Absolute dimensions of detached eclipsing binaries. III. The metallic-lined system YZ Cassiopeiae
aa r X i v : . [ a s t r o - ph . S R ] N ov Mon. Not. R. Astron. Soc. , 000–000 (0000) Printed 22 September 2018 (MN L A TEX style file v2.2)
Absolute dimensions of detached eclipsing binaries. III. Themetallic-lined system YZ Cassiopeiae
K. Pavlovski, J. Southworth, V. Kolbas and B. Smalley Department of Physics, University of Zagreb, Bijeniˇcka cesta 32, 10000 Zagreb, Croatia Astrophysics Group, Keele University, Newcastle-under-Lyme, ST5 5BG, UK
22 September 2018
ABSTRACT
The bright binary system YZ Cassiopeiae is a remarkable laboratory for studying the Amphenomenon. It consists of a metallic-lined A2 star and an F2 dwarf on a circular orbit, whichundergo total and annular eclipses. We present an analysis of 15 published light curves and42 new high-quality ´echelle spectra, resulting in measurements of the masses, radii, effectivetemperatures and photospheric chemical abundances of the two stars. The masses and radiiare measured to 0.5% precision: M A = 2 . ± .
012 M ⊙ , M B = 1 . ± .
007 M ⊙ , R A = 2 . ± .
011 R ⊙ and R B = 1 . ± .
006 R ⊙ . We determine the abundance of 20elements for the primary star, of which all except scandium are super-solar by up to 1 dex. Thetemperature of this star ( ± K) makes it one of the hottest Am stars. We also measurethe abundances of 25 elements for its companion ( T eff = 6880 ± K), finding all to besolar or slighly above solar. The photospheric abundances of the secondary star should berepresentative of the bulk composition of both stars. Theoretical stellar evolutionary modelsare unable to match these properties: the masses, radii and temperatures imply a half-solarchemical composition ( Z = 0 . ± . ) and an age of 490–550 Myr. YZ Cas thereforepresents a challenge to stellar evolutionary theory. Key words: stars: fundamental parameters — stars: binaries: eclipsing — stars: binaries:spectroscopic — stars: abundances — stars: chemically peculiar — stars: individual: YZ Cas
Eclipsing binary star systems are of fundamental importance to as-trophysics as they are the primary source of direct measurementsof the physical properties of stars (Andersen 1991; Torres et al.2010). The masses and radii of double-lined systems can be mea-sured to precisions of better than 1% using only geometric argu-ments (e.g. Clausen et al. 2008). Obtaining effective temperature( T eff ) measurements converts them into excellent distance indica-tors (Ribas et al. 2005; Bonanos et al. 2006). Detached eclipsingbinaries (dEBs) hold an additional advantage as their propertiescan be used to investigate the predictive abilities of evolutionarymodels for single stars.Metallic-lined A stars (Am stars) are defined to be those whichshow unusually weak Ca and/or Sc lines and overly strong metalliclines for their spectral type as derived from Balmer line profiles(Conti 1970; Gray & Corbally 2009). They were first describedas a class of chemically peculiar stars by Titus & Morgan (1940).Their abundance anomalies are explained as the product of chemi-cal stratification caused by radiative levitation and gravitational set-tling (Michaud 1970; Turcotte et al. 2000; Talon et al. 2006). Pul-sations have recently been found in many Am stars (Smalley et al.2011; Balona et al. 2011).These physical phenomena can produce observable effects in quiet radiative atmospheres. The Am phenomenon occurs onlyin stars with rotational velocities slower than about 100 km s − (Abt & Levy 1985; Budaj 1996, 1997). Am stars are preferentiallyfound in binary systems, where tidal effects cause a slower ro-tation than for single stars (Abt 1961, 1965; Carquillat & Prieur2007). This means that they are well-represented in the knownpopulation of dEBs , for example V364 Lac (Torres et al. 1999),V459 Cas (Lacy et al. 2004), WW Cam (Lacy et al. 2002) andRR Lyn (Tomkin & Fekel 2006). Whilst the Am phenomenon is a‘surface disease’, a very high bulk metal abundance was found forthe metallic-lined dEB WW Aur by Southworth et al. (2005c).In this work we present a detailed analysis of the dEB YZ Cas,a system which consists of a metallic-lined A2 star and an F2 dwarfof significantly lower mass, radius and T eff . This system providesan opportunity to study the Am phenomenon in a star for whichthe internal chemical composition can be inferred – from its lower-mass companion. YZ Cas is particularly well suited to such an anal-ysis. It is totally eclipsing and extensive photometric data is avail-able in the literature, so the radii of the stars can be determinedto unusually high precision. It is also bright and contains slowly A catalogue of well-studied detached eclipsing binaries is maintained at ∼ jkt/debdata/debs.html c (cid:13) Pavlovski et al.
Table 1.
Identifications, location, and combined photometric indices forYZ Cassiopeiae. YZ Cassiopeiae ReferenceFlamsteed designation 21 Cas 1Hipparcos number HIP 3572 2Henry Draper number HD 4161 3Bright Star Catalogue HR 192 4Bonner Durchmusterung BD +74 27 5 α
00 45 39.078 2 δ +74 59 17.06 2Hipparcos parallax ( m as) 11.24 ± B T ± V T ± J ± H ± K ± b − y ± v − b ± u − b ± References: (1) Flamsteed (1712); (2) Perryman et al. (1997); (3)Cannon & Pickering (1918); (4) Hoffleit & Jaschek (1991); (5) Argelander(1903); (6) This work; (7) 2MASS (Cutri et al. 2003); (8) Hilditch & Hill(1975), given as the mean and standard deviation of the six measurementsfor each Str¨omgren colour index (all taken outside eclipse). rotating stars, making spectroscopic analysis especially productive.Below we recap the long observational history of this object, modelthe available light curves, analyse new ´echelle spectroscopy of thesystem, deduce the physical properties of the two stars, and finallyconfront theoretical models with our results.
YZ Cas shows total and annular eclipses recurring on an orbitalperiod of 4.47 d. It has been the subject of many analyses overnearly a century, which we summarise below. Throughout thiswork we refer to the primary component as star A and the sec-ondary component as star B. Star A is substantially hotter, largerand more massive than star B, and is the component in inferior con-junction at the midpoint of primary eclipse. The spectral type ofthe YZ Cas system is given as A2 in the
Henry Draper Catalogue (Cannon & Pickering 1918); this measurement pertains to star A,which dominates the light of the system.The eclipsing nature of YZ Cas was announced by Stebbins(1924), who is credited with the discovery, and further elaboratedby Huffer (1925). It was chosen at the Washburn Observatory asa photometric comparison star for 23 Cas (a spectroscopic binarywhich has not been observed to be variable). The resulting lightcurve was presented by Huffer (1928).Plaskett (1926) determined the first spectroscopic orbit of thesystem, based on radial velocity (RV) measurements of star A. Un-der the assumption that both stars had a mass of 2 M ⊙ he measuredtheir radii and orbital separation. Plaskett detected the Rossiter-McLaughlin effect (Rossiter 1924; McLaughlin 1924) during pri-mary minimum, with an amplitude slightly greater than 3 km s − .The shape of the Rossiter-McLaughlin anomaly is consistent withalignment between the orbital and stellar rotational axes, albeit withlow significance.Kron (1939b) used the annular nature of secondary eclipse to measure the colour index of star B, based on photoelectric photom-etry from Huffer (1931), and transformed this into an approximatespectral type of F4.Kron (1939a) presented and tabulated a light curve of the sys-tem with full coverage of the eclipses, obtained using a photo-electric photometer with an effective wavelength of 4500 ˚A (Kron1939c). He used this to measure the fractional radii and limb dark-ening (LD) coefficients of the two stars (see also Kron 1938). Kron(1942) used the same instrument but with a red filter and red-sensitive light detector to obtain a second light curve in a passbandcovering 5200–8200 ˚A (half maximum response). He found frac-tional radii in good agreement with those from his blue light curve,and determined a colour of star B corresponding to a spectral typeof F5. Serkowski (1961) reconsidered the determination of the LDcoefficients from the Kron light curves.McNamara (1951) presented two new light curves in a blueand a UV passband, obtained using similar equipment and methodsto Kron (1939a). The main aim was to investigate the LD at bluewavelengths. Grygar et al. (1972) presented a detailed discussionof LD from the two Kron and the two McNamara light curves, andrevised measurements of the physical properties of the stars.Koch et al. (1965) measured the spectral type of star A to beA2-A3 and its equatorial rotational velocity to be V eq = 34 ± km s − . Perry & Stone (1966) obtained a single-lined spectro-scopic orbit from 92 RVs. Their measured orbital eccentricity, e = 0 . ± . (most likely a probable rather than standarderror), suggests that the orbit is circular (Lucy & Sweeney 1971).Lacy (1981) studied YZ Cas using the Kron light curves andnew spectra from which RVs were measured for star B for thefirst time. Lacy adopted photometric parameters which were theweighted average of eight independent studies of the same lightcurves (Kron 1939a, 1942), a procedure which may lack statis-tical validity. Lacy also obtained individual Str¨omgren photomet-ric indices of the two component stars from published photometry.Lacy’s RVs yielded measurements of the masses of the stars to aprecision of 0.5%, allied with values of V eq of ± km s − forstar A and ± km s − for star B.Shortly after publication of the study of Lacy (1981),de Landtsheer (1983) presented a large number of photoelectric ob-servations of YZ Cas. These were obtained in four colours in theUtrecht photometric system (Provoost 1980; Heintze & Van Gent1989), and were analysed using the code of Wilson & Devinney(1971) in order to obtain the physical properties of the stars.de Landtsheer & Mulder (1983) obtained two high-resolution UVspectra from the International Ultraviolet Explorer satellite andconfirmed the metallic nature of star A. The abundance of star Bwas not measured, so the internal (rather than photospheric) abun-dance of star A remained unknown. de Landtsheer & De Greve(1984) found a subsolar metal abundance ( Z = 0 . where Z isthe mass fraction of metals) in a comparison with theoretical stel-lar evolutionary models, in disagreement with Lacy’s finding of asupersolar bulk metal abundance of Z = 0 . ± . .Finally, Papouˇsek (1989) obtained 16 000 photoelectric obser-vations in six passbands, from the 60 cm reflector of the UniversityObservatory, Brno. He used these to study the physical propertiesand LD of the system. c (cid:13) , 000–000 he eclipsing binary YZ Cassiopeiae Table 2.
