20 pages

Late stages of the evolution of A-type stars on the main sequence: comparison between observed chemical abundances and diffusion models for 8 Am stars of the Praesepe cluster

of 20
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Late stages of the evolution of A-type stars on the main sequence: comparison between observed chemical abundances and diffusion models for 8 Am stars of the Praesepe cluster
  Astronomy & Astrophysics  manuscript no. Gpraesepe c  ESO 2007October 2, 2007 Late stages of the evolution of A-type stars on the main sequence:comparison between observed chemical abundances and diffusionmodels for 8 Am stars of the Praesepe cluster L. Fossati 1 , S. Bagnulo 2 , R. Monier 3 , S. A. Khan 4 , O. Kochukhov 5 , J. Landstreet 4 , G. Wade 6 , and W. Weiss 1 1 Institut fur Astronomie, Universit¨at Wien, T¨urkenschanzstrasse 17, 1180 Wien, Austria.e-mail:; 2 Armagh Observatory, College Hill, Armagh BT61 9DG, Northern Ireland. e-mail: 3 Groupe de Recherches en Astronomie et Astrophysique du Languedoc UMR 5024, Universit´e Montpellier II, Place E. Bataillon,34095 Montpellier, France. e-mail: 4 Department of Physics & Astronomy, University of Western Ontario, London, N6A 3K7, Ontario, Canada.e-mail:; 5 Department of Astronomy and Space Physics, Uppsala University, 751 20, Uppsala, Sweden. e-mail: 6 Physics Dept., Royal Military College of Canada, PO Box 17000, Station Forces, K7K 4B4, Kingston, Canada. e-mail: Received  /  Accepted ABSTRACT Aims.  We aim to provide observational constraints on di ff  usion models that predict peculiar chemical abundances in the atmospheresof Am stars. We also intend to check if chemical peculiarities and slow rotation can be explained by the presence of a weak magneticfield. Methods.  We have obtained high resolution, high signal-to-noise ratio spectra of eight previously-classified Am stars, two normalA-type stars and one Blue Straggler, considered to be members of the Praesepe cluster. For all of these stars we have determinedfundamental parameters and photospheric abundances for a large number of chemical elements, with a higher precision than wasever obtained before for this cluster. For seven of these stars we also obtained spectra in circular polarization and applied the LSDtechnique to constrain the longitudinal magnetic field. Results.  No magnetic field was detected in any of the analysed stars. HD 73666, a Blue Straggler previously considered as an Ap(Si)star, turns out to have the abundances of a normal A-type star. Am classification is not confirmed for HD 72942. For HD 73709 wehave also calculated synthetic  ∆ a  photometry that is in good agreement with the observations. There is a generally good agreementbetween abundance predictions of di ff  usion models and values that we have obtained for the remaining Am stars. However, theobserved Na and S abundances deviate from the predictions by 0.6 dex and  ≥ 0.25 dex respectively. Li appears to be overabundant inthree stars of our sample. Key words.  stars: abundances – stars: atmospheres – stars: fundamental parameters – stars: chemically peculiar 1. Introduction Main sequence A-type stars present spectral peculiarities, usu-ally interpreted as due to peculiar photospheric abundances andabundance distributions which are believed to be produced bythe interaction of a large variety of physical processes (e.g. di ff  u-sion, magnetic field, pulsation and various kinds of mixing pro-cesses).An interesting problem that has yet to be addressed is howthese peculiarities change during main sequence evolution. Thechemical composition of field A-type stars have been studiedby several authors, e.g. Hill & Landstreet (1993), Adelman etal. (2000). However, it is not straightforward to use the re-sults of these investigations to study how photospheric chem-istry evolves during a star’s main sequence life. First, the src-inal composition of the cloud from which stars were born isnot known and is likely somewhat di ff  erent for each field star.It is therefore not possible to discriminate between evolution-ary e ff  ects and di ff  erences due to srcinal chemical composition. Send o  ff   print requests to : L. Fossati Secondly, it is di ffi cult to estimate the age of field stars with theprecision necessary for such evolutionary studies (for a discus-sion of this problem see Bagnulo et al. 2006).From this point of view, A-type stars belonging to open