mA7 -0.75 0.16 -0.85 0.12 0.13 0.14 0.25 0.19 0.09 0.09HD107276 Am / Ka5mA7 -0.13 0.27 -0.08 0.04 0.51 0.08 0.25 0.18 0.21 0.18HD108486 AmkA3hA5mA7 -0.69 0.20 -0.86 0.04 0.60 0.34 -0.02 0.18 0.02 0.21HD106999 Am -0.15 0.10 -0.02 0.04 1.04 0.11 0.40 0.18 0.26 0.09
HD107513 Am / kA7hF0mF0 -0.36 0.10 -0.15 0.12 0.10 0.10 0.22 0.16 0.12 0.12HD SpT CaII σ Ca ScII σ Sc TiII σ Ti VII σ V CrII σ Cr HD107966
A3V / A3IV -0.08 0.08 -0.13 0.08 -0.10 0.05 - - 0.08 0.09HD108382 A4V / A3IV -0.19 0.04 0.12 0.26 -0.09 0.17 - - 0.24 0.25HD106887 A4m -0.11 0.21 0.03 0.34 0.12 0.15 - - 0.13 0.12
HD107655
A0V - - -0.20 0.11 0.02 0.08 - - 0.56 0.09
HD107168
A8m / kA5hA5mF0 0.23 0.16 -0.18 0.10 0.48 0.08 - - 0.48 0.20HD109307 A4Vm / A3IV-V 0.14 0.06 0.10 0.11 -0.07 0.10 - - 0.03 0.05HD108642 A2m / kA2hA7mA7 -0.33 0.18 -1.17 0.20 -0.18 0.07 0.86 0.13 0.18 0.05HD107276 Am / Ka5mA7 -0.24 0.16 -0.06 0.23 0.04 0.22 0.90 0.10 -0.08 0.11HD108486 AmkA3hA5mA7 -0.22 0.15 -0.50 0.31 -0.20 0.13 0.70 0.29 0.08 0.15HD106999 Am 0.10 0.16 0.41 0.27 0.14 0.21 0.69 0.29 0.03 0.20
HD107513 Am / kA7hF0mF0 - - -0.25 0.10 0.01 0.07 0.65 0.32 0.09 0.09HD SpT MnI σ Mn FeII σ Fe Co σ Co NiI σ Ni SrII σ Sr HD107966
A3V / A3IV -0.35 0.18 -0.13 0.05 0.65 0.38 -0.18 0.07 -0.28 0.33HD108382 A4V / A3IV -0.55 0.12 -0.14 0.10 - - -0.14 0.16 -0.26 0.02HD106887 A4m 0.02 0.21 0.21 0.15 - - 0.37 0.11 0.59 0.08
HD107655
A0V - - 0.08 0.05 - - 0.77 0.23 -0.23 0.20
HD107168
A8m / kA5hA5mF0 0.19 0.12 0.39 0.10 - - 0.60 0.10 0.79 0.31HD109307 A4Vm / A3IV-V -0.04 0.17 0.05 0.09 -0.06 0.35 0.12 0.08 0.46 0.10HD108642 A2m / kA2hA7mA7 -0.04 0.10 0.16 0.11 0.29 0.35 0.41 0.08 0.50 0.13HD107276 Am / Ka5mA7 - - 0.03 0.23 - - -0.14 0.10 -0.22 0.19HD108486 AmkA3hA5mA7 - - 0.20 0.09 - - 0.21 0.21 0.76 0.02HD106999 Am - - 0.08 0.13 - - 0.23 0.20 0.47 0.07
HD107513 Am / kA7hF0mF0 - - -0.02 0.07 - - -0.23 0.10 -0.06 0.19HD SpT YII σ Y ZrII σ Zr BaII σ Ba HD107966
A3V / A3IV 0.00 0.13 0.38 0.12 0.04 0.26HD108382 A4V / A3IV 0.25 0.21 0.16 0.17 -0.24 0.03HD106887 A4m 0.78 0.16 0.61 0.15 1.40 0.26
HD107655
A0V 0.80 0.15 0.78 0.12 0.72 0.32
HD107168
A8m / kA5hA5mF0 0.97 0.12 0.93 0.13HD109307 A4Vm / A3IV-V 0.56 0.07 0.60 0.11 1.17 0.23HD108642 A2m / kA2hA7mA7 0.88 0.12 0.75 0.12 1.79 0.24HD107276 Am / Ka5mA7 0.12 0.10 -0.15 0.20 0.44 0.21HD108486 AmkA3hA5mA7 0.56 0.12 0.67 0.13 1.57 0.21HD106999 Am 0.27 0.11 0.49 0.08 0.78 0.07
HD107513 Am / kA7hF0mF0 0.08 0.11 0.13 0.20 0.53 0.15 This can be established independently of errors in the absolutevalues of the oscillator strengths, since all stars will be a ff ectedin the same manner. Seven lines of quality B of C I have been synthesized. In a graphof [
CFe ] versus T e ff , this element displays a di ff erent behaviour inF stars and A stars. The F stars show very little scatter aroundtheir mean iron abundance (about -0.01 dex) whereas the A andAm stars exhibit a pronounced spread in abundances of about0.75 dex which is much larger than the estimated typical uncer- Table 5.
Abundances relative to hydrogen and to the solar value, [ XH ] = log( XH ) ⋆ − log( XH ) ⊙ for the F stars. The HD numbers initalics are those for which the uncertainties have been calculated as explained in Appendix A. For the others, the quantities labeledas σ are standard deviations. HD SpT CI σ C OI σ O NaI σ Na MgII(MgI) σ Mg SiII σ Si HD106103
F5V -0.04 0.08 -0.32 0.15 -0.04 0.04 0.30(0.02) 0.21(0.09) 0.10 0.10HD106293 F5V -0.07 0.27 - - 0.01 0.18 0.45(0.10) 0.18(0.06) 0.26 0.11HD106691 F5IV -0.07 0.16 -0.26 0.15 -0.06 0.20 0.30(0.10) 0.18(0.07) 0.12 0.08HD106946 F2V -0.02 0.20 0.02 0.15 0.09 0.14 0.45(0.13) 0.18(0.04) 0.26 0.07
HD107611
F6V 0.05 0.10 - - -0.10 0.04 0.23(0.10) 0.18(0.15) 0.22 0.12
HD109530
F2V 0.08 0.09 0.10 0.14 0.16 0.12 0.39(-0.01) 0.11(0.13) 0.13 0.28HD107877 F6 - - - - -0.13 0.03 0.40 0.18 - -HD108154 F5 - - - - -0.13 0.21 0.15 0.18 - -HD108226 F5 - - - - 0.04 0.23 0.22 0.18 - -HD108976 F6 V - - - - 0.02 0.20 0.13 0.18 - -HD109069 F0 V - - - - - - 0.41 0.18 - -HD SpT CaII σ Ca ScII σ Sc TiII σ Ti VII σ V CrII σ Cr HD106103
F5V -0.17 0.21 0.00 0.08 0.00 0.10 0.44 0.15 0.05 0.10HD106293 F5V - - -0.07 0.15 0.01 0.16 0.62 0.19 0.11 0.09HD106691 F5IV -0.23 0.19 -0.01 0.16 0.00 0.15 0.42 0.33 0.04 0.12HD106946 F2V - - -0.04 0.17 0.21 0.17 0.60 0.44 0.11 0.18
HD107611
F6V - - -0.05 0.08 0.14 0.06 0.68 0.10 0.10 0.08
HD109530
F2V - - 0.06 0.14 0.25 0.10 0.45 0.33 0.02 0.08HD107877 F6 - - - - 0.18 0.23 0.64 0.26 0.33 0.08HD108154 F5 - - - - 0.18 0.13 0.12 0.36 0.38 0.08HD108226 F5 - - - - 0.24 0.14 0.57 0.25 0.38 0.08HD108976 F6 V - - - - 0.09 0.15 0.56 0.12 0.22 0.08HD109069 F0 V - - - - -0.01 0.21 0.40 0.02 - -HD SpT MnI σ Mn FeII σ Fe Co σ Co NiI σ Ni SrII σ Sr HD106103
F5V -0.06 0.07 0.09 0.05 -0.01 0.26 0.13 0.07 0.20 0.15HD106293 F5V 0.04 0.09 0.19 0.17 0.44 0.26 0.21 0.22 -0.01 0.00HD106691 F5IV 0.08 0.10 -0.08 0.16 0.64 0.14 -0.04 0.08 0.15 0.02HD106946 F2V 0.08 0.07 0.18 0.08 0.29 0.25 -0.02 0.18 0.12 0.10
HD107611
F6V 0.01 0.06 0.09 0.05 -0.10 0.33 -0.06 0.05 0.14 0.17
HD109530
F2V -0.20 0.31 0.15 0.09 -0.02 0.24 0.21 0.11 0.27 0.20HD107877 F6 0.15 0.12 0.15 0.13 -0.30 0.28 0.12 0.08 - -HD108154 F5 0.13 0.13 0.01 0.11 -0.35 0.30 0.09 0.08 - -HD108226 F5 0.11 0.07 0.05 0.11 -0.30 0.34 0.08 0.08 - -HD108976 F6 V -0.46 0.12 -0.01 0.14 -0.29 0.16 0.03 0.13 - -HD109069 F0 V - - -0.03 0.17 0.45 0.37 0.29 0.14 - -HD SpT YII σ Y ZrII σ ZrII
BaII σ Ba HD106103
F5V 0.00 0.08 0.04 0.11 0.86 0.10HD106293 F5V 0.01 0.06 -0.04 0.09 0.64 0.23HD106691 F5IV 0.04 0.08 0.50 0.15 0.65 0.21HD106946 F2V 0.24 0.19 0.38 0.09 0.59 0.16
HD107611
F6V 0.10 0.12 0.13 0.17 0.84 0.10
HD109530