The wavelengths of maximum transmission ( λ cen ) and the fullwidths at half maximum response (FWHM) of the passbands of the lightcurves used in this work. When possible the FWHM is given as the actualwavelength interval rather than just the width of the passband.Passband λ cen ( ˚A) FWHM ( ˚A) ReferenceKron blue 4500 (wide) Kron (1939c)Kron red 6000 5250–8220 Kron (1942)McNamara UV 3575 850 McNamara (1951)McNamara blue 4525 3500–5700 McNamara (1951)Utrecht 472 4730 105 Provoost (1980)Utrecht 672 6719 100 Provoost (1980)Utrecht 782 7812 115 Provoost (1980)Utrecht 871 8798 130 Provoost (1980)U1 3500 260 Papouˇsek (1989)U2 3810 155 Papouˇsek (1989)B 4075 190 Papouˇsek (1989)G 4903 100 Papouˇsek (1989)V 5405 145 Papouˇsek (1989)O 5822 100 Papouˇsek (1989) Hipparcos
We obtained 42 high-resolution spectra of YZ Cas in October 2007,using the Nordic Optical Telescope (NOT) at La Palma, Spain,equipped with the FIbre-fed Echelle Spectrograph (FIES). Thisspectrograph is housed in a separate climate-controlled buildingand has a high thermal and mechanical stability. The wavelengthscale was established from thorium-argon exposures taken regu-larly throughout the observing nights. We used fibre 4 in fibrebundle B, giving complete spectral coverage in the interval 3640–7360 ˚A at a reciprocal dispersion ranging from 0.023 ˚A px − in theblue to 0.045 ˚A px − in the red. The resolution of the instrumentis roughly 3.5 px, giving a resolving power of 48 000. An exposuretime of 300 s was used for all spectra, resulting in continuum signalto noise ratios in the region of 300 in the B and V bands.The basic steps for data reduction (bias subtraction, flat-fielding, correction for scattered light, extraction of orders, andwavelength calibration) were performed with IRAF . Removal ofthe instrumental blaze function for YZ Cas is not trivial, becausethe broad Balmer lines from star A extend over entire ´echelle or-ders. In such cases we interpolated blaze functions from adjacentorders which are well-defined, using a semi-manual approach and JAVA routines written by VK.
We present no new photometry in this work, due to the large num-ber and good quality of existing datasets. Published observationswere taken from a range of sources, discussed in Sect. 1.1, result-ing in a total of 15 separate light curves. All of these were observedin non-standard passbands, so we give a summary of them in Ta-ble 2. We did not use the data from Huffer (1928) as they are sparseand relatively scattered. IRAF is distributed by the National Optical Astronomy Observatory,which is operated by the Association of Universities for Research in As-tronomy (AURA) under cooperative agreement with the National ScienceFoundation.
Kron (1939a) and Kron (1942) presented photoelectric pho-tometry of YZ Cas in one blue and one red wide passband, obtainedusing a 1 m reflector at the Lick Observatory. The two Kron lightcurves were obtained from the papers by optical character recogni-tion using the
TESSERACT software. Due to changes in the opticalpath of the instrument between the variable star and its comparisonstar (23 Cas), many of the nights of data were shifted in magnitudeto obtain a good internal agreement. Becase of this, the observa-tions taken well outside eclipse carry essentially no information.We have therefore rejected data taken more than 0.05 phase unitsfrom the midpoint of an eclipse. A few observations were enclosedin square brackets to indicate that they are less reliable, and we alsorejected these. Each Kron datapoint is the sum of six individual de-flections on the chart recorder, meaning they have a relatively highprecision.McNamara (1951) observed YZ Cas in one UV and one bluepassband, with central wavelengths of 3575 ˚A and 4525 ˚A for anA0 star. The UV passband was selected using a Corning O-5840glass filter, and the blue passband with Corning C-5562 and O-3389 glass. The passband widths were not specified by McNamara(1951), so have been obtained from diagrams in Dobrowolski et al.(1977) and are given in Table 2. The blue passband has more than50% transmission from ± ˚A to ± ˚A, but a longred tail of roughly 30% transmission extends beyond the edge ofFig. 20 in Dobrowolski et al. (1977) at 7500 ˚A. The photometricdata were obtained from McNamara (1951) as a function of orbitalphase.de Landtsheer (1983) obtained light curves of YZ Cas in fourdifferent narrow passbands in the Utrecht photometric system, cen-tred on wavelengths 474, 672, 782 and 871 nm. These were aimedat wavelength intervals containing no telluric lines and the feweststellar spectral lines possible, with the motivation of improving theunderstanding of continuum LD through study of dEBs. Only oneset of Utrecht filters is known to exist, and these were affixed to a40 cm reflector then operated at Ausserbin in the Swiss Alps. Thefour Utrecht light curves had to be obtained from the paper by copy-ing out by hand. They have a much larger scatter than the Kron data,but the morphology of the light variation of YZ Cas means they arestill valuable data.YZ Cas was bright enough to be observed by the Hipparcos satellite (Perryman et al. 1997) and also the
Tycho experiment onboard
Hipparcos (Høg et al. 1997). All three datasets suffer from ashortage of points within the total phases of secondary minimum,but in the case of the
Hipparcos passband the scatter is sufficientlysmall that the points on the ascending branch of the secondary min-imum provide adequate constraints on the depth of the minimum.Papouˇsek (1989) presented possibly the most extensive obser-vations of YZ Cas. These comprise about 16 000 observations in anintermediate-band filter system denoted U , U , B , G , V and O ,in order of increasing wavelength. These are tabulated in his pa-per as sets of normal points covering phases between 0 and 1, andwere converted into machine-readable format using TESSERACT .As they are specified as magnitudes versus phase we do not haveinformation on the times of individual observations, only that thedata were obtained in the years 1973–1974. c (cid:13) , 000–000 Pavlovski et al.
Table 3.
Literature times of minimum light of YZ Cas and the observedminus calculated ( O − C ) values of the data compared to the ephemerisderived in this work.Time of minimum Cycle O − C value Reference(HJD − ± − − ± − − ± − − ± − − ± − − ± − ± − − ± − ± − − ± − ± − − ± − − ± − ± − − ± − − ± − ± − − ± − ± − ± − − ± − ± − − ± − ± ± − ± − ± − ± ± − ± ± − ± − ± − References: (1) Kukarkin (1928); (2) Plaskett (1926); (3) Huffer (1931);(4) Zverev (1936); (5) Hassenstein (1954); (6) Skoberla (1936); (7)Kron (1939a); (8) Dombrovskij (1964); (9) Lavrov & Lavrova (1988);(10) Papouˇsek (1989); (11) de Landtsheer (1983); (12) Diethelm & Lines(1986); (13) This work, based on the
Hipparcos and
Tycho B and V obser-vations; (14) Brelstaff (1994); (15) P. Svoboda (in Br´at et al. 2008). The available times of minimum light for YZ Cas extend from1923 to 2008. Most of these are listed by Kreiner et al. (2001) andPapouˇsek (1989), to which we added a few additional values froma literature search. We rejected several recent visual timings whichwere either discrepant or too uncertain to be useful. A straight linewas fitted to these timings as a function of HJD, from which wedetermined the orbital ephemeris:
Min I = HJD 2 445 583 . . × E The individual times of minimum light are given in Table 3 alongwith their references and residuals versus the fitted ephemeris. Thetime system used in this work is UTC, which does not compensatefor leap seconds; the time of primary eclipse is only specified towithin 9 s so this does not have a significant effect on the results. http://code.google.com/p/tesseract-ocr/ A small number of timings refer to secondary rather than pri-mary minimum. We assumed these to represent phase 0.5, and theirresiduals support this assumption. The minimum timings are there-fore consistent with a circular orbit.