clus-ters are much more interesting objects. Compared to field stars,A-type stars in open clusters have three very interesting proper-ties:- they were all presumably born from the interstellar gas withan approximately uniform composition;- they all have approximately the same age (to within a fewMyr);- their age can be determined much more precisely than forfield stars.Few abundance analyses of A-type stars in open clustershave been carried out. Those that have been published have usu-ally focused on a limited numbers of stars.Varenne & Monier (1999) have determined the abundancesof eleven chemical elements for a large sample of stars regularlydistributed in spectral type along the main sequence in order to  2 L. Fossati et al.: Chemical abundance analysis of A-type stars in the Praesepe cluster sample the expected masses uniformly. All these stars were anal-ysed in a uniform manner using spectrum synthesis. St¨utz et al.(2006) have performed a detailed abundance analysis for fiveA-type stars of the young open cluster IC 2391. Folsom et al.(2007) have performed a detailed abundance analysis for fourAp  /  Bp stars and one normal late B-type star of the open clusterNGC 6475.A goal of this programme is to determine photospheric abun-dance patterns in A-type star members of clusters of di ff  erentages. This is crucial in order to:  i)  investigate the chemical di ff  er-ences between normal and peculiar stars inside the same cluster, ii)  study the evolution with time of abundance peculiarities bystudying clusters of various ages,  iii)  set constraints on the hy-drodynamical processes occurring at the base of the convectionzone in the non magnetic stars and  iv)  study the e ff  ects of dif-fusion in the presence of a magnetic field for the magnetic (Ap)stars in the cluster. The abundance analysis will be performed inan homogeneous way applying a method described in this firstwork.Praesepe (NGC 2632), a nearby intermediate-age open clus-ter (log t   =  8 . 85  ±  0 . 15, Gonz´alez-Garc´ıa et al. 2006), is an es-pecially interesting target because it includes a large number of A-type stars, among which are many Am stars. Furthermore, be-cause the cluster is relatively close to the sun (d  =  180  ±  10pc, Robichon et al. 1999), many of the member A-type stars arebright enough to allow us to obtain high resolution spectra withintermediate class telescopes.We dedicate this first paper to the Am stars of the Praesepecluster, searching for magnetic fields in these objects and dis-cussing the di ff  erences between ”normal” 1 A-type stars and Amstars in the cluster.We also compare our results with previous works and withtheoretical chemical evolution models. In particular we take intoaccount di ff  usion models by Richer et al. (2000). We want toprovide observational constraints to the theory of the evolutionof the abundances in normal and chemically peculiar stars. Ourdetailed abundance analysis could provide information about theturbulence occurring in the outer stellar regions in Am stars withwell determined age. In particular, our analysis can give con-straints to define the depth of the zone mixed by turbulence,since it is the only parameter characterising turbulence (Richeret al. 2000). A systematic abundance analysis of normal and pe-culiar stars in clusters could provide information on the srcinof the mixing process and show if only turbulence is needed toexplain abundance anomalies, or if other hydrodynamical pro-cesses occur.We tackle this problem using new and more precise new-generation spectrographs providing a wider wavelength cover-age together with newer analysis codes and procedures (e.g.Least-SquaresDeconvolutionandsyntheticlineprofilefittingin-stead of equivalent width measurements.)The observed stars, the instruments employed and the targetselection are described in Sect. 2. The data reduction and a dis-cussionofthecontinuumnormalisationareprovidedinSect.2.3.In Sect. 3 and 4 we describe the models and the procedure usedto perform the abundance analysis. Our results are summarisedin Sect. 5. Discussion and conclusions are given in Sect. 6 and 7respectively. 