F2V 0.06 0.20 -0.13 0.14 0.43 0.14HD107877 F6 - - 0.14 0.17 0.56 0.10HD108154 F5 - - 0.03 0.17 0.54 0.10HD108226 F5 - - -0.06 0.17 0.59 0.10HD108976 F6 V - - -0.50 0.17 0.29 0.10HD109069 F0 V - - -0.04 0.17 0.37 0.10 tainty of about 0.15 dex on [ CH ]. Also all A stars display deficien-cies in carbon. There seems thus to be real star-to-star variationin [ CFe ]. Another way to quantify the relative dispersion of A andAm stars with respect to the F stars is to compute the mean abun-dance for carbon and the associated dispersion for each group ofstars, σ A and σ F . For A and F stars, these mean abundances anddispersions for carbon and all other chemical elements are col-lected in Table 6. For carbon, the dispersion σ A for the A starsis about 4 times higher than for F stars.The behaviour of carbon with respect to iron is not clear when [ CH ] is displayed versus [ FeH ]. As previously found by Hill (1995),an anticorrelation appears more clearly when [
CFe ] is displayedagainst [
FeH ] for Am and normal A stars (Figure 7). We find aslope of − . ± .
48 , not very di ff erent from the value obtainedby Hill (1995) ( − . ± .
31 for 15 A stars).We have checked whether the C lines studied here mightbe a ff ected by non-LTE e ff ects especially for A stars. Non-LTE abundance corrections of carbon have been calculated byRentzsch-Holm (1996) for a set of main sequence stars rang- . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 11 Fig. 6.
Abundance patterns for the ”normal” A (a), Am (b) and F stars of Coma Berenices cluster. A maximum ± ff ective temperatures from 7000 K to 12000 K, sur-face gravities from log g = . MH ] = − . ff ective temperatures below10000 K, the non-LTE abundance corrections were found to bealways negative. Only three lines of our list were studied byRentzsch-Holm (1996): λ λ λ / H].
Oxygen
The derived oxygen abundances are the weighted means of 16quality B and 6 quality C + lines. In a graph of [ OH ] versus T e ff (Figure 7), the F stars abundances are slightly scattered aroundtheir mean value. All A and Am stars exhibit underabundancesof oxygen and their scatter is again larger than that of the F stars.The total spread in oxygen abundance for the A and Am stars isabout 0.8 dex, significantly larger than the maximum uncertaintyof 0.12 dex for A stars. This again suggests real star-to-star vari-ations in [O / H].In a graph of [ OH ] versus [ FeH ], oxygen seems to be only veryloosely anticorrelated with iron. The anticorrelation appears
Fig. 7.
Left panel: Abundance of carbon, oxygen and sodium versus e ff ective temperature. The dotted line corresponds to the solarvalue and the dashed one to the mean value determined for the F stars of the cluster. Right panel: [C / Fe], [O / Fe] and [Na / H] versus[Fe / H]. The filled dots correspond to normal A stars, the filled squares correspond to Am stars and the filled diamonds correspondto F stars. In the plot representing [Na / H] versus [Fe / H], the dashed line corresponds to the solar [Na / Fe] ratio. The error bars in theright panel represent the mean uncertainties for the displayed abundances.more clearly in [
OFe ] versus [
FeH ] with a slope of − . ± . ff ec-tive temperatures less than 10000 K and log g around 4.0, thesecorrections always remain less than -0.03 dex, well below the es-timated uncertainty. The observed anticorrelation of [O / Fe] ver-sus [Fe / H] should therefore remain after non-LTE correctionshave been made. The large star-to-star variations in [O / H] wouldnot be a ff ected by these corrections either. Sodium
Sodium abundances were derived from 9 lines of quality A to C.The sodium abundances for the F stars display very little scat-ter around their mean value (-0.01 dex), very close to solar. AllA and Am stars display overabundances of Na. For these stars,the spread in [
NaH ] is about 0.9 dex, much larger than the typicaluncertainty (0.10 dex). This suggests there are real star-to-starvariations in [
NaH ]. The sodium abundance does not appear to becorrelated to the iron abundance: in Figure 7, about half of thedata lie above the line of solar ratio while the other half are scat-tered around it.Quantitative information on non-LTE corrections for Na abun-dances in A stars is scarce. Non-LTE corrections have been . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 13
Fig. 8.
Left panel: Abundance of magnesium, silicon and calcium versus e ff ective temperature. The dotted line represents the solarvalue and the dashed one represents the mean abundance of F stars. Right panel: [Mg / H], [Si / Fe] and [Ca / H] versus [Fe / H]. Thesymbols are the same as in Figure 7. The dashed lines represent the solar ratios.calculated by Bikmaev et al. (2002) for two cool A8IV-V starswhose fundamental parameters are close to the coolest A star ofour sample. The largest corrections amount to -0.35 dex and oc-cur for the two lines λ FeH ], it is di ffi cult topredict what the non-LTE corrections would be for the A andAm stars studied here. Provided the corrections would be smallfor F stars, lowering the A and Am star abundances by about0.35 dex would certainly improve the correlation of [ NaH ] with[
FeH ]. Magnesium
Seven lines of Mg II of quality B to D were used for the A starsand 5 lines of Mg I of qualities B to C for the F stars. In graphs of [
MgH ] versus T e ff (Figure 8), the F stars and the A and Amstars display fairly large and comparable scatter. The maximumspread in [ MgH ] is about 0.30 dex and 0.55 dex for the F and theA and Am stars respectively. The maximum uncertainty in [
MgH ]being about 0.18 dex, there does not appear to be significant star-to-star variations. Almost all stars exhibit LTE overabundances.We noticed that the Mg II λ MgH ] abundances (derived from all lines excluding the MgII triplet) do not appear to be correlated to [
FeH ]. Most data lieabove the line of the solar ratio [
MgFe ]. In their analysis, Hill &Landstreet (1993) found that the ratio [
MgFe ] runs parallel to andabove the line of solar [
MgFe ]. They attribute this to the use of the
Fig. 9.