The 15 available light curves of YZ Cas have been individuallymodelled using the
JKTEBOP code code (Southworth et al. 2004a;Southworth 2013), which is based on the EBOP program developedby P. B. Etzel (Popper & Etzel 1981; Etzel 1981; Nelson & Davis1972). This code represents the stellar figures as biaxial spheroids,with the spherical approximation used for the calculation of eclipsefunctions. This approximation is more than adequate for well-detached EBs such as YZ Cas. Tests for third light and orbital ec-centricity returned small and insignificant values, so these parame-ters were held at zero for the final solutions. The orbital period wasfixed at the value found in Sect. 3, but the ephemeris zeropoint wasincluded as a fitted parameter. The brightness outside eclipse wasfitted for, as were the sum and ratio of the fractional radii ( r A + r B and k ), the orbital inclination ( i ), and the central surface brightnessratio of the stars ( J ).YZ Cas has a rich history of use for measuring the LD ofintermediate-mass stars. For the Kron and Utrecht light curves wewere able to fit for the linear LD coefficient for star A ( u A ) butnot for star B ( u B ). We therefore fixed u B to theoretical values(Van Hamme 1993; Claret 2000, 2004; Claret & Hauschildt 2003)appropriate for the photometric passband used. There is mount-ing observational evidence that the more complex two-parameterLD laws can provide a significant improvement (Southworth et al.2007a,b, 2009), so we tried the quadratic and logarithmic laws. Theuse of these laws did not yield better fits, so we adopted the linearlaw for our final results. This is in line with the experience of otherswhen analysing dEBs (e.g. Lacy et al. 2010).Uncertainties in each solution were found using 10 000 MonteCarlo simulations (Southworth et al. 2004b, 2005c). u B was fixedat theoretical values when finding the best fits but perturbed by ± r A and i , with a correlation coefficient of − . , and arisesbecause these parameters govern the eclipse duration. The other isbetween k and u A , with a correlation coefficient of − . , and isexpected as the two parameters affect the depth of primary mini-mum.For the Kron data we noticed a systematic mismatch betweenthe data and the best fit immediately before and after the minima.We therefore set the coefficients of the reflection effect to be fittedparameters, in effect allowing for different normalisations for pri-mary and secondary eclipse. This approach yielded a significantlybetter fit, but had a negligible effect on the values of the derivedparameters. JKTEBOP is written in
FORTRAN
77 and the source code is available at c (cid:13) , 000–000 he eclipsing binary YZ Cassiopeiae The fitted light curves are shown in Figs. 1 and 2, and theparameters of the fitted models in Tables 4, 5 and 6. The parame-ters of the fits unfortunately are in poor agreement with each other,suggesting that the Monte Carlo and residual-permutation uncer-tainties are underestimated. It is also the case that the two lightcurves which yield the most precise parameters (the Kron datasets)suffer from systematic errors due to optical path changes in the in-strument, so their uncertainties have an additional reason to be un-derestimated. The light curves with the greatest discrepancy withrespect to the others is U from Papouˇsek (1989), which has thebluest passband of all datasets studied in the current work. Thislight curve was observed from a low-altitude site in mainland Eu-rope. Our experience of such observatories is that the atmosphereis almost always unstable on hourly timescales, leading to extinc-tion variations and thus systematic errors. These are particularlypronounced in the blue, where atmospheric extinction is greater.Fig. 3 shows the photometric parameters r A , r B , i , u A andlight ratio as a function of wavelength. YZ Cas has sometimes in thepast been taken to indicate possible variations of radius with wave-length. Aside from the discrepant results for the bluemost passband(Papouˇsek U ), this figure shows no evidence for a variation withwavelength of r A , r B or i . The quantity u A has a weak dependenceon wavelength and in general higher values than expected theoreti-cally.For the final photometric parameters we have calculated theweighted mean and its reduced χ ( χ ν ) of the geometrical quanti-ties (i.e. those which do not depend on wavelength). As expectedgiven the discussion above, we find χ ν > for all quantities. i isleast problematic at χ ν = 1 . and k is the most difficult at χ ν =5 . . This is a very similar situation to that often found in studiesof transiting planetary systems (Southworth 2008, 2010, 2012). Wehave therefore multiplied the errorbars of the final weighted-meanvalues by p χ ν to yield reliable errorbars. The large number ofavailable light curves and careful treatment of uncertainties meansthat our results are robust despite the excess χ ν found for someparameters. Final values of the geometrical parameters are given inTable 6 and show that r A and r B have been measured to better than0.5% precision.Fig. 4 shows the output from Monte Carlo simulations for eachlight curve and for the parameters r A + r B , k , r B and i . The pointsplotted represent best-fit values for each of the synthetic datasetsgenerated for the Monte Carlo simulations. It can be seen that thegeometric parameters are actually more precisely defined for theoldest four datasets, and least well-defined for the Hipparcos sur-vey data. The main parameters of the fit, r A + r B and k , are veryweakly correlated (e.g. correlation coefficient . for the Utrecht672 dataset), whereas the derived parameters r A and r B are morecorrelated ( . ). A strong correlation occurs between i and r B ( − . ), as can be seen in the lower panels of Fig. 4. The light ratio between the stars is very well determined forYZ Cas, in the passbands for which we have light curves, due toit showing total and annular eclipses. For the spectroscopic analy-sis in Sect. 6, however, we needed a measurement of the continuumlight ratio as a function of wavelength.We therefore fitted two synthetic spectra from
ATLAS
Figure 3.
The photometric parameters r A , r B , i , u A and light ratio as afunction of wavelength. The Utrecht results are shown using blue lines,the Papouˇsek (1989) results using red lines, and other results using blacklines. The horizontal lines show the widths of the passbands, but the widepassbands are indicated only for the wavelength-dependent quantities in thelower two panels. The dashed horizontal lines show the adopted final val-ues and the dotted horizontal lines indicate the size of the errorbars on thesequantities. Papouˇsek (1989). The synthetic spectra were used only to fill inthe gaps between different passbands, so the choice of their atmo-spheric parameters was unimportant. Finally, we fitted a second-order polynomial to the light ratio as a function of wavelength(from 4000 to 7000 ˚A) to give a smooth and continuous proxy tothe true light ratio of the system. Star A produces between 86%and 94% of the system light in this wavelength region, being moredominant at bluer wavelengths due to its higher T eff (Fig. 3). c (cid:13) , 000–000 Pavlovski et al.
Figure 1.
Light curves of YZ Cas from Kron (1939a), Kron (1942), McNamara (1951), de Landtsheer (1983) and
Hipparcos (coloured points) compared tothe
JKTEBOP best fits (black curves). The residuals of the fits are offset from zero to appear at the base of the figure.
Table 4.
Model parameters of the Kron and Utrecht light curves of YZ Cas. The upper part of the table contains fitted parameters and the lower part containsderived parameters. N obs is the number of datapoints in each light curve and the scatter is the rms of the residuals around the best fit.Parameter Kron blue Kron red Utrecht 472 Utrecht 672 Utrecht 781 Utrecht 871 J ± ± ± ± ± ± r A + r B ± ± ± ± ± ± k ± ± ± ± ± ± u A ± ± ± ± ± ± u B i ( ◦ ) 88.18 ± ± ± ± ± ± r A ± ± ± ± ± ± r B ± ± ± ± ± ± ± ± ± ± ± ± N obs
407 565 827 829 835 833Scatter (mmag) 3.9 5.0 14.7 14.1 14.3 14.6c (cid:13) , 000–000 he eclipsing binary YZ Cassiopeiae Figure 2.
Light curves of YZ Cas from Papouˇsek (1989) (coloured points) compared to the
JKTEBOP best fits (black curves). The residuals of the fits are offsetfrom zero to appear at the base of the figure.
Table 5.