1 We consider as ”normal” A-type stars all the A-type stars that areclassified neither as Am nor Ap 2. Observations and data reduction 2.1. Instruments  We observed six stars of the Praesepe cluster using theESPaDOnS (Echelle SpectroPolarimetric Device forObservations of Stars) spectropolarimeter at the Canada-France-Hawaii Telescope (CFHT) from January 8th to 10th2006. Spectra were acquired in circular polarisation.Spectra of an additional four stars were obtained with theELODIE spectrograph at the Observatoire de Haute Provence(OHP) from January 4th to 6th 2004.An additional circular polarisation spectrum of HD 73709was obtained with the MuSiCoS spectropolarimeter at the 2-mBernard Lyot Telescope (TBL) of the Pic du Midi Observatoryon 7 March 2000. The magnetic field derived form this observa-tion has been discussed by Shorlin et al. (2002). 2.1.1. ESPaDOnS ESPaDOnS consists of a table-top cross-dispersed echellespectrograph fed via a double optical fiber directly from aCassegrain-mounted polarization analysis module. In ”polari-metric” mode, the instrument can acquire a Stokes  V  , Q  or  U  stellar spectrum throughout the spectral range 3700 to 10400 Åwith a resolving power of about 65000. A complete polarimet-ric observation consists of a sequence of 4 sub-exposures, be-tween which the retarder is rotated by  ±  90 degrees (Donati etal. 1997; Wade et al. 2000). In addition to the stellar exposures, asingle bias spectrum and ThAr wavelength calibration spectrum,as well as a series of flat-field exposures, were obtained at thebeginning and end of each night. 2.1.2. ELODIE ELODIE is a cross-dispersed echelle spectrograph at the 1.93-m telescope at the OHP observatory. Light from the Cassegrainfocus is fed into the spectrograph through a pair of optical fibers.Two focal-plane apertures are available (both 2 arc-sec wide),one of which is used for starlight and the other can be used foreither the sky background or the wavelength calibration lamp,but can also be masked. The spectra cover a 3000 Å wavelengthrange (3850–6800 Å) with a mean spectral resolution of 42000.ELODIE was decommissioned in mid-August 2006. 2.1.3. MuSiCoS MuSiCoS, like ESPaDOnS, consists of a table-top cross-dispersed echelle spectrograph fed via a double optical fiber di-rectly from a Cassegrain-mounted polarization analysis module.MuSiCoS provides a Stokes  V  ,  Q  or  U   stellar spectrum from4500 to 6600 Å with a mean resolving power of 35000. A moredetailed description of the instrument and of the observing pro-cedures are reported by Donati et al. (1999). MuSiCoSwas de-comissioned in December 2006. 2.2. Target selection  The target selection for the run with the ESPaDOnS spectro-graph was performed taking into account previously-publishedpeculiar spectral classifications of the stars, together with their υ sin i  (if any), giving priority to slowly-rotating stars. Data con-cerning stars of the cluster were collected from the WEBDA  L. Fossati et al.: Chemical abundance analysis of A-type stars in the Praesepe cluster 3 database 2 (Mermillod & Paunzen 2003) and the SIMBADdatabase, operated at CDS, Strasbourg, France.The stars observed with the ELODIE spectrograph were thebright stars included in the analysis provided by Burkhart &Coupry (1998), to allow a comparison with their work.The star observed with MuSiCoS was analysed by Shorlinet al. (2002), with the LSD technique, to measure the magneticfield strength.Four stars of the sample were accepted as cluster mem-bers by the HIPPARCOS survey (Robichon et al. 1999) whereasthe others have been confirmed as members by di ff  erent stud-ies, such as those by Kharchenko et al. (2004) and Wang et al.(1995).The complete sample of stars observed and analysed in thispaper is listed in Table 1. Seven of the stars are spectroscopicbinaries and one is a  δ  Scuti star. Of the eleven stars observed,eight were previously classified as Am stars, two as normal A-type stars and one as an Ap(Si) star. 2.3. Data reduction  The ESPaDOnS spectra were reduced using the Libre-ESpRITpackage  3 .The ELODIE spectra were automatically reduced by a stan-dard data reduction pipeline described by Baranne et al. (1996).The MuSiCoS spectrum was reduced according to the pro-cedure described by Wade et al. (2000) and Shorlin et al. (2002).The sample of stars includes objects with a high  υ sin i  (up to   130kms − 1 ) for which the continuum normalisation is a criti-cal reduction procedure. For this reason, all of the spectra werenormalised without the use of any automatic continuum fittingprocedure. We considered the single echelle orders of the spec-tra, which were normalised and then merged. It was not possibleto determine a correct continuum level short wards of the H γ   line(4340.462 Å), as there were not enough continuum windows inthe spectrum at these shorten wavelengths. 