Left panel: Abundance of scandium, titanium and vanadium versus e ff ective temperature. The dotted line represents the solarvalue and the dashed one represents the mean abundance of F stars. Right panel: [Sc / H], [Ti / Fe] and [V / H] versus [Fe / H]. Thesymbols are the same as in Figure 7. The dashed lines represent the solar ratios.very strong Mg II λ λλ Silicon
Nineteen lines of Si II of quality C to E have been synthesized.For the F stars, the silicon abundances hardly show any scatteraround their mean value, + S iH ].The silicon abundance does not show any convincing correlationwith the iron abundance. Most of our data fall slightly above theline representing a solar silicon to iron ratio. Note that Hill &Landstreet (1993) did find a tight correlation between [
S iH ] and[
FeH ]. All their data fall close to the line representing a solar sil-icon to iron ratio. The lines we analysed di ff er from theirs andlead to overabundances possibly because of incorrect oscillatorstrengths or non-LTE e ff ects. . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 15 Fig. 10.
Left panel: Abundance of chromium, manganese and iron versus e ff ective temperature. The dotted line represents the solarvalue and the dashed one represents the mean abundance of F stars. Right panel: [Cr / H] and [Mn / H] versus [Fe / H]. The symbolsare the same as in Figure 7. The dashed lines represent the solar ratios.
Calcium
Twelve lines of Ca II of quality C and D were used to derivethe calcium abundance for all A and Am stars and for only twoF stars. The abundance of the F stars is subsolar ( < [ CaH ] > ∼− . dex ). The two normal A stars are slightly deficient in cal-cium and the Am stars show modest deviations (both over andunderabundances) around the solar abundance. Their maximumspread in [Ca / H] is 0.50 dex which is only marginally signif-icant compared to the maximum estimated uncertainty (about0.20 dex).In a graph displaying the calcium abundance versus that of iron(Figure 8), the Am stars, as expected, can easily be discriminatedfrom the normal A and F stars. They all fall to the right in a re-gion characterized by overabundances of iron and they are eitherCa-deficient or Ca-rich. For the normal A and F stars, [
CaH ] has a fairly uniform value (-0.20) dex and does not appear to varywith [
FeH ]. Scandium
Eleven lines of quality D of Sc II were used to derive the scan-dium abundance. The scandium abundances of F stars are scat-tered very little around the solar value. About half of the A andAm stars are deficient in scandium (3 Am stars are close to solarwhile 4 exhibit deficiencies ranging from -0.2 dex to -1.2 dex).The total spread in [
S cH ] for these stars is about 1.20 dex, signifi-cantly larger than the typical uncertainty of 0.10 dex. There thusseems to be real star-to-star variation in [
S cH ].As for calcium, scandium does not exhibit any clear correlationor anticorrelation with respect to iron in the diagram of [
S cH ] ver-
Fig. 11.
Left panel: Abundance of cobalt, nickel and strontium versus e ff ective temperature. The dotted line represents the solarvalue and the dashed one represents the mean abundance of F stars. Right panel: [Co / H], [Ni / Fe] and [Sr / H] versus [Fe / H]. Thesymbols are the same as in Figure 7. The dashed lines represent the solar ratios.sus [
FeH ] (Figure 9). The scandium abundances of the F and nor-mal A stars, which are all very close to solar, do not depend on[
FeH ]. Most Am stars lie in the lower right part of the diagram:they are all iron-rich and 4 out of 7 are deficient in scandium.
Twenty six lines of Ti II, most of them of quality D, have beensynthesized. The oscillator strengths for this element are not assecure as for iron and chromium. One should not expect theseabundances to be too reliable on an absolute scale. The F starsshow little scatter around their mean [
TiH ] value, about + TiH ] is about 0.68dex which is larger than the typical uncertainty of about 0.08dex. There seems to be real star-to-star variation for this element(Figure 9). The titanium abundance appears to be loosely corre-lated to that of iron (correlation coe ffi cient 0.66) for normal Fand A stars. The [ TiFe ] ratios of these normal stars appear to beclose to or slightly higher than solar as found by Lemke (1989)for a sample of 16 normal A stars. Hill & Landstreet (1993)found a strong correlation of [Ti / H] with [Fe / H](correlation co-e ffi cient 0.95), their [ TiFe ] ratios being slightly above the solarvalue.Nine lines of V II were synthesized whose uncertainties are notspecified. These lines are intrinsically weak in all spectra. Theabundances derived for this element must be taken with cau-tion. The vanadium abundances for the F stars are fairly scat- . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 17
Fig. 12.
Left panel: Abundance of yttrium, zirconium and barium versus e ff ective temperature. The dotted line represents the solarvalue and the dashed one represents the mean abundance of F stars. Right panel: [Y / H], [Zr / Fe] and [Ba / H] versus [Fe / H]. Thesymbols are the same as in Figure 7. The dashed lines represent the solar ratios.tered around their mean value, + VH ] and [ FeH ] (Figure 9). Most of the data fot thenormal A stars and the F stars fall above the line representingthe solar [
VFe ] ratio which may be due to poorly determined os-cillator strengths. Hill & Landstreet (1993) found a correlationof [ VH ] with [ FeH ] using lines other than ours; their ratios [
VFe ] arealso significantly above solar.Eleven lines of Cr II of quality D have been synthesized. As fortitanium, the A stars are only a little more scattered than the Fstars. The total spread in [
CrH ] for the A and Am stars is about0.56 dex (0.36 dex for the F stars) while the maximum uncer-tainty on [
CrH ] is about 0.20. Should the data for HD107168 be re-moved, the spread in [
CrH ] drops to 0.30 dex and is not significant. The evidence for star-to-star variations in [
CrH ] is therefore ratherweak. The chromium abundance appears to be loosely correlatedwith that of iron (correlation factor = CrFe ] ratio (Figure 10). Hill & Landstreet (1993) alsofound a correlation of [
CrH ] with [
FeH ], their ratios [
CrFe ] being onlymarginally above solar.Twenty lines of Mn I of quality B to C + were synthesized. The Fstars show little scatter around the solar value. The total spreadfor the A and Am stars is about 0.6 dex which is larger thanthe maximum uncertainty of 0.2 dex, suggesting real star-to-starvariation. The two normal A stars are deficient in manganese.The manganese abundance appears to be well correlated withthat of iron (correlation factor = MnFe ]run parallel to and below the line representing the solar ratio (Figure 10). Hill & Landstreet (1993) found a correlation of [
MnH ]with [
FeH ] using lines other than ours; their ratios [
MnFe ] are onlymarginally above solar.
Iron, cobalt and nickel:
Twenty seven lines of Fe II of quality C to E were synthesized.In a graph [
FeH ] versus T e ff (Figure 10), the F stars are onlyslightly scattered around their mean value, + < [ FeH ] > = − . ± . dex , based on the analysis ofthe equivalent widths of a few Fe I lines for 14 F stars of thecluster. Our usage of di ff erent techniques (model atmospheresand line synthesis) and di ff erent lines (Fe II) probably accountsfor the di ff erence in iron abundance. The total spread in [ FeH ] forthe A and Am stars is about 0.53 dex which is larger than themaximum estimated uncertainty (0.10 dex). There are thus realstar-to-star variations in [
FeH ].For cobalt, our analysis is based on 12 lines of Co I, whose er-rors in the oscillator strengths are unknown. Most of these linesare weak and often are blended with lines whose atomic param-eters are not necessarily accurately known. We therefore do notexpect these abundances to be reliable. In a graph of [
CoH ] versus T e ff (Figure 11), the cobalt abundances are much more scatteredfor the F stars than for any other chemical element; this scatter isprobably largely due to blending species. There is only one datapoint for the normal A stars. We feel that the abundances are notreliable enough to claim to find real star-to-star variations in [ CoH ]and not to find a correlation with [
FeH ].Fifty six lines of Ni I of quality C + to D have been synthesized.In the graph [ NiH ] versus T e ff , nickel behaves in a similar manneras iron (Figure 11). The F stars are fairly well grouped aroundtheir mean abundance ( + NiH ]. The nickel abundance istightly correlated with that of iron (correlation coe ffi cient 0.92).Most of the ratios [ NiFe ] are close to solar.