Model parameters of the Papouˇsek (1989) light curves of YZ Cas. See Table 4 for other details.Parameter U U B G V OJ ± ± ± ± ± ± r A + r B ± ± ± ± ± ± k ± ± ± ± ± ± u A ± ± ± ± ± ± u B i ( ◦ ) 89.96 ± ± ± ± ± ± r A ± ± ± ± ± ± r B ± ± ± ± ± ± ± ± ± ± ± ± N obs
175 175 183 164 158 164Scatter (mmag) 4.7 4.5 5.6 4.9 5.2 5.2
Table 7.
The orbital elements of YZ Cas from SPD. We fixed the orbitalperiod in our analysis to the value found by Lacy (1981).Parameter This work Lacy (1981)Orbital period P (d) 4.4672235 4.4672235Orbital eccentricity e ± K A ( km s − ) 73.05 ± ± K B ( km s − ) 124.78 ± ± q ± ± The small light contribution of star B makes the measurement ofits orbital motion and atmospheric parameters relatively challeng-ing, even from ´echelle spectra. We have therefore used the spec-tral disentangling (SPD) approach for our spectroscopic analysis.SPD of time-series spectra of binary star systems (Simon & Sturm1994) is a powerful technique which enables the determinationof the optimal set of the orbital parameters of a binary system,and the individual spectra of the components, simultaneously andself-consistently. A discussion of SPD and its practical applica-tions can be found in reviews by Pavlovski & Hensberge (2010)and Pavlovski & Southworth (2012). c (cid:13) , 000–000 Pavlovski et al.
Table 6.
Model parameters of the McNamara (1951) and
Hipparcos light curves of YZ Cas. The final parameter values, obtained from all fifteen light curves,are given in the last column. See Table 4 for other details.Parameter McNamara UV McNamara blue
Hipparcos
Final values J ± ± ± r A + r B ± ± ± ± k ± ± ± ± u A ± ± ± u B i ( ◦ ) 88.81 ± ± ± ± r A ± ± ± ± r B ± ± ± ± ± ± ± N obs
894 893 124Scatter (mmag) 6.6 7.4 7.0
Figure 4.
Plots of the variations of best-fit parameters for synthetic datasets generated for the Monte Carlo analysis. From left to right the panels show thelight curves in the order they are given in Table 2. For legibility the x-axis labels for the parameters r A + r B and r B are only specified for alternate panels,and the tick values are multiplied by 100 (in effect omitting the leading “0.” in each case). The light curves are colour-coded according to Figs. 1 and 2. As YZ Cas is a dEB, we have followed the analysis approachdescribed in detail by Hensberge et al. (2000). SPD was performedwith the FDB
INARY code (Iliji´c et al. 2004) which implementsthe Fourier approach of Hadrava (1995). Since no spectra wereobtained during totality in eclipse, there is an ambiguity in thedetermination of the continuum level. Therefore, SPD was per-formed in pure separation mode and renormalisation of disentan-gled spectra of the individual components to their continua wasdone with the light ratio derived from the light curves in Section4.1 (Pavlovski & Southworth 2012). Since FDB INARY is based ondiscrete Fourier transforms there is no limitation on the length orresolution of the spectra to be analysed, so far as the basic prescrip-tions for Fourier disentangling are fulfilled.There is no indication from photometry (Sections 3 and 4) orprevious spectroscopy (Koch et al. 1965; Lacy 1981) that the orbitis eccentric. We therefore fixed e to zero in our analysis, and fit-ted only for the velocity amplitudes of the two stars ( K A and K B ).Spectral regions containing Balmer lines were avoided in the opti- http://sail.zpf.fer.hr/fdbinary misation of K A and K B , as SPD is very sensitive to minor imper-fections in continuum placement over these extremely broad lines.The large wavelength interval covered by our ´echelle spectra meansthat there is plenty of spectrum with a well-defined continuum andcontaining only metallic lines. SPD was performed and K A and K B calculated for a number of spectral intervals of widths rangingfrom 50 ˚A to 100 ˚A. The quality of the normalisation and mergingof the spectra is reflected in very small standard deviations of thevalues of K A ( ± − ) and K B ( ± − ).A second estimate of the uncertainties in K A and K B can beobtained using other techniques for least-squares analysis such asbootstrap resampling, but such analysis can require a lot of com-putation time. We have therefore performed a jackknife test wherea set of solutions are calculated, each one ignoring a single ob-served spectrum. We have previously found reasonable and realis-tic results using this approach for the dEB AS Cam (Pavlovski et al.2011). This was done for seven of the spectral regions, with theresult that the jackknife-derived uncertainties ( ∼ − and ∼ − ) are roughly one order of magnitude greater thanthe standard deviation in K A and K B for these spectral regions( ± − and ± − ). We accept the jackknife uncer- c (cid:13) , 000–000 he eclipsing binary YZ Cassiopeiae HbetaHgammaHdelta
Figure 5.
Determination of the T eff of YZ Cas A from its Balmer line pro-file. The renormalised disentangled profiles are shown by thin black lines,and theoretical lines profiles for T eff s of 9800, 9600 and 9400 K are shownby red, blue and green lines. From top to bottom the profiles are of H β , H γ and H δ . The surface gravity is fixed from the light and RV curve analyses. tainties as our final errorbars (Table 7); they correspond to measure-ment uncertainties of 0.5% in the masses of the two stars. This isa realistic result for 42 high-quality ´echelle spectra of a dEB con-taining slowly rotating stars in a circular orbit. The complexity of spectra of close binary systems due to their com-posite nature and continuously varying Doppler shifts makes theextraction of atmospheric parameters non-trivial. The SPD tech-nique is an important aid to such studies, as it allows the derivationof the individual spectra of the two stars without needing templatespectra for guidance. These separated spectra can then be analysedas if they were observed spectra of single stars, allowing the deter-mination of the T eff s and chemical composition of the two compo-nents of the binary system. These numbers are in turn important forthe use of dEBs as distance indicators (e.g. Hensberge et al. 2000;North et al. 2010) and as critical tests of stellar evolutionary theory(e.g. Pavlovski et al. 2009; Brogaard et al. 2011).The first uses of SPD to estimate T eff s from disentangled indi-vidual component spectra were made for the close binaries DH Cepby Sturm & Simon (1994) and Y Cyg by Simon et al. (1994). Theseauthors also estimated helium abundances for the components inthe systems. On these grounds Hensberge et al. (2000) constructeda self-consistent complementary approach in the analysis of closebinary stars. In a detailed study of the high-mass double-lined dEBV578 Mon, Hensberge et al. were able to determine the basic phys-ical properties more accurately than possible using ‘standard’ tech-niques. In a follow-up study Pavlovski & Hensberge (2005) pre-sented the first abundance analysis using the broad wavelengthrange available in disentangled ´echelle spectra.In this work we follow the approach of Hensberge et al.(2000). The disentangled spectra have to be normalised to their in-trinsic continua as this cannot be done by SPD. SPD essentially attibutes spectral features to the individual stars according to theirorbital motion, and this is not possible for continuum flux unlessspectra were obtained during eclipse. The light factor (LFs) whichgive the continuum level of each star can be obtained either fromthe light curves (using the light ratios from the best-fitting eclipsemodel) or by constrained fitting of the Balmer lines of both com-ponents simultaneously (Tamajo et al. 2011). We proceeded to estimate the T eff of star A by fitting its Balmer lineprofiles and then fine-tuning the value using the ionisation balanceof the many Fe lines in the spectrum. The Balmer lines in A-typestars are sensitive both to T eff and surface gravity, log g . In thecase of YZ Cas, this degeneracy can be easily sidestepped sinceour analysis of the light and RV curves results in a measurement of log g for both stars to an accuracy of better than 0.01 dex; this is anintrinsic advantage for spectroscopic analysis of EBs.Spectral disentangling was performed on three spectral inter-vals of 150–250 ˚A width, centred on H β , H γ and H δ . In the SPD ofearly-type stars these are the most difficult and uncertain spectralregions since Doppler shifts due to orbital motion are much smallerthan the intrinsic widths of the Balmer lines. In order to minimisesystematic errors in the normalisation and merging of ´echelle or-ders containing broad Balmer lines we developed a semi-automaticprocedure. A high-order polynomial fit of the blaze function wascalculated in adjacent well-defined (‘cleaner’) ´echelle orders. Thenthe blaze function was constructed for ´echelle orders containingbroad Balmer lines by interpolation and scaling between adjacentorders. We used the light ratio measured from the photometric anal-ysis (Sect. 4.1) to normalise the disentangled spectra to the intrinsiccontinuum of each star. Since the disentangled (separated) spectraof the components have to be corrected for the dilution effect, inthis particular case by factors of about 1.08 for star A, and 14.3 forstar B, any imperfections in the continuum placement or the deter-mination of the light ratio would be scaled up by the same amount.Because of the large light ratio between the components, as well astheir atmopsheric properties, in this work the effective temperaturewas determined from Balmer lines only for star A. It was then usedfor deriving the T eff of star B through the photometric analysis, andas a starting point in determination of the T eff of star A throughdetailed spectroscopic analysis.A grid of local thermodynamic equilibrium (LTE) syntheticspectra was calculated using the UCLSYN code (Smith 1992;Smalley et al. 2001) and