3. Calculation of model atmospheres Model atmospheres were calculated with the LTE codeLL   (version 8.4), which uses direct sampling of the lineopacity (Shulyak et al. 2004), and allows the computation of model atmospheres with individualised (not scaled solar) abun-dance patterns. This allows us to compute self-consistent modelatmospheres that match the actual abundances of chemically pe-culiar stars and to thereby minimise systematic errors (Khan &Shulyak 2007).We used the VALD database (Piskunov et al. 1995; Kupkaet al. 1999; Ryabchikova et al. 1999) as a source of spectralline data, including lines that srcinate from predicted levels.We then performed a preselection procedure to eliminate thoselines that do not contribute significantly to the line opacity.For this procedure we utilised model atmospheres calculated byATLAS9 (Kurucz 1993a) with fundamental parameters corre-sponding to each star in our sample. Because at this stage of theanalysis we did not know the photospheric abundances of thesample stars, we employed the Opacity Distribution Function(ODF) tables for solar abundances (Kurucz 1993b). The line se-lection criterion required that the line-to-continuum opacity ratio 2 3 See also Donati et al. (1997) at the center of each line, at any atmospheric depth, be greaterthan 0.05 %.To compute atmosphere models with individual abundancepatterns, we used an iterative procedure. The initial model at-mosphere was calculated with the solar abundances taken fromAsplund et al. (2005). Then, these abundances were modified ac-cordingtotheresultsofspectroscopicanalysis,andafinal modelatmosphere was iterated.A logarithmic Rosseland optical depth scale log τ ross  wasadopted as an independent variable of atmospheric depth span-ning from + 2 to − 6 . 875 and subdivided into 72 layers. Opacitieswere sampled with a 0.1Å wavelength step. Since  a posteri-ori  we found no magnetic field in the atmospheres of our stars(Sect. 6.1), the excess line blanketing due to a magnetic field(Kochukhov et al. 2005; Khan & Shulyak 2006a,b) was ne-glected. The value of the microturbulence velocity  υ mic  wasadopted according to the results of the spectroscopic analysis(Sect. 4.2).Convection was treated according to the CM approach(Canuto & Mazzitelli 1992). It has been argued that CM shouldbe preferred over the MLT approach (Kupka 1996). For exam-ple, Smalley & Kupka (1997) showed that the CM models giveresults that are generally superior to standard MLT ( α  =  1 . 25)models. Furthermore, to calculate a part of the NEMO model at-mosphere grid, Heiter et al. (2002) adopted the free parameter α  =  0 . 5 for the MLT approach as it produced results quite sim-ilar to those of the CM method. Consequently, we have takenconvection (CM approach) into account for the whole set of cal-culations to ensure correct modelling for lower e ff  ective temper-atures, as many of the stars analysed in this paper have e ff  ec-tive temperatures below 9000K. To test the importance of theconvection treatment on the abundance analysis, as applied tothis sample of stars, we performed several numerical tests usingmodel atmospheres calculated with no convection treatment atall. The results revealed di ff  erences of about 0.005dex, clearlywithin the errors associated with the analysis. 4. Spectral analysis 4.1. High precision search for magnetic field  One of the main goals of our analysis is to search for weakmagnetic fields in the Am stars of Praesepe, and to check if the Ap(Si) star HD 73666 has one, since many Ap stars showa strong magnetic field. For these reasons, we observed someAm stars and HD 73666 with the ESPaDOnS spectrograph thatprovides the opportunity to obtain high resolution spectra in cir-cular polarization. To detect the presence of a magnetic fieldand to infer the longitudinal magnetic field we used the Least-Squares Deconvolution technique (hereafter LSD). LSD is across-correlationtechniquedeveloped for thedetectionand mea-surement of weak polarization signatures in stellar spectral lines.The method is described in detail by Donati et al. (1997) andWade et al. (2000). 