A&A 425, 171-177 (2004)
DOI: 10.1051/0004-6361:20035922

HD 153720 - A SB2 system with twin metallic-line components[*],[*]

A. V. Yushchenko 1,2 - V. F. Gopka 2 - V. L. Khokhlova 3 - D. L. Lambert 4 - C. Kim 5 - Y. W. Kang 1

1 - Department of Astronomy, Sejong University, Seoul, 143-747, Korea
2 - Odessa Astronomical Observatory, Odessa National University, Park Shevchenko, Odessa, 65014, Ukraine
3 - Deceased (Institute of Astronomy, Russian Academy of Sciences, Pyatnitskaya ul. 48, Moscow, 109017, Russia)
4 - The W. J. McDonald Observatory, University of Texas, Austin, TX 78712, USA
5 - Department of Earth Science Education, Chonbuk National University, Chonju 561-756, Korea

Received 20 December 2003/ Accepted 25 May 2004

We report the results of abundance determinations for the components of the SB2 star HD 153720 from high resolution (R=60 000) echelle high signal-to-noise spectra of the wavelength region 3595-10 260 Å taken with the 2.7 m telescope of the McDonald Observatory We found the values of the atmospheric parameters of the primary to be effective temperature $T_{{\rm eff}}$ = 7425 K and surface gravity $\log~g$ = 4.0 cgs, and of the secondary to be $T_{{\rm eff}}$ = 7125 K and $\log~g$ = 3.9 cgs. The microturbulent velocity is $v_{{\rm micro}}$ = 2.7  ${{\rm km~s^{-1}}}$ for both components, and the projected rotational velocity is ${v\sin i}$  = 15  ${{\rm km~s^{-1}}}$ also for both components. The abundances of about 20 elements were determined with the method of spectrum synthesis. The components of HD 153720 are metallic-line stars. Possible inconsistencies between old and new measurements of radial velocities may be explained by the existence of third body in this system. A review of recent high resolution spectral observations of eight A4-F1 binaries shows that only one of these systems can be classified as normal.

Key words: stars: abundances - stars: atmospheres - stars: evolution - stars: chemically peculiar - stars: binaries: spectroscopic - stars: individual: HD 153720 

1 Introduction

HD 153720 (BD+75 608) of magnitude $V=6\hbox{$.\!\!^{\rm m}$ }82$ and spectral class F0 was first noted by Young (1942) to be a spectroscopic binary. The orbit of this unevolved double-lined spectroscopic binary was determined by Heard & Hurkens (1975) using forty-six 12 Å/mm spectrograms (1971-1974). They did not mention the chemical composition of the components, but pointed out that there was no difficulty in distinguishing primary from secondary, and that the difference in their line strengths was not great. Budaj (1996) used HD 153720 as an example of a binary with normal components, i.e., neither star exhibited chemical peculiarities. We observed HD 153720 as part of an investigation of atmospheric abundances of SB2 systems whose components have equal masses and are unevolved main-sequence stars.

It was started by one of the authors of this paper (Vera L. Khokhlova). She died last September (2003); this and the next papers of the series are devoted to her memory.

The stars of this program studied to date are AR Aur (Khokhlova et al. 1995), 46 Dra (Tsymbal et al. 1998) and 66 Eri (Yushchenko et al. 1999). Preliminary results for HD 153720 were reported by Yushchenko et al. (2002).

One might suppose that unevolved stars of equal mass comprising a binary should have very similar, if not identical, compositions. This supposition is questionable, however, for main sequence A and F stars for which chemical peculiarities are extremely common. Do the components of HD 153720 have the same chemical composition?