ATLAS (cid:2) MH (cid:3) = 0 . The grid covers T eff s from 8400 K to10 400 K in steps of 200 K, for the known surface gravity of star A( log g = 3 . ). The T eff of star A was determined by minimis-ing the difference between the reconstructed disentangled spec-tra and the synthetic spectra in the spectral ranges of H δ (4070–4170 ˚A), H γ (4290–4370 ˚A) and H β (4800–4900 ˚A). Minimisationwas performed separately on the blue and red wings. Metallic linescontaminating the Balmer profiles were masked out during thisprocess. Minimisation from these six wings of Balmer lines gave T eff = 9670 ± K for star A, where the errorbar is the 1 σ error.In Fig. 5 we show the quality of fit to the observed (disentangled)spectra of the Balmer lines by the synthetic spectra.In the spectra of the components several species appear in twoor even three ionization stages, e.g. Si for star A. The most numer-ous lines in the spectra of both components are Fe I and Fe II lines,and they are well suited to the determination of both T eff and mi-croturbulent velocity ( v turb ). T eff was determined from Fe I and c (cid:13) , 000–000 Pavlovski et al. Fe II lines under the condition that abundances should not dependon ionisation state or excitation potential (Gray 2008). We deter-mined v turb by requiring the Fe abundance to be independent ofequivalent width (EW) (Magain 1984).The measurements of EWs were obtained by direct integra-tion of the line profiles and abundances for Fe I and Fe II calculatedusing UCLSYN . We have followed the critical selection of Fe I andFe II lines from Qiu et al. (2001) in deriving T eff and v turb , but allidentified Fe lines were measured. The lines were selected on thebasis of the reliability of their atomic data, mostly gf values, andtheir strength. We agree with Qiu et al. (2001) that including alllines, most of which are weak lines with EW < m ˚A, results in alarger scatter and less reliable abundances. In total we have 51 Fe I and 30 Fe II reliable lines for star A (see Table 8), with EWs rang-ing from 5 to 100 m ˚A. For the cooler star B there are many more Felines, 216 Fe I and 24 Fe II , and these have EWs of 5–70 m ˚A. AllFe lines detected were checked against new entries for gf values inthe VALD line lists (Kupka et al. 2000, and references therein).For star A we found T eff = 9520 ± K, in acceptable agree-ment with the value from fitting the Balmer lines, and v turb =3 . ± . km s − . In the case of star B, we estimated a prelimi-nary T eff from its colour indices because its Balmer lines are muchweaker. We refined the value using the ionisation balance of Fe,finding T eff = 6880 ± K and v turb = 1 . ± . km s − . Theobserved spectra are quite sensitive to v turb so we have been ableto constrain the values well. The derived value for star A is typicalfor Am stars (e.g. Landstreet et al. 2009) whilst the value for star Bfits the trend found for stars of similar T eff (e.g. Bruntt et al. 2010;Doyle et al. 2013).de Landtsheer & Mulder (1983) estimated the T eff of star Afrom an IUE low-resolution calibrated spectrum, in a determina-tion also involving the distance and line-of-sight interstellar ex-tinction for YZ Cas. The authors concluded that T eff = 10 600 ± K, arguing against the value of
K resulting from visualspectral classification (Hill et al. 1975). Ribas et al. (2000) usedintermediate-band photometry and grids of spectra from model at-mospheres to revise the T eff scale using a sample of dEBs. Theyfound T eff = 9100 ± K for star A and T eff = 6600 ± K forstar B. Our spectroscopic results, based both on Balmer line profilesand Fe ionisation balance, corroborate their estimates to within 1 σ . For stars with T eff s similar to those of the components of YZ Cas, alarge number of spectral lines arise from neutral and singly ionisedspecies. The number of good lines for abundance determination is,however, decreased due to the rotation of the component stars. Weselected good lines based on the quality of their atomic data and theamount of blending with other lines.We used UCLSYN to measure EWs and calculate the abun-dances. Table 8 contains the derived abundances for all elementsidentified in the spectrum of either star, in two forms. Firstly, thenumber density ǫ is given on a scale where log ǫ (H) = 12 . . Sec-ondly, the value for element X is converted into logarithmic abun-dance versus the Sun ([X/H]) using the standard solar abundancesfrom Asplund et al. (2009), the most recent critical evaluation ofthe solar abundances. The errorbars on these quantities are ther.m.s. errors of the results for individual lines, so are indicative of http://ams.astro.univie.ac.at/vald Table 8.
Measured chemical abundances in the photospheres of the compo-nent stars of YZ Cas, derived in the LTE approximation. N is the numberof lines used for each element.Ion YZ Cas A YZ Cas B N log ǫ (X) [X/H] N log ǫ (X) [X/H]C 55 8.85 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± − ± ± ± ± ± ± − ± ± ± ± ± ± ± ± ± ± − ± ± − ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± − ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Figure 6.
Comparison between abundances derived for star A (red filledcircles) and star B (blue filled squares). the number and quality of the lines used rather than the true uncer-tainties. Additional contributions to the uncertainties arise from thequality of the atomic data, the quality of the spectra and their nor-malisation, and uncertainties in the measured T eff , log g and v turb values. The spectra are of high quality and their normalisation iswell-defined away from the Balmer lines, so these should not giverise to significant systematic errors. Similarly, the atmospheric pa- c (cid:13) , 000–000 he eclipsing binary YZ Cassiopeiae rameters are measured to high precision so will not cause muchuncertainty.The abundances in Table 8 are visualised in Fig. 6. We findthat the chemical abundances for star B are close to solar; a cal-culation of the metallicity from the abundances listed in Table 8for star B and replacing missing elements with abundances givenby Asplund et al. (2009) returns a value of Z = 0 . ± . (where Z is the mass fraction of metals) which is slightly morethan 1 σ away from the solar value of Z = 0 . derived inAsplund et al. (2009). These values should be appropriate for thebulk chemical composition of the system as a whole because thephotosphere is expected to represent the internal composition for a6880 K dwarf star. The abundances for star A are all significantlysuper-solar except for Sc, confirming the Am nature of this star.Zn, Y, Zr and Ba are all overabundant by 1 dex relative to solar. Wehave not been able to measure the abundances of any rare-earth el-ements, even though a strong over-abundance of these is a hallmarkof the Am phenomenon (Wolff 1983).We calculated a provisional set of abundances for star A us-ing model atmospheres of solar abundance rather than abundancestuned to the true metallicity of the star. For our final abundancemeasurements (Table 8), we used model atmospheres with scaled-solar metallicity and a metal abundance of [X/H] = T eff . We found thatthe use of model atmospheres with [X/H] = T eff (10 600 K) andmeasured a higher v turb ( . ± . km s − ) than we find, and thisis probably why they derived exceptionally high abundance values(from 0.80 dex for Fe to 3.4 dex for Co relative to solar). We do notconfirm these overabundances for five of the six chemicals (Si, Cr,Mn, Fe, Ni). We did not obtain an abundance for the sixth, Co, butthe value from de Landtsheer & Mulder (1983) is almost certainlyalso wrong. The disentangled spectra of YZ Cas have S/N values of about 1500and 200 for star A and star B, respectively, and are rich in informa-tion. We first measured the instrumental broadening profile usingboth ThAr emission lines and and telluric absorption lines, findingFWHM = v sin i by line-profile fitting of complexblends using the UCLSYN code.In each disentangled spectrum several spectral regions wereselected containing lines with good atomic data. Line-profile fittingthen yielded v A sin i = 29 . km s − and v B sin i = 15 . km s − .The formal errors on these determinations are ± − and ± − , respectively. Additional contributions to these uncer-tainties come from the instrumental broadening, its variation withwavelength, continuum normalisation, microturbulence and macro- Table 9.