4.2. Atmosphere fundamental parameters  The initial values of the e ff  ective temperature ( T  e ff  ), surfacegravity (log g ) and metallicity of our sample of stars were es-timated via Str¨omgren photometry (Hauck & Mermilliod 1998).The initial value of the microturbulence velocity ( υ mic ) wasdetermined using the following relation (Pace et al. 2006): υ mic  =  − 4 . 7log( T  e ff  ) + 20 . 9 kms − 1 .  (1)  4 L. Fossati et al.: Chemical abundance analysis of A-type stars in the Praesepe cluster Table 1.  Basic datas of the observations for the program stars.  1 : positions from Perryman et al. (1997);  2 : positions from Hog et al. (1998). TheSNR are calculated at  ∼ 5000 Å in a bin of 0.5 Å. The exposure time is in seconds. With the ESPaDOnS and MuSiCoS instruments we obtainedStokes  I   and  V   spectra, which allowed us to attempt magnetic field detection. With ELODIE we obtained Stokes  I   spectra only. The HJD indicatethe Heliocentric Julian Date at the middle of the exposure. HD RA DEC HJD M v  Spectral Type Instrument Resolution SNR Exp. Time Remarks73430 08 39 03.585  + 19 59 59.08  2 2453746.125 8.33 A9V ESPaDOnS 65000 220 180073575 08 39 42.6548  + 19 46 42.440  1 2453747.116 6.66 F0III ESPaDOnS 65000 250 2400  δ Sct variable73666 08 40 11.4528  + 19 58 16.073  1 2453745.063 6.61 Ap(Si) ESPaDOnS 65000 660 1600 SB1, Blue Straggler72942 08 36 17.4422  + 20 20 29.421  1 2453746.098 7.48 Am ESPaDOnS 65000 350 1600 SB?73045 08 36 48.0033  + 18 52 58.111  1 2453745.091 6.82 Am ESPaDOnS 65000 290 2400 SB173730 08 40 23.466  + 19 50 05.91  2 2453745.123 7.99 Am ESPaDOnS 65000 330 200073618 08 39 56.496  + 19 33 10.76  2 2453009.612 7.30 Am ELODIE 42000 160 3600 SB173174 08 37 36.995  + 19 43 58.48  2 2453010.574 7.76 Am ELODIE 42000 120 2700 SB1, triple system73711 08 40 18.099  + 19 31 55.17  2 2453010.432 7.51 Am ELODIE 42000 90 3600 SB173818 08 40 56.935  + 19 56 05.47  2 2453012.102 8.69 Am ELODIE 42000 80 2400 SB173709 08 40 20.748  + 19 41 12.24  2 2451611.473 7.68 Am MuSiCoS 35000 120 2400 SB1, quadruple system The uncertainties associated with  T  e ff  , log g  and  υ mic  determinedphotometrically and from Eq. (1) are quite large (about 300Kin  T  e ff   and 0.25 in log g , due to uncertainties in the photometricmeasurements and intrinsic scatter in the calibrations). We haveadjusted these parameters spectroscopically to get more accuratevalues for an abundance analysis.The best value of the e ff  ective temperature was estimatedusing the abundance–  χ excit  correlation, calculated fitting theabundances of di ff  erent selected lines of an element, in theabundance–excitation potential plane. This correlation is sensi-tive to e ff  ective temperature variations. This property allows usto determine the best value of the e ff  ective temperature, whichwe derived by eliminating the abundance–  χ excit  correlation. Forthis step, the abundances were calculated using a modified ver-sion (Tsymbal 1996) of the WIDTH9 code (Kurucz 1993a), us-ing equivalenth widths for the more slowly-rotating stars ( υ sin i <  30 kms − 1 ), and by fitting line cores (as described in 4.3) forthe rapid rotators.In a similar way, the best value for log g  was found by elimi-nating any systematic di ff  erence in abundance derived from dif-ferent ionisation stages of the same chemical element.Finally, for stars with  υ sin i  less than 30 kms − 1 , the  υ mic value was determined by eliminating any abundance–equivalentwidth correlation.To check the degree of refinement reached with the spec-troscopic method, we systematically performed the followingcheck. For each star with  υ sin i  <  30 kms − 1 we have performedthe abundance analsysis of Fe   and Fe   lines using three dif-ferent approaches, i.e., by calculating log g  and  T  e ff   from (i)Geneva photometry, (ii) from Str¨omgren photometry, and (iii)with the spectroscopic method outlined above. For each of thesethree models, we have calculated the abundances of Fe   and Fe  lines for  υ mic  = 0, 1, 2, ..., 6 kms − 1 . Finally, for each of these 18sets of   T  e ff  , log g  and  υ mic  values, we have calculated the stan-dard deviation from the mean Fe   and Fe   values. In all caseswe found that the parameters that we have identified with themethod described above are those that minimise the abundancescatter.In Fig. 1 we give an example of the result of our check, forthe Am star HD 73730. The result of the check for all the otherstars of the sample is similar to the one shown in Fig. 1.For faster rotators an insu ffi cient number of unblended linesare available with which to calculate the equivalent width. Themore slowly-rotating Am stars of the sample show a mean  υ mic value around 2.7 kms − 1 . For this reason, for fast rotators, weretained the value found using Eq. (1), which was always com-patible, within the errors, with 2.7 kms − 1 . 