Our study of strontium is based on only two lines at 4077.71Å and 4215.52 Å whose accuracies are not specified in NIST.Errors in gf values may result in a considerable zero point shiftwith respect to the Sun. We therefore do not attach too muchsignificance to apparently large absolute over-or underabun-dances. The F star strontium abundances show little scatteraround their mean value, + ffi cient 0.90). Strontiumabundances vary more rapidly than those of iron as found byLemke (1990).Five lines of yttrium whose accuracies are unknown weresynthesized. The same holds for zirconium. Yttrium and zirco-nium are found to be nearly solar in F stars (Figure 12). Theseelements are overabundant by about 0.50 dex for 6 of the Amstars and by about 0.40 dex in the normal A stars. The A andAm stars display fairly large spreads in abundances of Y andZr, about 1.0 dex, much larger than the associated uncertainties, Table 6.
Mean abundances and their respective dispersion forFe, C, O, Ni, Sc, Si, Ba, Y, Zr and Sr for the F, normal A and Amstars.
Elements F stars σ F A stars σ A [ CH ] -0.01 0.06 -0.48 0.24[ OH ] -0.11 0.18 -0.34 0.31[ S iH ] 0.18 0.07 0.22 0.20[
S cH ] -0.02 0.04 -0.23 0.38[
FeH ] 0.07 0.09 -0.14 0.16[
NiH ] 0.09 0.11 0.11 0.28[
S rH ] 0.15 0.08 0.25 0.43[ YH ] 0.07 0.08 0.47 0.34[ ZrH ] 0.04 0.34 0.46 0.33[
BaH ] 0.58 0.17 0.84 0.70 indicating real star-to-star variations in [ YH ] and [ ZrH ].Five lines of quality B of barium have been synthesized. Thiselement is found to be overabundant in all A, Am and F starsby large amounts (up to 1.10 dex). The spread for [
BaH ] is about1.8 dex, much larger than the typical uncertainties, indicatingreal star-to-star variations in [
BaH ] (Figure 12). However, theseoverabundances may reflect a non-LTE e ff ect, namely anoverionization of barium in A stars. In Vega, the non-LTE BaII abundances are lower than the LTE abundances by about0.30 dex as demonstrated by Gigas (1988) and Lemke (1990)using the KIEL code. Non-LTE corrections for Ba should besensitive to [Fe / H] which can di ff er much from star to star.Detailed calculations for the respective fundamental parametersand iron abundances of each star should therefore be performed . A byproduct of this analysis has been to determine microturbu-lent velocities for each star. Figure 13 displays the derived ξ t versus e ff ective temperature. The overall variation agrees wellwith that found by Coupry & Burkhart (1992) who found that ξ t varies from 0 km · s − for late B-type stars, up to about 3 km · s − for mid-A type stars down to around 2 km · s − for early F-typestars. Gray (2001) also found that ξ t varies diminishes 3 km · s − for mid-A type stars to about 1 km · s − for solar-type stars. The inferred rotational velicities v e sin i are lower than 102km · s − . None of the found abundances appears to depend on v e sin i . The profiles of [X / H] versus v e sin i are flat as shownfor instance for iron in Figure 14. This result is not surprising.Detailed calculations of di ff usion in the presence of meridionalcirculation carried out by Charbonneau & Michaud (1991) re-vealed that, in stars rotating at less than v e sin i =
100 km · s − ,meridional circulation has little influence on chemical separationonce the helium superficial convective zone has disappeared.Accordingly, the abundances should not present any positive nornegative trend (slope) with v e sin i in this velocity regime. Theabsence of fast rotators in Coma Berenices (ie. v e sin i > · s − ) prevents us from investigating the behaviour of the var-ious abundances with rotational velocity above that velocity. . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 19 Normal A starsAm starsF stars
Fig. 13.
Variation of the derived microturbulence velocities withe ff ective temperature. Fig. 14.
Abundance of iron versus v e sin i for A,Am and F stars.
5. Astrophysical implications
The most important result of this study is the evidence of largestar-to-star abundance variations for A stars in Coma Berenices.These stars appear to display much larger star-to-star variationsin their abundances than the F stars do for the following chemicalelements: C, O, Na, Sc, Ti, Mn, Fe, Ni, Sr, Y, Zr and Ba. In con-trast, the abundances of Mg, Si, Ca and Cr do not show signif-icant star-to-star variations. For the Hyades, Varenne & Monier(1999) found a similar behaviour for the abundances of C, O,Na, Sc, Fe and Ni. Monier (2005) also found large star-to-starvariations in [Fe / H], [Ni / H] and [Si / H] for several A stars of theUma group.
We theorize that this peculiar behaviour is a signature of theoccurence of transport processes competing with radiative dif-fusion (eg. rotational mixing in radiative zone, Zahn 2005) .Indeed, if radiative di ff usion was the only process at work,the microscopic di ff usion velocity of a given chemical element should be the same in stars of similar e ff ective temperaturesand surface gravities. We would therefore expect similar surfacecompositions for stars of similar fundamental parameters. We have compared the derived abundances to the predictions ofrecent evolutionary models, calculated with the Montr´eal codeusing slightly di ff erent assumptions and physics. This code treatsradiative di ff usion in detail and allows the inclusion of turbulentdi ff usion. Schatzman (1969) proposed two possible physical ori-gins of turbulent di ff usion: loss of angular momentum while thestar descends towards the Main Sequence or meridional circula-tion. For the F stars, the found abundances have been compared to thepredictions of Turcotte et al’s (1998) evolutionary models at theage of Coma Berenices. Their models predict the evolution ofthe abundances for an F star of a given mass consistently with theinternal structure. They include the e ff ects of gravitational sed-imentation and of radiative di ff usion for 28 chemical elements( Z ≤
28) but no macroscopic mixing (wind, accretion, merid-ional circulation, turbulence). These models are relevant for Fstars with masses in the range 1.1 M ⊙ to 1.5 M ⊙ from the pre-main sequence state up to hydrogen core exhaustion. The impactof the abundance variations with time on the structure of the staris taken into account. Monochromatic OPAL opacities for eachelement are used to recalculate the opacity corresponding to theabundances and local conditions during the evolution. Atomicdi ff usion has an important e ff ect on the opacities for stars moremassive than 1 . M ⊙ . In these objects, the abundances of Fe andother iron-peak elements were found to substantially vary withtime.For the light elements C and O, the predicted large underabun-dances by Turcotte et al. (1998) for stars with T e f f > < [ CH ] > = − . ± < [ OH ] > = − . ± ≤ T e f f ≤ M ⊙ , the predicted Mg and Si underabundancesare not seen either in our data. For the iron peak elements Fe andNi, the predicted overabundances due to microscopic di ff usionare not observed either. For A stars, we have compared the found abundances withthe predictions of the recent models of Richer et al. (2000). Inthese new models, the e ff ect of atomic and turbulent di ff usionwere calculated for stars of 1.45-3 M ⊙ . They showed that thesuperficial abundances of the 28 species calculated in theirmodels depend, in a star of a given mass, on essentially theinitial metallicity and the deph of the zone mixed by turbulence.Figures 10 and 11 from Richer et al. (2000) represent thevariations of surface abundances with time of the 28 elementsfor stars of 2-2.5-3 M ⊙ . For the 2 M ⊙ model, Richer et al. (2000)found that the ratio of the abundances of carbon and oxygenat 450 Myr to their initial values decrease with time. Weobserved the same trend for C and O in Am stars using the mean abundances of C and O in the F stars as initial abundances ofthese elements in the cluster.In addition, we have specifically computed a series of mod-els for two Am stars of similar e ff ective temperatures but dif-ferent rotational velocities: HD108642, a slow rotator ( v e sin i = · s − ) and HD106887, a faster rotator ( v e sin i = · s − ). Masses for these two stars should be in the range1.8 to 2.0 M ⊙ . Seven models for a 1.8 M ⊙ mass star hav-ing di ff erent turbulent coe ffi cients, D T (equation 1 in Richeret al. 2000) at the age of the Coma were computed. Theadopted initial and homogeneous abundances for these mod-els are collected in Table 7. The models use an Eggleton-Faulkner-Flannery equation of state (Eggleton et al. 1973) in-cluding Coulomb correction on the pressure (labeled as CEFFmodels) (see also Christensen-Dalsgaard & Daeppen 1992).The nuclear energy generation follows the prescriptions ofBahcall & Pinsonneault (1992). These models take into accountgravitational settling, thermal di ff usion and radiative accelera-tions. The detailed treatment of atomic di ff usion is describedin Turcotte et al. (1998) and the radiative accelerations are fromTurcotte et al. (1998) with correction for redistribution fromGonzalez et al. (1995) and LeBlanc et al. (2000). These mod-els are self consistent as the Rosseland opacity and radia-tive accelerations are recomputed at each time step in eachlayer for the exact local chemical composition using OPALmonochromatic opacities for 24 elements. Convection and semi-convection are modeled as di ff usion processes as described inRicher et al. (2000) and Richard et al. (2001). The initial metal-licity is taken from Friel & Boesgaard (1992).The models are compared to the abundance patterns ofHD108642 in Figure 15 and for HD106887 in Figure 16. Noneof the models reproduces entirely the characteristic abundancepattern, ie. marked underabundances of light elements and over-abundances of iron-peak elements. They basically di ff er by theamount of turbulent di ff usion included. The model that best ap-proaches the abundance patterns for elements with Z >
20 forthese two stars is the model labeled as 1.8T5.3D200K-3 follow-ing the syntax of Table 1 of Richer et al. (2000). In this model,the turbulent di ff usion coe ffi cient D T varies with density as: D T = ω D (He) (cid:16) ρ ρ (cid:17) n where n = D (He) is the atomic di ff usion coe ffi cient of He atthe density ρ = ρ ( T ) in the iron convection zone (see equation1 of Richer et al. 2000, here ω = · and log T = . ff usion 1.8T5.3D25K-3 and 1.8T5.3D500-3 roughly accountfor the abundances of elements with Z <
15 but predict too largeoverabundances of iron-peak elements.Part of the discrepancy between observed and theoretical abun-dance patterns could be due to non-LTE e ff ects. Correcting themagnesium abundance for the magnesium triplet at λ ∼ ff ected by non-LTE e ff ects,will also improve the agreement with this model. However, theinclusion of competing processes in the models such as rota-tional mixing in the radiative zone for A stars, internal waves forF stars could also help improve the agreement. Table 7.
Initial chemical composition
Element Mass fractionH . . . . . . . . . . . . . . . . . . 7.03 × − He ( a ) . . . . . . . . . . . . . . 2.7995 × − C ( b ) . . . . . . . . . . . . . . 2.935 × − N . . . . . . . . . . . . . . . . . . 9.000 × − O . . . . . . . . . . . . . . . . . . 8.189 × − Ne . . . . . . . . . . . . . . . . . 1.675 × − Na . . . . . . . . . . . . . . . . . 3.396 × − Mg . . . . . . . . . . . . . . . . . 6.377 × − Al . . . . . . . . . . . . . . . . . . 5.519 × − Si . . . . . . . . . . . . . . . . . . 6.878 × − P . . . . . . . . . . . . . . . . . . . 5.944 × − S . . . . . . . . . . . . . . . . . . . 3.592 × − Cl . . . . . . . . . . . . . . . . . . 7.642 × − Ar . . . . . . . . . . . . . . . . . . 9.170 × − K . . . . . . . . . . . . . . . . . . 3.396 × − Ca . . . . . . . . . . . . . . . . . . 6.368 × − Ti . . . . . . . . . . . . . . . . . . 3.396 × − Cr . . . . . . . . . . . . . . . . . . 1.698 × − Mn . . . . . . . . . . . . . . . . . 9.340 × − Fe . . . . . . . . . . . . . . . . . . 1.219 × − Ni . . . . . . . . . . . . . . . . . . 7.557 × − (a) He = . × − (b) C is 1% of C Z −1.5−1.0−0.5 0.0 0.5 1.0 1.5 [ N / H ] H HeLi BeB C N O F NeNaMgAlSi P S ClAr K CaScTi V CrMnFeCoNiCu M T5.3D500−3 445 Myr, T eff = 8113 K, logg= 4.18 1.8 M T5.3D25k−3 451 Myr, T eff = 8164 K, logg= 4.19 1.8 M T5.3D200k−3 451 Myr, T eff = 8190 K, logg= 4.19 HD108642
Fig. 15.
Comparison of the observed abundance pattern ofHD108642 (A2m) with the predictions of three models calcu-lated for a mass of 1.8 M ⊙ and di ff erent amounts of turbulentdi ff usion. Observed abundances are represented as triangles witherror bars. Note that the models do not predict the surface abun-dances of Sc, V and Co.
6. Conclusion
High and medium resolution spectra of 11 A and 11 F starmembers of the Coma Berenices open cluster have been synthe-sized in order to determine the abundances of C, O, Na, Mg, Si,Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Sr, Y, Zr and Ba. In graphsrepresenting the abundance [X / H] versus e ff ective temperature,the A stars display abundances that are more scattered around . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 21 Z −1.5−1.0−0.5 0.0 0.5 1.0 1.5 [ N / H ] H HeLi BeB C N O F NeNaMgAlSi P S ClAr K CaScTi V CrMnFeCoNiCu M T5.3D500−3 445 Myr, T eff = 8113 K, logg= 4.18 1.8 M T5.3D25k−3 451 Myr, T eff = 8164 K, logg= 4.19 1.8 M T5.3D200k−3 451 Myr, T eff = 8190 K, logg= 4.19 HD106887
Fig. 16.
Comparison of the observed abundance pattern ofHD106887 (A4m) with the same models.the mean value than the F stars. Large star to star variations aredetected for A stars for C, O, Na, Sc, Ti, Mn, Fe, Ni, Sr, Y, Zrand Ba which we interpret as evidence of transport processescompeting with radiative di ff usion.The chemical pattern found for the A and F dwarfs ofComa resembles that found for the Hyades (Varenne &Monier 1999) and the UMa group (Monier 2005). Themean iron abundance derived for the F stars is found to be < [ FeH ] > = . ± . dex , slightly higher than the metallicityderived by Friel & Boesgaard (1992). The abundances ofmanganese, nickel, strontium and barium are strongly correlatedwith the iron abundance for A and Am stars. The ratios [C / Fe]and [O / Fe] appear to be anticorrelated with [Fe / H]. The ratio[Ti / Fe] is solar as found by Lemke (1989).The Am stars in Coma Berenices are found to be deficient inlight elements (C and O), they are not all deficient in calciumand scandium but are all overabundant in metallic and heavyelements (Fe, Co, Ni, Sr, Y, Zr and Ba). The two normal Astars have almost solar abundances. The F stars have solarabundances for almost all the elements except for Mg, Si, V andBa.The abundance patterns predicted by current state of theart evolutionary models following the prescriptions ofRicher et al. (2000) have been compared to the observedabundance patterns of two Am stars of the cluster, one aslow rotator (HD108642) and one a moderately fast rotator(HD106887). These models were calculated with di ff erentstrengths of the turbulent di ff usion coe ffi cient. None of the mod-els reproduces entirely the characteristic abundance pattern, ie.marked underabundances of light elements and overabundancesof iron-peak and heavier elements. Part of the discrepancy mayarise from non-LTE e ff ects. In this respect, non-LTE abundancedeterminations for many of the elements analysed here (whenfeasible, ie. when atomic data and model atoms are available)are highly desirable. However it is likely that the inclusion ofcompeting processes such as rotational mixing in the radiativezones (which will vary from star to star) should also help reproduce the observed abundance patterns and the large scatterof abundances of several elements in A stars. Appendix A: Determination of uncertainties
Six major sources are included in the uncertainty determina-tions: uncertainty on the e ff ective temperature ( σ T e ff ), on thesurface gravity ( σ log g ), on the microturbulent velocity ( σ ξ t ),on the apparent rotational velocity ( σ v e sin i ), the oscillatorstrength ( σ log g f ) and the continuum placement ( σ cont ). Theseuncertainties are supposed to be independent, so that the totaluncertainty σ tot i for a given transition (i) is: σ tot i = σ T e ff + σ g + σ ξ t + σ v e sin i + σ g f + σ cont . (A.1)The mean abundance < [ XH ] > is then computed as a weightedmean of the individual abundances [X / H] i derived for eachtransition (i): < [ XH ] > = P i ([ XH ] i /σ tot i ) P i (1 /σ tot i ) (A.2)and the standard deviation, σ sd is given by:1 σ sd = N X i = (1 /σ tot i ) (A.3)where N is the number of lines per element. This procedure hasbeen applied to 4 A stars (HD107966, HD107655, HD107168and HD107513) and 3 F stars (HD106103, HD107611 andHD109530). In the fastest rotators, the pseudo-continuum, f λ pseudo , we see that regions free of lines in the observed spec-tra are actually a ff ected by the rotational broadening of neigh-bouring lines. The level of the actual continuum, f λ cont , can berecovered by carefully synthesizing the pseudo-continuum win-dows, yielding the theoretical flux, F λ pseudo , and the actual cor-responding continuum flux, F λ cont , at the appropriate velocity v e sin i ± ∆ ( v e sin i ). The observed intensity level in these pseudo-continuum windows was then multiplied by ( F λ cont / F λ pseudo ) to re-cover the observed continuum level: f λ cont = F λ cont F λ pseudo f λ pseudo . (A.4)The maximum and minimum allowed rotational velocity v e sin i ± ∆ ( v e sin i ) yields 2 ratios F λ cont F λ pseudo ( v e sin i max ) and F λ cont F λ pseudo ( v e sin i min ). The corresponding observed normalized lineprofiles were then used to derive the corresponding changes inabundances due to the di ff erent locations of the continuum. Thistest was performed in several spectral regions (excluding over-laping regions of 2 orders) and yields a maximum uncertainty ofabout 0.07 dex on the abundances.For the other stars, the final abundances are averages. Theerrors on the elemental abundances are standard deviation as-suming a Gaussian distribution of the abundances derived fromeach line¯ x = P i x i N (A.5) σ = P i ( x i − ¯ x ) N (A.6)where ¯ x is the mean value of the abundance, N the number oflines of the element and σ the standard deviation. Acknowledgements.
We warmly thank the OHP night sta ff for the support dur-ing the observing runs. This research has used the SIMBAD, WEBDA, VALD,NIST and Kurucz databases. References
Bahcall, J. N., & Pinsonneault, M. H. 1992, Reviews of Modern Physics, 64, 885Baranne, A., et al. 1996, A&AS, 119, 373Biermann, L., L¨ubeck, K. 1948, Zeitschrift fur Astrophysik, 25, 325Biemont, E., Grevesse, N., Hannaford, P., & Lowe, R. M. 1981, ApJ, 248, 867Bikmaev, I. F., et al. 2002, A&A, 389, 537Burkhart, C., & Coupry, M. F. 2000, A&A, 354, 216Boesgaard, A. M. 1987, ApJ, 321, 967Bounatiro, L. 1993, A&AS, 100, 531Castelli, F., & Kurucz, R. L. 2003, IAU Symposium, 210, 20PCayrel, R., Cayrel de Strobel, G., & Campbell, B. 1988, The Impact of VeryHigh S / N Spectroscopy on Stellar Physics, 132, 449Charbonneau, P., & Michaud, G. 1991, ApJ, 370, 693Coupry, M. F., & Burkhart, C. 1992, A&AS, 95, 41Christensen-Dalsgaard, J., & Daeppen, W. 1992, A&A Rev., 4, 267Dommanget, J., & Nys, O. 1995, Bulletin d’Information du Centre de DonneesStellaires, 46, 3Edvardsson, B., Andersen, J., Gustafsson, B., Lambert, D. L., Nissen, P. E., &Tomkin, J. 1993, A&A, 275, 101Eggleton, P. P., Faulkner, J., & Flannery, B. P. 1973, A&A, 23, 325Erspamer, D., & North, P. 2002, A&A, 383, 227Friel, E. D., & Boesgaard, A. M. 1992, ApJ, 387, 170Fuhr, J. R., Martin, G. A., & Wiese, W. L. 1988, New York: American Instituteof Physics (AIP) and American Chemical Society, 1988,Giddings, J. R. 1981, Ph.D. Thesis,Gigas, D. 1988, A&A, 192, 264Gillet, D., et al. 1994, A&AS, 108, 181Gonzalez, J.-F., LeBlanc, F., Artru, M.-C., & Michaud, G. 1995, A&A, 297, 223Grevesse, N., & Sauval, A. J. 1998, Space Science Reviews, 85, 161Hill, G. M. 1995, A&A, 294, 536Hill, G. M., & Landstreet, J. D. 1993, A&A, 276, 142Holweger, H., Ste ff en, M., & Gigas, D. 1986, A&A, 163, 333Horne, K. 1986, PASP, 98, 609Hubeny, I., & Lanz, T. 1992, A&A, 262, 501Hui-Bon-Hoa, A., & Alecian, G. 1998, A&A, 332, 224Hui-Bon-Hoa, A., Burkhart, C., & Alecian, G. 1997, A&A, 323, 901Je ff ries, R. D. 1999, MNRAS, 304, 821Kostyk, R. I., & Orlova, T. V. 1983, Astrometriia i Astrofizika, 49, 39Kupka, F., Piskunov, N., Ryabchikova, T. A., Stempels, H. C., & Weiss, W. W.1999, A&AS, 138, 119Kurucz, R. L. 1979, ApJS, 40, 1Kurucz, R. L. 1992, Revista Mexicana de Astronomia y Astrofisica, vol. 23, 23,45Kurucz, R. L. 1992, Revista Mexicana de Astronomia y Astrofisica, vol. 23, 23,181Kurucz, R. L. 2005, Memorie della Societa Astronomica Italiana Supplement, 8,14Lambert, D. L., McKinley, L. K., & Roby, S. W. 1986, PASP, 98, 927LeBlanc, F., Michaud, G., & Richer, J. 2000, ApJ, 538, 876Lemke, M. 1989, A&A, 225, 125Lemke, M. 1990, A&A, 240, 331Magazzu, A., & Cowley, C. R. 1986, ApJ, 308, 254Miles, B. M., & Wiese, W. L. 1969, Atomic Data, Vol. 1, p.1, 1, 1Monier, R. 2005, A&A, 442, 563Monier, R., & Richard, O. 2004, IAU Symposium, 224, 209Napiwotzki, R., Schoenberner, D., & Wenske, V. 1993, A&A, 268, 653Nissen, P. E. 1981, A&A, 97, 145Pickering, J. C., Thorne, A. P., & Perez, R. 2002, ApJS, 138, 247Przybilla, N., Butler, K., Becker, S. R., Kudritzki, R. P., & Venn, K. A. 2000,A&A, 359, 1085Przybilla, N., Butler, K., Becker, S. R., & Kudritzki, R. P. 2001, A&A, 369, 1009Rentzsch-Holm, I. 1996, A&A, 312, 966Rentzsch-Holm, I. 1997, A&A, 317, 178Richard, O., Michaud, G., & Richer, J. 2001, ApJ, 558, 377 Richard, O., Michaud, G., & Richer, J. 2002, IAU Colloq. 185: Radial andNonradial Pulsationsn as Probes of Stellar Physics, 259, 270Richer, J., Michaud, G., & Turcotte, S. 2000, ApJ, 529, 338Savanov, I. S. 1996, Astronomy Reports, 40, 196Rodriguez, E., Lopez de Coca, P., Rolland, A., Garrido, R., & Costa, V. 1994,A&AS, 106, 21Schatzman, E. 1969, A&A, 3, 331Sigut, T. A. A., & Landstreet, J. D. 1990, MNRAS, 247, 611Smalley, B. 2004, IAU Symposium, 224, 131Soderblom, D. R., Oey, M. S., Johnson, D. R. H., & Stone, R. P. S. 1990, AJ, 99,595Ste ff en, M. 1985, A&AS, 59, 403Takeda, Y. 1995, PASJ, 47, 287Tody, D. 1993, ASP Conf. Ser. 52: Astronomical Data Analysis Software andSystems II, 52, 173Trumpler, R. J. 1938, Lick Observatory Bulletin, 18, 167Turcotte, S., Richer, J., & Michaud, G. 1998, ApJ, 504, 559Turcotte, S., Richer, J., Michaud, G., Iglesias, C. A., & Rogers, F. J. 1998, ApJ,504, 539Varenne, O. 1999, A&A, 341, 233Varenne, O., & Monier, R. 1999, A&A, 351, 247Ward, L. 1985, MNRAS, 213, 71Zahn, J.-P. 2005, EAS Publications Series, 17, 157 . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 23 Table 8.
Linelist used for abundance determination. The ”A” term is used for a g f accurancy lower than 3%, ”B” lower than 10%, ”C + ” lowerthan 18%, ”C” lower than 25%, ”D + ” lower than 40 %, ”D” lower than 50 % and ”E” higher than 50 %. If no accurancy is available, we usedthe E ( i.e. // kurucz.harvard.edu / LINELISTS / GFALL / )for the gfall.dat linelist. Element λ (Å) log g f Accuracy references Element λ (Å) log g f Accuracy references(Ionization) (Ionization)CI 4371,3670 -1,962 B NIST SiII 4072,7090 -2,367 SL90CI 4932,0490 -1,658 B NIST SiII 4075,4520 -1,403 SL90CI 5052,1670 -1,303 B NIST SiII 4128,0540 0,306 C NISTCI 5380,3370 -1,616 B NIST SiII 4130,8940 0,464 C NISTCI 5793,1200 -2,063 B NIST SiII 4190,7240 -0,351 NISTCI 5800,6020 -2,337 B NIST SiII 4198,1330 -0,611 NISTCI 6587,6100 -1,003 B NIST SiII 4621,4180 -0,540 D KuruczSiII 4621,6960 -1,675 D KuruczOI 3947,2950 -2,095 B NIST SiII 4621,7220 -0,387 D KuruczOI 3947,4810 -2,244 B NIST SiII 5041,0240 0,174 D + NISTOI 3947,5860 -2,467 B NIST SiII 5055,9840 0,441 D + NISTOI 3947,9530 -1,761 B FMW SiII 5056,3170 -0,535 E NISTOI 4368,1930 -2,665 B NIST SiII 5466,4320 -0,190 D KuruczOI 4368,2420 -1,964 B NIST SiII 5669,5630 0,266 NISTOI 4368,2580 -2,818 B NIST SiII 5688,8170 0,106 NISTOI 5329,6730 -2,063 C + NIST SiII 5957,5590 -0,349 D NISTOI 5329,6810 -1,473 C + NIST SiII 5978,9300 -0,061 D NISTOI 5329,6900 -1,268 C + NIST SiII 6347,1100 0,230 C NISTOI 5330,7260 -2,416 C + NIST SiII 6371,3710 -0,080 C NISTOI 5330,7350 -1,570 C + NISTOI 5330,7410 -0,983 C + NIST CaII 3933,6630 -0,135 C NISTOI 6155,9610 -1,363 B NIST CaII 3968,4690 -0,180 C NISTOI 6155,9710 -1,011 B NIST CaII 4472,0500 -2,694 KuruczOI 6155,9890 -1,120 B NIST CaII 4479,4330 -2,994 KuruczOI 6156,7370 -1,487 B NIST CaII 4489,1790 -0,726 KuruczOI 6156,7550 -0,898 B NIST CaII 4489,1790 -2,157 KuruczOI 6156,7780 -0,694 B NIST CaII 4489,1790 -0,613 KuruczOI 6158,1490 -1,841 B NIST CaII 5001,4790 -0,517 D NISTOI 6158,1720 -0,995 B NIST CaII 5019,9710 -0,257 D NISTOI 6158,1870 -0,409 B NIST CaII 5021,1380 -1,217 D NISTCaII 5285,2660 -1,153 D NISTNaI 4494.1800 -1.840 C NIST CaII 5307,2240 -0,853 D NISTNaI 4497.6570 -1.574 B NISTNaI 4668.5590 -1.310 C NIST ScII 4246,8220 0,242 D NISTNaI 4978.5410 -1.210 C NIST ScII 4314,0830 -0,100 D NISTNaI 4982.8130 -0.961 C NIST ScII 4320,7320 -0,250 D NISTNaI 5889.9500 0.112 A NIST ScII 4324,9960 -0,440 D NISTNaI 5895.9240 -0.191 A NIST ScII 4374,4570 -0,418 D NISTNaI 6154.2260 -1.547 A NIST ScII 4670,4070 -0,576 D NISTNaI 6160.7470 -1.230 C NIST ScII 5031,0210 -0,400 D NISTScII 5239,8130 -0,765 D NISTMgII 4384,6370 -0,792 D NIST ScII 5526,7900 0,020 D NISTMgII 4390,5140 -1,706 D NIST ScII 5657,8960 -0,603 D NISTMgII 4390,5720 -0,530 D NIST ScII 6604,6010 -1,310 D NISTMgII 4427,9940 -1,201 C + NISTMgII 4481,1260 0,730 B BL48 TiII 4163,6440 -0,130 D PTPMgII 4481,1500 -0,570 B BL48 TiII 4287,8730 -1,790 PTPMgII 4481,3250 0,575 B BL48 TiII 4290,2190 -0,850 PTPMgI 4702,9910 -0,374 C NIST TiII 4294,0990 -0,930 PTPMgI 5167,3213 -0,856 B NIST TiII 4300,0420 -0,440 D PTPMgI 5172,6844 -0,380 B NIST TiII 4316,7940 -1,420 D KO83MgI 5183,6034 -0,158 B NIST TiII 4386,8440 -0,960 PTPMgI 5528,4050 -0,498 B + TiII 4394,0590 -1,780 PTPTiII 4395,0310 -0,540 A PTPTiII 4399,7720 -1,190 PTP
Table 8. continued.
Element λ (Å) log g f Accuracy references Element λ (Å) log g f Accuracy references(Ionization) (Ionization)TiII 4411,0720 -0,670 D PTP FeII 4233,1720 -2,000 C FMWTiII 4417,7140 -1,190 PTP FeII 4258,1540 -3,400 D FMWTiII 4443,8010 -0,720 D PTP FeII 4273,3260 -3,258 D FMWTiII 4468,4920 -0,600 D FMW FeII 4296,5720 -3,010 D FMWTiII 4488,3250 -0,510 D PTP FeII 4385,3870 -2,570 D FMWTiII 4501,2700 -0,770 D PTP FeII 4416,8300 -2,600 D FMWTiII 4549,6210 -0,470 D PTP FeII 4472,0900 -1,791 KuruczTiII 4571,9710 -0,320 D PTP FeII 4472,6200 -2,340 KuruczTiII 4589,9580 -1,780 D PTP FeII 4472,9290 -3,430 KuruczTiII 4629,2740 -2,240 Kurucz FeII 4491,4050 -2,690 C FMWTiII 4657,2000 -2,240 Kurucz FeII 4508,2880 -2,210 C FMWTiII 4805,0850 -1,100 D PTP FeII 4515,3390 -2,490 C FMWTiII 5129,1560 -1,400 D KO83 FeII 4520,2240 -2,600 C FMWTiII 5188,6870 -1,050 D PTP FeII 4522,6340 -2,030 C FMWTiII 5336,7860 -1,590 A PTP FeII 4541,5240 -3,050 C FMWFeII 4555,8900 -2,290 FMWVII 4475.6700 -1.440 Kurucz FeII 4576,3400 -3,040 FMWVII 4528.5000 -0.960 Kurucz FeII 4582,8350 -3,100 C FMWVII 4532.1700 -0.760 Kurucz FeII 4620,5210 -3,280 D FMWVII 4538.6200 -1.800 Kurucz FeII 4635,3160 -1,650 D FMWVII 4558.4500 -0.930 Kurucz FeII 4656,9810 -3,630 E FMWVII 4564.5900 -1.450 Kurucz FeII 4666,7580 -3,330 D FMWVII 4577.1300 -2.140 Kurucz FeII 4923,9270 -1,320 C FMWVII 4590.5000 -0.780 Kurucz FeII 5197,5770 -2,100 C FMWVII 4600.1800 -1.360 Kurucz FeII 5276,0020 -1,940 C FMWFeII 5316,6150 -1,850 C FMWCrII 4558,6500 -0,660 D Kurucz FeII 5506,1950 0,950 D FMWCrII 4588,1990 -0,643 KuruczCrII 4592,0490 -1,217 D FMW CoI 4466.8800 -0.540 KuruczCrII 4616,6290 -1,291 SL90 CoI 4469.5400 -0.330 KuruczCrII 4618,8030 -1,110 D Kurucz CoI 4471.5400 -0.770 KuruczCrII 4634,0700 -0,990 Kurucz CoI 4530.9500 0.150 KuruczCrII 4812,3370 -1,995 D Kurucz CoI 4533.9800 -0.500 KuruczCrII 5237,3290 -1,160 D FMW CoI 4549.6500 -0.330 KuruczCrII 5308,4400 -1,810 D FMW CoI 4565.5800 -0.220 KuruczCrII 5313,5900 -1,650 D FMW CoI 4581.5900 -0.150 KuruczCrII 5502,0670 -1,990 D FMW CoI 4594.6300 -0.080 KuruczCoI 4596.8900 -0.010 KuruczMnI 4451,5860 0,278 B FMW CoI 4625.7600 -0.370 KuruczMnI 4453,0120 -0,490 C + FMW CoI 4629.3600 -0.190 KuruczMnI 4457,0440 -0,555 C + FMWMnI 4458,2540 0,042 C + FMW NiI 4468.4340 -1.642 KuruczMnI 4461,0790 -0,380 C + FMW NiI 4470.4720 -0.310 D KuruczMnI 4462,0310 0,320 C + FMW NiI 4480.5610 -1.491 KuruczMnI 4464,6820 -0,104 B FMW NiI 4490.0490 -2.108 KuruczMnI 4470,1440 -0,444 B FMW NiI 4490.5250 -2.324 KuruczMnI 4472,8060 -0,583 B FMW NiI 4512.9860 -1.470 D KuruczMnI 4490,0900 -0,521 B FMW NiI 4519.9790 -2.880 D + FMWMnI 4498,9020 -0,343 B FMW NiI 4521.3220 -0.949 KuruczMnI 4502,2130 -0,344 B FMW NiI 4523.6940 -1.305 KuruczMnI 4709,7120 -0,339 B FMW NiI 4528.5260 -1.127 KuruczMnI 4739,0870 -0,490 B FMW NiI 4542.2370 -1.308 KuruczMnI 4754,0420 -0,085 B FMW NiI 4546.9200 -0.271 KuruczMnI 4761,5120 -0,138 B FMW NiI 4551.2170 -0.880 D FMWMnI 4762,3670 0,426 B FMW NiI 4559.9210 -1.737 KuruczMnI 4783,4270 0,042 B FMW NiI 4572.0410 -0.536 KuruczMnI 4823,5240 0,144 B FMW NiI 4588.4110 -0.745 KuruczMnI 5255,3260 -0,763 B FMW NiI 4592.5220 -0.370 KuruczNiI 4596.3830 -0.704 KuruczNiI 4600.3550 -0.610 FMW . Gebran, R. Monier and O. Richard: Chemical composition of A and F dwarf members of the Coma Berenices open cluster 25
Table 8. continued.
Element λ (Å) log g f Accuracy references Element λ (Å) log g f Accuracy references(Ionization) (Ionization)NiI 4604.9820 -0.250 D Kurucz SrII 4077.7090 0.151 NISTNiI 4606.2190 -1.000 D FMW SrII 4215.5200 -0.169 NISTNiI 4609.9050 -0.580 KuruczNiI 4617.8620 -0.525 Kurucz YII 4883.6840 0.070 kuruczNiI 4631.0170 -0.957 Kurucz YII 4900.1200 -0.09 kuruczNiI 4648.6460 -0.100 D Kurucz YII 4982.1290 -1.290 kuruczNiI 4886.7050 -1.780 Kurucz YII 5087.4160 -0.170 kuruczNiI 4886.9760 -1.120 Kurucz YII 5200.4060 -0.570 kuruczNiI 4900.9670 -1.670 E FMWNiI 4904.4070 -0.170 D FMW ZrII 4149.2170 -0.030 kuruczNiI 4912.0200 -0.800 D FMW ZrII 4156.2400 -0.776 kuruczNiI 4913.9680 -0.630 D FMW ZrII 4161.2100 -0.720 kuruczNiI 4918.3620 -0.240 D FMW ZrII 4208.9800 -0.460 kuruczNiI 4918.7060 -0.780 Kurucz ZrII 4496.9600 -0.810 Bal81NiI 4925.5590 -0.770 D KuruczNiI 4935.8310 -0.350 D FMW BaII 4554,0290 0,163 B KuruczNiI 4937.3410 -0.390 D FMW BaII 4934,0760 -0,156 B NBSNiI 4976.3260 -3.100 C + NIST BaII 5853,6680 -1,510 B NISTNiI 5003.7410 -2.800 C + NIST BaII 6141,7130 -0,810 B NISTNiI 5080.5280 0.330 kurucz BaII 6496,8970 -1,010 B NISTNiI 5081.1070 0.300 kuruczNiI 5084.0890 0.090 kuruczNiI 5096.8540 -0.900 kuruczNiI 5099.9270 -0.100 kuruczNiI 5102.9660 -2.620 C + NISTNiI 5137.0740 -1.990 C + NISTNiI 5476.9040 -0.890 C + NISTNiI 5553.6900 -3.240 C + NISTNiI 5587.8580 -2.140 C + NISTNiI 5592.2620 -2.590 C + NISTNiI 5711.8880 -2.260 C + NISTNiI 5754.6560 -2.340 C + NISTNiI 5892.8720 -2.340 C ++