The physical properties of the YZ Cas system.Parameter Star A Star BOrbital separation ( R ⊙ ) . ± . Mass ( M ⊙ ) 2.263 ± ± R ⊙ ) 2.525 ± ± log g [cm s − ] 3.988 ± ± V synch ( km s − ) 28.61 ± ± v sin i ( km s − ) 29.2 ± ± T eff (K) 9520 ±
120 6880 ± log( L/ L ⊙ ) ± ± M bol 1 ± ± . ± . Calculated assuming L ⊙ = 3 . × W (Bahcall et al. 1995) and M bol ⊙ = 4 . (Zombeck 1990). turbulence. We therefore adopt larger errorbars of ± − forboth v sin i measurements.If both components show synchronous rotation the ratio oftheir velocities should equal the ratio of their radii. We find v A sin i/v B sin i = 0 . ± . and k = 0 . ± . , sothese values are consistent to within 0.5 σ . This matches our expec-tation that the components should be rotating synchronously withthe orbital motion. Using the photometric and spectroscopic results from Sections 4and 5, we have calculated the physical properties of the YZ Cassystem (Table 7). This was done using the
JKTABSDIM code(Southworth et al. 2005a), which propagates uncertainties via aperturbation analysis. We adopted the physical constants tabulatedby Southworth (2011). The masses and radii are measured to pre-cisions of 0.5%, representing a useful improvement over previ-ous analyses. The high precision of these determinations is due toseveral circumstances. Firstly, we had access to extensive obser-vational data: 15 published light curves plus 42 new high-quality´echelle spectra. Secondly, the intrinsic character of the systemmakes it well-suited to detailed analysis. The total eclipses meanthe light curve solutions are very well-defined, and the slow rota-tion and large number of spectral lines means the orbital velocitiesof the two components are measurable to high precision. Comparedto Lacy (1981), we find very similar radii but masses smaller by 2–3 σ . We have also measured the atmospheric parameters of bothstars to high precision. Our new T eff measurements are precise andsupport previous optical rather than UV determinations. The ro-tational velocities of both stars are consistent with synchronocity,as expected from the timescales for rotational synchronisation (seeZahn 1975, 1977).From the known masses and radii and the T eff s obtained inSect. 6 we have calculated the luminosities (expressed logarithmi-cally with respect to the Sun in Table 7) and absolute bolometricmagnitudes of the two stars. From these we have measured the dis-tance to the system using empirical surface brightness relations (seeSouthworth et al. 2005a and Kervella et al. 2004) and by the usualmethod involving bolometric corrections (e.g. Southworth et al.2005b). For this process we adopted apparent magnitudes in the B and V bands from Høg et al. (1997) and in the JHK s bands c (cid:13)000
JKTABSDIM code(Southworth et al. 2005a), which propagates uncertainties via aperturbation analysis. We adopted the physical constants tabulatedby Southworth (2011). The masses and radii are measured to pre-cisions of 0.5%, representing a useful improvement over previ-ous analyses. The high precision of these determinations is due toseveral circumstances. Firstly, we had access to extensive obser-vational data: 15 published light curves plus 42 new high-quality´echelle spectra. Secondly, the intrinsic character of the systemmakes it well-suited to detailed analysis. The total eclipses meanthe light curve solutions are very well-defined, and the slow rota-tion and large number of spectral lines means the orbital velocitiesof the two components are measurable to high precision. Comparedto Lacy (1981), we find very similar radii but masses smaller by 2–3 σ . We have also measured the atmospheric parameters of bothstars to high precision. Our new T eff measurements are precise andsupport previous optical rather than UV determinations. The ro-tational velocities of both stars are consistent with synchronocity,as expected from the timescales for rotational synchronisation (seeZahn 1975, 1977).From the known masses and radii and the T eff s obtained inSect. 6 we have calculated the luminosities (expressed logarithmi-cally with respect to the Sun in Table 7) and absolute bolometricmagnitudes of the two stars. From these we have measured the dis-tance to the system using empirical surface brightness relations (seeSouthworth et al. 2005a and Kervella et al. 2004) and by the usualmethod involving bolometric corrections (e.g. Southworth et al.2005b). For this process we adopted apparent magnitudes in the B and V bands from Høg et al. (1997) and in the JHK s bands c (cid:13)000 , 000–000 Pavlovski et al.
Figure 7.
The locations of the components of YZ Cas in the log T eff versus log g plane. The Geneva evolutionary tracks (Mowlavi et al. 2012) for theirmasses are plotted for the metallicity values Z = 0 . (green lines), 0.010(red lines) and 0.014 (blue lines). from Skrutskie et al. (2006). Tabulated bolometric corrections weretaken from Bessell et al. (1998) and Girardi et al. (2002). All dis-tance measurements agree well, and we take the K s -band surface-brightness-based distance of . ± . pc as our final value.The good agreement between the optical ( BV ) and IR( JHK s ) distance measurements argues against the presence of sig-nificant interstellar absorption; by requiring consonant distances wefind an upper limit of only E ( B − V ) = 0 . mag (3 σ ). We havealso interpreted the light ratio derived from the Papouˇsek (1989) V light curve as a light ratio in the Johnson V band to obtain aseparate distance estimate for the two stars. These agree to withintheir errorbars, so the T eff s and radii in Table 7 pass this consistencycheck.Both stars in YZ Cas are situated outside the δ Sct instabilitystrip (Dupret et al. 2005; Uytterhoeven et al. 2011): star A is hotterthan the blue edge of the fundamental radial mode, and star B iscooler than its red edge. With a T eff of 9520 K, star A is among thehottest of the Am stars. Considerable attention has been devoted to the metallicity of thecomponent stars of YZ Cas. The relatively large difference in theirmasses means they could be a stringent test of the stellar evolu-tionary models (Lastennet & Valls-Gabaud 2002). These authorsfound a problem in the discrepancy in metallicity estimated frommatching the physical properties of the system to the predictionsof theoretical stellar models ( Z = 0 . ), and the metallicityderived for star A from UV spectroscopy ( Z = 0 . ± . ;de Landtsheer & Mulder 1983). Lastennet et al. (2001) challengedthe UV metallicity determination on the basis that photometric methods do not indicate a high Z and are in fact compatible withsolar metallicity.Our detailed abundance analysis based on optical ´echelle spec-tra has revealed the photospheric chemical composition of bothstars. Star A shows a high metallicity which cannot be taken asan indication of its internal composition. Star B, however, showsa solar metallicity which should be representative of its bulk metalabundance.We have compared the masses, radii and T eff s of the compo-nents of YZ Cas to tabulated predictions from several sets of the-oretical stellar evolutionary models. An immediate results is thatthe best fit to the observed properties is found for a subsolar metalabundance, whereas a solar metallicity results in predicted T eff swhich are too small to match the observed values. We also wereable to infer a precise age, τ , for the system. The Teramo models(Pietrinferni et al. 2004) give a good fit for Z = 0 . and τ =490 Myr, with a slight preference for models with core overshoot-ing over canonical models. The VRSS models (VandenBerg et al.2006) agree very well with the measured properties for Z = 0 . and τ = 545 Myr, with an acceptable fit also being found for Z =0 . and τ = 530 Myr. The PARSEC models (Bressan et al.2012) for Z = 0 . and τ = 545 Myr are almost identical tothe VRSS ones, so also fit well. Finally, an investigation of the Y models (Demarque et al. 2004) resulted in a good fit for Z = 0 . and τ = 550 Myr, where a higher Z causes the predicted T eff s tobecome too low and a lower Z underpredicts the