0123456 microturbulence velocity   s   t  a  n   d  a  r   d   d  e  v   i  a   t   i  o  n   (   F  e   1   &   2   l   i  n  e  s   )   Parameters from Geneva photometryParameters from Stroemgren photometryParameters from Spectroscopy Fig.1.  Standard deviation for Fe   and Fe   lines, selected for HD 73730,as a function of   υ mic  using fundamental parameters obtained fromGeneva photometry (straight line), from Str¨omgren photometry (dottedline) and from spectroscopy (dashed line). Following Landstreet (1998) it has been generally adoptedthat for Am stars  υ mic  is between 4–5 kms − 1 . By contrast in thepresent work we determined values between 2.3–3.1 kms − 1 , forthe slowly rotating stars.The chemical elements used to determine the fundamentalparameters were di ff  erent for each star, depending mainly on thevalue of   υ sin i , and are described in Sect. 5.For the lines used to eliminate the correlations discussedabove we assumed that non-LTE and stratification e ff  ects werenegligible (Ryabchikova et al. 2007).The wavelength range of the ESPaDOnS orders was wideenough to perform a safe continuum normalisation of the H α and H  β  lines. For the spectra obtained with this instrument it wastherefore possible to use these two hydrogen lines to provide anadditional constraint on the fundamental parameters. In particu-lar, for e ff  ective temperatures less than 8000K, the wings of theHydrogen lines are very sensitive to the e ff  ective temperature.On the other hand, at higher e ff  ective temperatures the wings areparticularly sensitive to log g . We did not use the hydrogen linesof the spectra obtained with ELODIE and MuSiCoS because thewavelength ranges of the orders containing the H α  and H  β  lineswere too narrow to include the entire line profile.  L. Fossati et al.: Chemical abundance analysis of A-type stars in the Praesepe cluster 5 The spectroscopic procedure of refining the fundamental pa-rameters reduces the errors of   T  e ff   and log g  to typically 200Kand 0.2, respectively, while the estimated error for  υ mic  is around0.2kms − 1 . These errors are estimated taking into account di ff  er-ent noise sources, including continuum normalisation errors anduncertainty in the convective model. The errors associated withthe e ff  ective temperature and log g  have been confirmed by a fitof the H α  and H  β  line profiles using the models calculated withthe parameters obtained from the estimated errors.In some cases a star’s high rotational velocity prevented usfrom recovering  T  e ff   and log g  spectroscopically with good pre-cision. In these cases we ultimately adopted an average of pa-rameters determined spectroscopically and those obtained viaStr¨omgren photometry.For the spectra obtained with the ELODIE spectrograph wealso attempted to check for the presence of a magnetic field bylooking for a correlation between the abundance derived fromeach line and its Land´e factor. We also conducted a search formagnetically-split lines (e.g. Fe   line at 6149.258 Å) in spec-tra of the very slowest rotators. In fact, for  υ sin i  more thana few kms − 1 this method cannot be used for detecting typi-cal Ap-star magnetic fields. For the spectra obtained with theESPaDOnS spectrograph we used the LSD technique to mea-sure the velocity-resolved Stokes  V   profile and the longitudinalmagnetic field.With the correct fundamental parameters, we determined thebest value of   υ sin i  fitting it on the Fe   lines at 5434.524 Å and5576.089 Å. We chose these two Fe   lines because their broad-ening parameters are well known and are not a ff  ected by mag-netic field broadening, since their Land´e factor is close to 0. Theestimated error on  υ sin i  is about 5%.Theradial velocity,inkms − 1 ,was determinedbyperformingthe median of the results obtained by fitting several lines of theobserved spectrum to a synthetic one. 4.3. Element abundance analysis  Once we had obtained the best values for the fundamental pa-rameters, we determined the final elemental abundances by di-rect fitting of the observed spectra with synthetic models. Wesynthesised the model spectra with