Our previous investigations of the chemical composition of the individual components of A-type binaries showed strong differences for some elements. For AR Aur (Khokhlova et al. 1995), a system with mass ratio MA/MB=1.09, the abundances of the majority of the elements are similar, but on the whole differ from the normal (i.e., solar) values. For Al, Ni, Pt, and Hg, there are differences between the components of more than 1 dex. A similar situation was found for 46 Dra (Tsymbal et al. 1998) with mass ratio  MA/MB=1.14: the abundances are similar for all elements except Al, Ga, Sr, Pt. A more interesting case is 66 Eri (Yushchenko et al. 1999), with MA/MB=0.98 (Yushchenko et al. 2001). In the spectrum of the main component, only lines of elements with $Z\leq26$ and barium lines were found. The spectrum of the secondary component showed numerous lines of heavy elements, indicating overabundances of these elements of between 2 and 6 dex. A similar binary system was analysed by Catanzaro et al. (2003): HD 191110 ( MA/MB=1.08). We may mention also the paper by Adelman et al. (1998) on 46 Dra, where abundance differences for Sr and Pt of more than 1 dex exist between the primary and the secondary.

HD 153720 was originally chosen as a comparison star for our program. We expected HD 153720 to be a system of unevolved components with solar chemical composition. Our abundance analyses, the first for the star, show that HD 153720 is a system of metallic-line stars.

2 Spectral observations and data reduction

The two spectra of HD 153720 used in this study were obtained in 1996 with the echelle spectrograph of the 2.7-m McDonald Observatory telescope (Tull et al. 1995). Sixty-three spectral orders covered the wavelength range 3895-10 260 Å at a resolving power of 60 000 and signal to noise ratio > 200. The spectral coverage is incomplete beyond about 5600 Å.

Initial reduction of the spectra was done at the University of Texas by V. Woolf with the participation of V. L. Khokhlova using the KPNO-IRAF software package. Further reduction and calculations were performed using the URAN (Yushchenko 1998) package. The location of the continuum was determined taking into account the synthetic spectrum. The heliocentric Julian dates of our spectra can be found in Table 4. It should be noted that one of the spectra was obtained at a time when the spectral lines of the components were unresolved at a separation of less than 5  ${{\rm km~s^{-1}}}$.

This spectrum was used only to provide an estimate of radial velocity with an error near 1  ${{\rm km~s^{-1}}}$. The spectrum with the lines separated by 104  ${{\rm km~s^{-1}}}$ was used for the determination of atmospheric parameters and for abundance calculations.

3 Radial velocities and orbital elements

To measure the radial velocities of the components of HD 153720 we selected unblended lines of neutral iron. The data are only available in electronic form at http://www.edpsciences.org. The results are listed in Table 4. This table gives the heliocentric Julian dates of the observations, the corresponding orbital phases (according to the ephemerides from Table 5), designations for the binary components, the mean radial velocity, the number of measured spectral lines, and the errors of the mean.

Our data and 46 observations of Heard & Hurkens (1975) permit us to determine new orbital elements of the system. In Table 5 we show two sets of new orbital elements and the Heard & Hurkens (1975) result. The residuals of our radial velocity values from the first solution are near 4  ${{\rm km~s^{-1}}}$ and significantly larger than the errors of our measurements but may approximate the errors of of the Heard & Hurkens (1975) measurements. The residuals are decreased if we correct our radial velocities by -4  ${{\rm km~s^{-1}}}$. The second solution was made with our velocities shifted by 4  ${{\rm km~s^{-1}}}$ with respect to the earlier results.

3.1 Third body?

The 4  ${{\rm km~s^{-1}}}$ difference may be explained in two ways. First, it is quite common to attribute such a difference to a third body in the system. In this case, the amplitude of the variations of the $\gamma$-velocity is at least 3  ${{\rm km~s^{-1}}}$. The orbital period of a third body should be significantly larger than 3 years - the time interval covered by the observations of Heard & Hurkens (1975). If the period is of the order of several dozen years, the mass of the third body can be less than a solar mass. Second, a systematic error in the old observations of Heard & Hurkens (1975) may exist. However, for 66 Eri we found no systematic shift between our observations (Yushchenko et al. 2001) and those of Young (1976). The observations of Heard & Hurkens (1975) were made with the 1.88 m telescope of the David Dunlap Observatory. There is no evidence for a systematic shift between the radial velocities obtained at the David Dunlap and modern measurements with coudeé spectrographs at the McDonald Observatory: see, for example Scarfe et al. (1994), and Fekel & Tomkin (1993).

We found no evidence of third light in our analysis of chemical abundances in the visual spectral region. Indeed, the predicted combined flux of the two stars is consistent with the measured fluxes from the ultraviolet (TD1 observations, Thompson et al. 1978), through the optical (UBV photometry), to the infrared (IRAS observations, Moshir et al. 1989). We conclude that, if a third body exists, it must be a low mass main sequence star.

Table 1: Radial velocities of components of HD 153720 .

Table 2: Orbital elements of HD 153720 .

4 Atmospheric parameters of the components and the ratio of luminosities

Given that the components are of a similar spectral class, we used calibrations for single stars to find the first approximation to their effective temperature and surface gravity. Following Kurucz (1995) we found $T_{{\rm eff}}$ = 7350${\rm K}$, $\log~g$ = 3.7 using $\it UBV$ colors and $T_{{\rm eff}}$ = 7200${\rm K}$, $\log~g$ = 4.0 using $\it uvby$ colors. So we selected the nearest root in the Kurucz (1995) grid of atmosphere models: the model with $T_{{\rm eff}}$ = 7250${\rm K}$, $\log~g$ = 4.0 with solar chemical composition and a microturbulent velocity of 2.0  ${{\rm km~s^{-1}}}$. Then, we calculated a synthetic spectrum for the whole observed region. Two copies of this spectrum were shifted in accordance with the observed radial velocities of the components and coadded with the assumption that  LA/LB=1.0. The composite synthetic spectrum was used for continuum placement and line identification.

To find more precise values of the components' parameters we used calculations of iron abundances obtained with different atmosphere models. This method was described in detail by Yushchenko et al. (1999), Yushchenko et al. (2004) and Gopka et al. (2004). Briefly, we calculate the iron abundance from individual lines for each point across a four-dimensional grid involving  $T_{{\rm eff}}$, $\log~g$, $v_{{\rm micro}}$, and LA/LB. For each grid point, the coefficients of correlations between the equivalent widths (and the excitation potentials) and the iron abundances, and the scatter of the abundances were calculated. On the assumption that  LA/LB = 1, the correct atmosphere parameters were selected as those providing zero (or very close to zero) correlation coefficients and minimal scatter in the derived iron abundances.

Next, we interpolated in the Kurucz (1995) grid to construct a mini-grid with effective temperatures 6750 ${\rm K}$ $\leq$  $T_{{\rm eff}}$ $\leq$ 7750 ${\rm K}$ ($\Delta$ $T_{{\rm eff}}$ = 25 ${\rm K}$) and surface gravities from 3.5 $\leq$ $\log~g$ $\leq$ 4.5 ($\Delta$$\log~g$ = 0.1). 54 lines of neutral iron in the spectrum of the primary and 44 lines in the spectrum of secondary were used for the calculations. These clean lines were selected from synthetic spectra of the components. Solar oscillator strengths of these lines were found using the Liège solar atlas (Delbouille et al. 1974), as described below in the section on abundance analysis. Calculations of iron abundances from neutral iron lines were made for all the above listed models with different microturbulent velocities (0.5 $\leq$  $v_{{\rm micro}}$ $\leq$ 8.0) and light ratio of the components ( $0.85\leq L_A/L_B \leq1.15$). The Kurucz (1995) WIDTH9 program was used.

The best set of parameters are $T_{{\rm eff}}$ $_A =7425 \pm50$ ${\rm K}$, $\log~g$ $_A =4.0 \pm0.15$, $v_{{\rm micro}}$ $_A=2.7\pm0.2~{\rm
km~s^{-1}}$ for component A, and $T_{{\rm eff}}$ $_B =7125 \pm50$ ${\rm K}$, $\log~g$ $_B =3.9 \pm0.15$, $v_{{\rm micro}}$ $_B=2.7\pm0.2~{\rm
km~s^{-1}}$, for component B. The ratio of luminosities is  $L_A/L_B =1.10 \pm0.025$. The errors are internal errors. High dispersion spectra permit us to use unblended iron lines of the components to find the atmosphere parameters. That is why the errors are of the same order as those found for single or SB1 stars; see, for example, our analyses of the barium star HD 202109 (Yushchenko et al. 2004) or the metal-poor halo star HD 221170 (Gopka et al. 2004). Of course, larger systematic errors can exist, and these errors will influence determinations of the parameters of both single and double stars.

The above-listed parameters were used for the abundance determinations. The ratio of luminosities was assumed to be 1.10 at wavelength 5800 Å. For other wavelengths it was changed in accordance with the ratios of the Kurucz (1995) predicted fluxes interpolated to the above mentioned parameters. All abundance calculations, except for the neutral iron lines, were made using the spectrum synthesis method. In this method it is necessary to know the line broadening parameters. In our case these parameters are the velocities of rotation and macroturbulence.

5 Synchronization of rotation, age and mass of HD 153720 

We estimate the projected rotational velocity by analyzing the profiles of iron lines with accurate oscillator strengths. We find that ${v\sin i}$  = 15  ${{\rm km~s^{-1}}}$ for both components. This value includes all possible line broadening mechanisms, including macroturbulence. The following argument shows that our estimate of the projected rotational velocity is a plausible value.

The Hipparcos parallax of HD 153720 is $10.56\pm0.50$ mas. Taking into account the visual magnitude of HD 153720, the bolometric corrections from Kurucz (1995), our parameters of atmosphere models and the flux ratio of the components, we find the radii of the components to be  $R_A=1.86~R_{\odot}$, and  $R_B=1.92~R_{\odot}$ and their luminosities to be  $L_A=9.54~L_{\odot}$ and  $L_B=8.55~L_{\odot}$. Given the radius and the ${v\sin i}$  = 15  ${{\rm km~s^{-1}}}$, we may estimate an upper limit to the period of rotation for a component: the upper limits are 6 $.\!\!^{\rm d}$3 and 6 $.\!\!^{\rm d}$5 for components A and B, respectively. These estimates are approximately half the orbital period.

Synchronization of orbital and rotational velocities requires that the components rotate with velocities near  $v_{\rm circle}=8.5{-}9$  ${{\rm km~s^{-1}}}$ in the case of a circular orbit. The eccentricity of the orbit is near 0.4. The maximum orbital velocity with this value of the eccentricity is approximately 50 percent higher than the mean orbital velocity. Tidal effects are strongest when the distance between stars is smallest, i.e., at maximum orbital velocity. Then, we expect that, at synchronization, the rotational velocities of the components will be higher than the values for circular orbits. The values of the pseudosynchronized velocity for an elliptical orbit can be calculated using the formula of Giuricin et al. (1984):

\begin{displaymath}v=v_{\rm circle}(1+e)^{1/2}/(1-e)^{3/2}.
\end{displaymath} (1)

For our case, this formula gives a result near 22  ${{\rm km~s^{-1}}}$. For an inclination of the orbit i near 43 degrees, this corresponds to the observed ${v\sin i}$  of 15  ${{\rm km~s^{-1}}}$ for both components. Then, the masses of the components are larger than  $2.0~M_\odot$.

However, evolutionary tracks (Claret 1995; Claret & Gimenez 1995; Claret 1997; Claret & Gimenez 1998) predict that a  $2.0~M_\odot$ star at the observed temperature and surface gravity has twice the observed luminosity. The evolutionary tracks show that the masses of the components are in the range  $1.48{-}1.65~M_\odot$ and the age in the range  $1.5{-}1.7\times 10^9$ years. At these lower masses, the angle i is the range 70 to 78 degrees and the rotational velocities of the components are near 16  ${{\rm km~s^{-1}}}$. These velocities of the components of HD 153720 are slower than the pseudosynchronized velocity. The difference between the predicted pseudosynchronized velocity (22  ${{\rm km~s^{-1}}}$) and the observed value is in the range of scattering of observed values for different binary systems investigated by Giuricin et al. (1984).

Note that from the Hipparcos-based diameters and spectroscopic surface gravities $\log~g$A=4.0, $\log~g$B=3.9 we find the masses of the stars to be  $M_A=1.27~M_\odot$ and  $M_B=1.07~M_\odot$, values lower than from the evolutionary tracks. To remove this discrepancy we should increase the spectroscopic surface gravities to the values $\log~g$ A=4.07-4.12, $\log~g$ B=4.04-4.09. These differences are within the errors of our determinations of the spectroscopic surface gravities of the components.

6 Abundance analysis

\end{figure} Figure 1: Part of the spectrum of HD 153720 . The axes are the wavelength in angstroms and relative flux. In the top part of the figure the points are the observed spectrum. The solid line is the sum of the synthetic spectra of the components. Scaled synthetic spectrum of the component A is shown separately in the middle part, of the component B - in the bottom part of the figure. Synthetic spectra of the components are shifted in accordance with the observed radial velocities and smoothed by instrumental and rotational profiles. The positions of the spectral lines, which were taken into account in synthetic spectra calculations are marked by short and long dashes (faint and strong lines). Some strong lines are identificated.
Open with DEXTER

\end{figure} Figure 2: The same as Fig. 1 in the vicinity of the Ba II line $\lambda $5853.688. The position of this line in the spectrum of component B is marked by a vertical broken line. Three coadded synthetic spectra near the position of bariun lines in component B correspond to the best barium abundance, and the abundances changed by $\pm $0.5 dex with respect to the best abundance.
Open with DEXTER

We calculated the abundances of all elements except iron by using synthetic spectra, as described in detail by Yushchenko et al. (2004). The synthetic spectra of the components were constructed using the SYNTHE code of Kurucz (1995) and the URAN code (Yushchenko 1998). We used atomic lines from the following databases: Kurucz (1995), Morton (2000), DREAM (Biemont et al. 2002) and VALD (Piskunov et al. 1995) with a few lines from other sources. We took into account the hyperfine structure and isotopic splitting of Mn, Cu, and Ba lines, as given by Kurucz (1995) and François (1996). The synthetic spectra of the components were shifted to the observed radial velocities and coadded, taking into account the flux ratio of the components. The value  LA/LB=1.1 was used for the wavelength of 5800 Å. For other wavelengths the flux ratio was scaled according to theoretical fluxes from the Kurucz (1995) models. To find unblended lines we compared the observed spectrum with the coadded synthetic spectrum of the system. To find the uncertainties of the derived abundances we made calculations for each component with two atmosphere models - that of the component and that of the other component. This procedure permits us to estimate the uncertainties due to the atmosphere model and due to the fitting method. The data for the individual lines can be found on the web sites "users.odessa.net/$^\sim$yua'' and "yushchenko.netfirms.com''. In Figs. 1 and 2, we present a part of the observed spectrum of HD 153720 , the synthetic spectra of the components and the coadded synthetic spectrum.

Table 3: Chemical composition of HD 153720 .

We also derived the solar abundances from the adopted lines. For this calculation, we use the Liège Solar Atlas (Delbouille et al. 1974) and the Grevesse & Sauval (1998) model atmosphere. The adopted values of the microturbulent and macroturbulent velocities are 0.8  ${{\rm km~s^{-1}}}$ and 1.8  ${{\rm km~s^{-1}}}$ respectively. The continuum in the Liège Solar atlas is corrected in accordance with Ardeberg & Virdefors (1979) and Rutten & van der Zalm (1984). The solar abundances are used with the stellar abundances to obtain the differential abundances which are independent of the adopted oscillator strengths.

Mean results for the investigated elements are listed in Table 3. The first columns in Table 3 are from left to right the atomic number, and the identification of atom or ion. Then follow four entries, two of three columns and two of two columns, concerning the abundances: the three columns are the mean logarithmic abundance of the element relative to the solar value, the rms error of one measurement and the number of lines contributing to the mean abundance. The first two of the four entries concern results for components A and B obtained with the model atmosphere of that component. The last two entries, each of two columns, give the abundances and the rms error of each component when analysed with the model atmosphere of the other component.

Figure 3 shows the abundances versus atomic number. The abundance patterns are those of metallic line stars. C, N, O, Mg, Al, Si, and Ca are underabundant with respect to the Sun. Cu and heavier elements are overabundant in both components. The abundances of other elements are approximately solar values or different for the A and B components. Both components of HD 153720 are metallic line stars.

7 Normal A stars are rare

The metallic line (Am) stars were first recognized as a group by Titus & Morgan (1940). They found that the Ca K line did not lead to the same spectral classification as the hydrogen lines for some Hyades A stars. Since then, it has been noticed that most if not all slowly rotating A and F stars exhibit abundance anomalies. The abundances of CNO are generally below solar, while those of iron- peak elements are generally above solar, and the species in between can be either under- or overabundant.

\end{figure} Figure 3: The abundances of chemical elements in the atmosphere of HD 153720  A and B with respect to their abundances in the solar atmosphere.
Open with DEXTER

The classical Am stars are characterized by a general enhancement of the metallic lines but a marked weakness of the lines of Ca and Sc when compared to normal A stars at the resolution of classification spectra. Lines of heavy elements are enhanced for most stars. The AmFm stars may arise as a result of radiatively driven diffusion processes (Michaud 1970). Binarity is a common feature of Am stars (Abt 1961) and tidal effects might be of great importance in controlling diffusion processes (Budaj 1996). The general catalog of Am and Ap stars (Renson et al. 1991) contains 6684 objects but only for a small number of these stars have chemical abundances been determined from high resolution spectra.

It should be noted that Am and other chemically peculiar stars are significantly more common than normal A stars. It is very hard to find a star with a normal (solar-like) chemical composition among A-type stars. Nonetheless, Budaj (1996) compiled a list of 61 apparently non-chemically peculiar binaries with spectral classes from A4 to F1.

Eight of Budaj's stars including HD 153720 have been analysed for their composition. For six of the eight, abundance anomalies indicative of chemical peculiarities have been seen in one or both components of these binaries: the six stars are comprised of HD 28910 (86 Tau - Varenne & Monier 1999; and Takeda & Sadakane 1997), HD 83808 (o Leo - Griffin 2002), HD 153720 (this paper), HD 178449 (Budaj & Iliev 2003), HD 205767 and HD 217792 (Erspamer & North 2003). Two stars, HD 32537 (9 Aqr - Hui-Bon-Hoa 2000) and R CMa (Tomkin & Lambert 1989) have been classified as normal, but it should be noted that the abundances of only six and five chemical elements were measured for HD 32537 and R CMa, respectively and, furthermore, the heaviest investigated element was Ni. If the abundances of heavier elements are measured, it is possible that the classification of the stars would be changed. Continued scrutiny of Budaj's binaries may uncover systems without chemical peculiarities but it is clear that it is difficult to find a normal star in this region of HR diagram.

8 Conclusion

In this paper, we present the first abundance analysis of the components of the SB2 system HD 153720 . The atmospheric parameters of the components were found from careful analysis of iron abundances calculated from individual atomic lines. We used differential spectrum synthesis to find the abundances of 21 elements in both components. In addition, Cu was found only in component A, and Ce - only in component B.

HD 153720 is a system with twin metallic-line (Fm) components. The star has not been classified previously as a metallic line star.

The Hipparcos parallax and our parameters of atmosphere models enable us to estimate the mass and the age of the components, and the inclination of the orbit. The rotational velocities of the components are probably slightly lower than the pseudosynchronized velocity.

There is a hint that this spectroscopic binary may have a third body. Our radial velocities are offset by 4  ${{\rm km~s^{-1}}}$ from the published data obtained about thirty years earlier.

The star HD 153720 highlights the difficulty of finding normal A-type stars as members of a spectroscopic binary. It appears in Budaj's (1996) list of such normal spectroscopic binaries, but our abundance analysis reveals chemical peculiarities characteristic of metallic line stars.

We would like to thank to L. Delbouille and G. Roland for sending us the Liège Solar Atlas. We thank V. Woolf for assistance with reduction of the McDonald spectra. We use data from NASA ADS, SIMBAD, CADC, VALD, NIST, and DREAM databases and we thank the teams and administrators of these projects.

Work by A.Y. and Y.K. was supported by the Astrophysical Research Center for the Structure and Evolution of the Cosmos (ARCSEC) of Korea Science and Engineering Foundation (KOSEF) through the Science Research Center (SRC) program.



Online Material

Table 4: Iron lines in the spectrum of HD 153720.

Table 5: Lines of other elements in the spectrum of HD 153720.

Copyright ESO 2004