radius of star B.In Fig. 7 we show an alternative approach where evolutionarytracks interpolated to the specific masses of the two stars are plot-ted in the log T eff versus log g diagram for three metallicities. Themodels used are the most recent versions from the Geneva group(Mowlavi et al. 2012), which have modest convective core over-shooting and do not account for stellar rotation. We found the best-fitting metallicities and ages to be Z = 0 . ± . and 420 Myrfor star Am and Z = 0 . ± . and 670 Myr for star B. Thiscorroborates our inferences from the four sets of models discussedabove.We therefore find that the physical properties of both starsmatch the predictions of theoretical models for a metallicity of Z = 0 . ± . and an age of 400–700 Myr. This is trou-bling because the photospheric abundances we find for star B showit to have an approximately solar chemical composition. We co-clude that either the theoretical models are incorrect or the pho-tospheric abundances of star B do not represent its bulk chemicalcomposition. We have presented a detailed analysis of the dEB YZ Cas based on15 published light curves and 42 new high-quality ´echelle spectra.The principal attraction of this object is that it contains two quitedifferent stars: one an A-type metallic-lined star and the other asmaller and less massive F2 dwarf. We were therefore able to inves-tigate the Am nature of star A using the chemically normal star Bas a reference.We have measured the masses and radii of both componentto a precision of 0.5%, due to the large amount of observationalmaterial as well as the co-operative nature of the system. Thetime-resolved spectra were analysed using spectral disentangling,allowing us to derive the individual spectra of the two stars aswell as their velocity amplitudes. From the separated spectra we c (cid:13) , 000–000 he eclipsing binary YZ Cassiopeiae obtained the atmospheric parameters ( T eff and v turb ) and photo-spheric chemical compositions of both stars.Star A shows clear signs of the Am phenomenon: enhancedmetals, depleted Sc, and over-abundances of Zn, Y, Zr and Baby 1 dex. By contrast, star B shows normal chemical abundanceswhich are consistent with a solar or slightly super-solar chemicalcomposition. Whilst the Am phenomenon is a ‘surface disease’, theabundances of star B should represent the bulk chemical composi-tion of both stars. It is therefore surprising that theoretical stellarevolutionary models require a significantly sub-solar metallicity toreproduce the properties of the YZ Cas system. Our results cannotdifferentiate between the possibilities that the model predictions arewrong and that the photospheric abundances of star B do not repre-sent the true chemical composition of either star.The Am stars are defined phenomenologically: inclassification-resolution spectra the Ca K line is appropriatefor an early-A spectral type and metallic lines point to a late-Ato F type, whilst the hydrogen Balmer lines are intermediate(Roman et al. 1948). With the widespread use of higher-resolutionspectra a new definition of the Am phenomenon has evolved.Conti (1970) defined it as an apparent surface underabundanceof Ca and/or Sc, and/or an apparent overabundance of Fe-groupand heavier elements. Modern spectroscopic studies have revealedmuch variety in observed abundance patterns, and difficulties existin delineating the border between normal and chemically peculiarA stars. Interesting discussions of the Am phenomenon in can befound in Adelman & Unsuree (2007) and Murphy et al. (2012).While Am stars were long considered to have quiet atmospheresthat were therefore not pulsating, high-precision photometry fromboth ground-based (SuperWASP) and space-based ( Kepler ) sur-veys have yielded the detection of more than 200 A and F dwarfswith Am characteristics and detectable pulsations (Smalley et al.2011; Balona et al. 2011). Important constraints of the origin of theAm phenomenon are also provided by their rotation and binarity(Fossati et al. 2008; Stateva et al. 2012).Our work on YZ Cas has yielded precise measurements of themass, radius and T eff of an Am star, plus the abundances of 26 ele-ments in its photosphere. Its F2 V companion has a solar chemicalcomposition which should reflect the internal composition of bothstars. Theoretical models cannot reproduce the physical propertiesof the stars for this composition. Further work is needed to un-derstand the nature of this discrepancy and to model the processesoccurrent in the atmospheres of Am stars. ACKNOWLEDGMENTS
We are grateful to Frank Verbunt, Miloslav Zejda, Robert Van Gentand Jerzy Kreiner for discussions and information. We thank thereferee, Colin Folsom, for a helpful report. JS acknowledges fi-nancial support from STFC in the form of an Advanced Fellow-ship. Based on observations made with the Nordic Optical Tele-scope, operated on the island of La Palma jointly by Denmark,Finland, Iceland, Norway, and Sweden, in the Spanish Observa-torio del Roque de los Muchachos of the Instituto de Astrof´ısica deCanarias. The following internet-based resources were used in re-search for this paper: the ESO Digitized Sky Survey; the NASA As-trophysics Data System; the SIMBAD database operated at CDS,Strasbourg, France; and the ar χ iv scientific paper preprint serviceoperated by Cornell University. REFERENCES
Abt, H. A., 1961, ApJS, 6, 37Abt, H. A., 1965, ApJS, 11, 429Abt, H. A., Levy, S. G., 1985, ApJS, 59, 229Adelman, S. J., Unsuree, N., 2007, Baltic Astronomy, 16, 183Andersen, J., 1991, A&ARv, 3, 91Argelander, F., 1903, Bonner Durchmusterung des nordlichenHimmels, Marcus and Weber’s Verlag, BonnAsplund, M., Grevesse, N., Sauval, A. J., Scott, P., 2009,ARA&A, 47, 481Bahcall, J. N., Pinsonneault, M. H., Wasserburg, G. J., 1995, Re-views of Modern Physics, 67, 781Balona, L. A., et al., 2011, MNRAS, 414, 792Bessell, M. S., 2000, PASP, 112, 961Bessell, M. S., Castelli, F., Plez, B., 1998, A&A, 333, 231Bonanos, A. Z., et al., 2006, ApJ, 652, 313Br´at, L., et al., 2008, Open European Journal on Variable Stars,94Brelstaff, T., 1994, Br. Astron. Assoc., Var. Star Sect. Circ., 81, 4Bressan, A., Marigo, P., Girardi, L., Salasnich, B., Dal Cero, C.,Rubele, S., Nanni, A., 2012, MNRAS, 427, 127Brogaard, K., Bruntt, H., Grundahl, F., Clausen, J. V., Frandsen,S., Vandenberg, D. A., Bedin, L. R., 2011, A&A, 525, A2Bruntt, H., et al., 2010, MNRAS, 405, 1907Budaj, J., 1996, A&A, 313, 523Budaj, J., 1997, A&A, 326, 655Cannon, A. J., Pickering, E. C., 1918, Annals of Harvard CollegeObservatory, 91, 1Carquillat, J., Prieur, J., 2007, MNRAS, 380, 1064Claret, A., 2000, A&A, 363, 1081Claret, A., 2004, A&A, 428, 1001Claret, A., Hauschildt, P. H., 2003, A&A, 412, 241Clausen, J. V., Torres, G., Bruntt, H., Andersen, J., Nordstr¨om, B.,Stefanik, R. P., Latham, D. W., Southworth, J., 2008, A&A, 487,1095Conti, P. S., 1970, PASP, 82, 781Cutri, R. M., et al., 2003, 2MASS All Sky Catalogue of pointsources., The IRSA 2MASS All-Sky Point Source Catalogue,NASA/IPAC Infrared Science Archivede Landtsheer, A. C., 1983, A&AS, 53, 161de Landtsheer, A. C., De Greve, J. P., 1984, A&A, 135, 397de Landtsheer, A. C., Mulder, P. S., 1983, A&A, 127, 297Demarque, P., Woo, J.-H., Kim, Y.-C., Yi, S. K., 2004, ApJS, 155,667Diethelm, R., Lines, R. D., 1986, IBVS, 2924Dobrowolski, J. A. ., Marsh, G. E., Charbonneau, J. Eng, J., Jose-phy, P. D., 1977, Applied Optics, 16, 1491Dombrovskij, V. L., 1964, Trudy Astronomicheskoj ObservatoriiLeningrad, 21, 19Doyle, A. P., et al., 2013, MNRAS, 428, 3164Dupret, M.-A., Grigahc`ene, A., Garrido, R., Gabriel, M., Scu-flaire, R., 2005, A&A, 435, 927Etzel, P. B., 1981, in Carling, E. B., Kopal, Z., eds., Photomet-ric and Spectroscopic Binary Systems, NATO ASI Ser. C., 69,Kluwer, Dordrecht, p. 111Flamsteed, J., 1712, Historia Coelestis, London, J Mathews 1712Fossati, L., Kolenberg, K., Reegen, P., Weiss, W., 2008, A&A,485, 257Girardi, L., Bertelli, G., Bressan, A., Chiosi, C., Groenewegen,M. A. T., Marigo, P., Salasnich, B., Weiss, A., 2002, A&A, 391,195 c (cid:13) , 000–000 Pavlovski et al.
Gray, D. F., 2008, The observation and analysis of stellar photo-spheres, Cambridge University Press, 2008Gray, R. O., Corbally, J., C., 2009, Stellar Spectral Classification,Princeton University Press, Princeton, USGrygar, J., Cooper, M. L., Jurkevich, I., 1972, Bulletin of the As-tronomical Institutes of Czechoslovakia, 23, 147Hadrava, P., 1995, A&AS, 114, 393Hassenstein, W., 1954, Publikationen des AstrophysikalischenObservatoriums zu Potsdam, 29 No. 98Heintze, J. R. W., Van Gent, R. H., 1989, Space Science Reviews,50, 257Hensberge, H., Pavlovski, K., Verschueren, W., 2000, A&A, 358,553Hilditch, R. W., Hill, G., 1975, MmRAS, 79, 101Hill, G., Hilditch, R. W., Younger, F., Fisher, W. A., 1975, Mem-oirs of the RAS, 79, 131Hoffleit, D., Jaschek, C. ., 1991, The Bright Star Catalogue, NewHaven, Conn.: Yale University Observatory, 1991, 5th ed.Høg, E., et al., 1997, A&A, 323, L57Huffer, C. M., 1925, Popular Astronomy, 33, 588Huffer, C. M., 1928, Publications of the Washburn Observatory,15, 101Huffer, C. M., 1931, Publications of the Washburn Observatory,15, 103Iliji´c, S., Hensberge, H., Pavlovski, K., Freyhammer, L. M., 2004,in Hilditch, R. W., Hensberge, H., Pavlovski, K., eds., Spectro-scopically and Spatially Resolving the Components of the CloseBinary Stars, vol. 318 of
ASP Conf. Ser. , p. 111Kervella, P., Th´evenin, F., Di Folco, E., S´egransan, D., 2004,A&A, 426, 297Koch, R. H., Olson, E. C., Yoss, K. M., 1965, ApJ, 141, 955Kreiner, J. M., Kim, C.-H., Nha, I.-S., 2001, An atlas of O-C dia-grams of eclipsing binary starsKron, G. E., 1938, PASP, 50, 173Kron, G. E., 1939a, Lick Observatory Bulletin, 19, 59Kron, G. E., 1939b, PASP, 51, 217Kron, G. E., 1939c, Lick Observatory Bulletin, 19, 53Kron, G. E., 1942, ApJ, 96, 173Kukarkin, B. W., 1928, Peremennye Zvezdy, 1, 6Kupka, F. G., Ryabchikova, T. A., Piskunov, N. E., Stempels,H. C., Weiss, W. W., 2000, Baltic Astronomy, 9, 590Kurucz, R. L., 1979, ApJS, 40, 1Lacy, C. H. S., 1981, ApJ, 251, 591Lacy, C. H. S., Torres, G., Claret, A., Sabby, J. A., 2002, AJ, 123,1013Lacy, C. H. S., Claret, A., Sabby, J. A., 2004, AJ, 128, 1340Lacy, C. H. S., Torres, G., Claret, A., Charbonneau, D.,O’Donovan, F. T., Mandushev, G., 2010, AJ, 139, 2347Landstreet, J. D., Kupka, F., Ford, H. A., Officer, T., Sigut,T. A. A., Silaj, J., Strasser, S., Townshend, A., 2009, A&A, 503,973Lastennet, E., Valls-Gabaud, D., 2002, A&A, 396, 551Lastennet, E., Valls-Gabaud, D., Jordi, C., 2001, in Vanbeveren,D., ed., The Influence of Binaries on Stellar Population Studies,vol. 264 of
Astrophysics and Space Science Library , p. 575Lavrov, M. I., Lavrova, N. V., 1988, Trudy Kazanskaia GorodkojAstronomicheskoj Observatorii, 51, 3Lucy, L. B., Sweeney, M. A., 1971, AJ, 76, 544Magain, P., 1984, A&A, 134, 189McLaughlin, D. B., 1924, ApJ, 60, 22McNamara, D. H., 1951, A two-colour photoelectric study of theeclipsing variable YZ Cassiopeiae., Ph.D. thesis, University of California, BerkeleyMichaud, G., 1970, ApJ, 160, 641Mowlavi, N., Eggenberger, P., Meynet, G., Ekstr¨om, S., Georgy,C., Maeder, A., Charbonnel, C., Eyer, L., 2012, A&A, 541, A41Murphy, S. J., Grigahc`ene, A., Niemczura, E., Kurtz, D. W., Uyt-terhoeven, K., 2012, MNRAS, 427, 1418Nelson, B., Davis, W. D., 1972, ApJ, 174, 617North, P., Gauderon, R., Barblan, F., Royer, F., 2010, A&A, 520,A74Papouˇsek, J., 1989, Folia Facultatis scientiarum naturalium Uni-versitatis Purkynianae Brunensis. Physica., 49, 59Pavlovski, K., Hensberge, H., 2005, A&A, 439, 309Pavlovski, K., Hensberge, H., 2010, in Prˇsa, A., Zejda, M., eds.,Binaries – Key to Comprehension of the Universe, vol. 435 of
Astronomical Society of the Pacific Conference Series , p. 207Pavlovski, K., Southworth, J., 2012, in Richards, M. T., Hubeny,I., eds., IAU Symposium, vol. 282 of
IAU Symposium , p. 359Pavlovski, K., Tamajo, E., Koubsk´y, P., Southworth, J., Yang, S.,Kolbas, V., 2009, MNRAS, 400, 791Pavlovski, K., Southworth, J., Kolbas, V., 2011, ApJ, 734, L29Perry, C. L., Stone, S. N., 1966, PASP, 78, 5Perryman, M. A. C., et al., 1997, A&A, 323, L49Pietrinferni, A., Cassisi, S., Salaris, M., Castelli, F., 2004, ApJ,612, 168Plaskett, J. S., 1926, Publications of the Dominion AstrophysicalObservatory Victoria, 3, 247Popper, D. M., Etzel, P. B., 1981, AJ, 86, 102Provoost, P., 1980, A&AS, 40, 129Qiu, H. M., Zhao, G., Chen, Y. Q., Li, Z. W., 2001, ApJ, 548, 953Ribas, I., Jordi, C., Torra, J., Gim´enez, ´A., 2000, MNRAS, 313,99Ribas, I., Jordi, C., Vilardell, F., Fitzpatrick, E. L., Hilditch, R. W.,Guinan, E. F., 2005, ApJ, 635, L37Roman, N. G., Morgan, W. W., Eggen, O. J., 1948, ApJ, 107, 107Rossiter, R. A., 1924, ApJ, 60, 15Serkowski, K., 1961, AJ, 66, 405Simon, K. P., Sturm, E., 1994, A&A, 281, 286Simon, K. P., Sturm, E., Fiedler, A., 1994, A&A, 292, 507Skoberla, P., 1936, Zeitschrift f¨ur Astrophysik, 11, 1Skrutskie, M. F., et al., 2006, AJ, 131, 1163Smalley, B., 1993, MNRAS, 265, 1035Smalley, B., Smith, K. C., Dworetsky, M. M., 2001, UCLSYNUserguideSmalley, B., et al., 2011, A&A, 535, A3Smith, K. C., 1992, Ph.D. thesis, University of LondonSouthworth, J., 2008, MNRAS, 386, 1644Southworth, J., 2010, MNRAS, 408, 1689Southworth, J., 2011, MNRAS, 417, 2166Southworth, J., 2012, MNRAS, 426, 1291Southworth, J., 2013, A&A, 557, A119Southworth, J., Maxted, P. F. L., Smalley, B., 2004a, MNRAS,349, 547Southworth, J., Maxted, P. F. L., Smalley, B., 2004b, MNRAS,351, 1277Southworth, J., Maxted, P. F. L., Smalley, B., 2005a, A&A, 429,645Southworth, J., Maxted, P. F. L., Smalley, B., 2005b, in Kurtz,D. W., ed., IAU Colloq. 196: Transits of Venus: New Views ofthe Solar System and Galaxy, p. 361Southworth, J., Smalley, B., Maxted, P. F. L., Claret, A., Etzel,P. B., 2005c, MNRAS, 363, 529Southworth, J., Bruntt, H., Buzasi, D. L., 2007a, A&A, 467, 1215 c (cid:13) , 000–000 he eclipsing binary YZ Cassiopeiae Southworth, J., Wheatley, P. J., Sams, G., 2007b, MNRAS, 379,L11Southworth, J., et al., 2009, MNRAS, 399, 287Stateva, I., Iliev, I. K., Budaj, J., 2012, MNRAS, 420, 1207Stebbins, J., 1924, Popular Astronomy, 32, 233Sturm, E., Simon, K. P., 1994, A&A, 282, 93Talon, S., Richard, O., Michaud, G., 2006, ApJ, 645, 634Tamajo, E., Pavlovski, K., Southworth, J., 2011, A&A, 526, A76Titus, J., Morgan, W. W., 1940, ApJ, 92, 256Tomkin, J., Fekel, F. C., 2006, AJ, 131, 2652Torres, G., Andersen, J., Gim´enez, A., 2010, A&ARv, 18, 67Torres, G., et al., 1999, AJ, 118, 1831Turcotte, S., Richer, J., Michaud, G., Christensen-Dalsgaard, J.,2000, A&A, 360, 603Uytterhoeven, K., et al., 2011, A&A, 534, A125Van Hamme, W., 1993, AJ, 106, 2096VandenBerg, D. A., Bergbusch, P. A., Dowler, P. D., 2006, ApJS,162, 375Wilson, R. E., Devinney, E. J., 1971, ApJ, 166, 605Wolff, S. C., 1983, The A-stars: Problems and perspectives.Monograph series on nonthermal phenomena in stellar atmo-spheresZahn, J., 1975, A&A, 41, 329Zahn, J., 1977, A&A, 57, 383Zombeck, M. V., 1990, Handbook of space astronomy and astro-physics, Cambridge University Press, 1990, 2nd editionZverev, M., 1936, Trudy Gosudarstvennogo AstronomicheskogoInstituta Sternberg, 8, 37 c (cid:13)000