Synth3 (Kochukhov 2006)and fit the cores of selected lines to get a value of the abundanceassociated with each line. We then calculated the mean and therelative standard deviation for each analysed element. The linecore fitting was performed with the ’Lispan’ and ’ATC’ codes(written by Ch. St¨utz) together with Synth3. The only free pa-rameter in the line core fitting procedure is the abundance of theline. The fitting procedure and the determination of the abun-dances was performed iteratively in order to obtained a betterdetermination of the abundances for blended lines. The error as-sociated with the derived abundance of each element is the stan-dard deviation from the mean abundance of the selected lines of that element. These errors do not take into account the uncer-tainties of the fundamental parameters and of the adopted modelatmosphere. To have an idea of these uncerteinties, we have cal-culated the abundances of Fe, Ti and Ni for HD 73730 with fivedi ff  erent models varing  T  e ff   of   ± 200 K and log g  of   ± 0.2 dexfrom the adopted model. We found a variation of less than 0.2dex for Fe, 0.1 dex for Ti and 0.1 for Ni due to the temperaturevaration. We did not find any significant abundance change var-ing log g . The uncertainty in temperature is probably the mainerror source on our abundances.The lines used to synthesise the spectra and selected to cal-culate the abundances for the various elements were extractedfrom the VALD database.The number of selected lines depends on the  υ sin i  valuefound for each star. We also checked the log gf   value of each se-lected line using the solar spectrum. Lines in the synthetic spec-trumoftheSunshowinglargedeviationsfromtheobservedsolarspectrum were rejected from the selection. For some elements(e.g. Co, Cu and Zn) we derived the abundance from one line.This does not mean that only one line is present in the spectrum,but that only one line was selected to derive the chemical abun-dance. Other lines of the same element are present, but they werenot selected because they are either too much blended or theirlog gf   value appeared to be in error based on the solar spectrum.For the elements which were not analysed we assumed solarabundances from Asplund et al. (2005). 4.4. Non-LTE effects  Some potassium lines were visible in the ESPaDOnS spectra(because of the extension of the spectra into the far red), makingpossible the calculation of the abundance of this element. In par-ticular the K   line at 7698.974Å was detected and was largelyunblended in the spectra of most stars. Unfortunately, this ele-mentpresentsstrongnon-LTEe ff  ects;forthisreasontheKabun-dances derived here are reported only as upper limits. The He, Oand Na lines were selected in such a way to reduce as much aspossible non-LTE e ff  ects. For He, we selected lines in the blueregion; for O, lines in the red region; and for Na, we used theNa   doublet at 5682 . 633  −  5688 . 205Å, which is known to beessentially une ff  ected by non-LTE e ff  ects (Ryabchikova, privatecommunication). For all the other elements that present impor-tant non-LTE e ff  ects (e.g. C, N, Al, S, Y, Ba), the abundances de-rived here should be considered as upper limits (Baumueller &Gehren 1997; Kamp et al. 2001; Przybilla et al. 2006, and refer-ences therein). However at present day it is not known how largeare the deviations from LTE for normal A-type stars and Amstars for the various elements. More non-LTE analysis should beperformed for this type of star. Since our sample of stars has sim-ilar fundamental parameters (mainly log g  and metallicity) thenon-LTE e ff  ects associated to each star should be of the samemagnitude. This allows us to compare the stars of our samplewith each other. 5. Results We have organised our sample of program stars according totheir literature classification as normal A-type stars (HD 73430,HD 73575), Blue Stragglers (HD 73666 – Ap(Si) star) and Amstars (HD 72942, HD 73045, HD 73730, HD 73618, HD 73174,HD 73711, HD 73818, HD 73709).Table 2 lists the initial and final set of fundamental parame-ters obtained for each program star.The derived abundances are illustrated in Fig. 13 to Fig. 22,and reported in Table 3.In the following sections we comment on individual stars,devoting special attention to the elements that characterise thepeculiarities of each star.
Related Documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks