Free Access
Issue
A&A
Volume 579, July 2015
Article Number A47
Number of page(s) 18
Section Stellar atmospheres
DOI https://doi.org/10.1051/0004-6361/201425541
Published online 25 June 2015

© ESO, 2015

1. Introduction

Type II Cepheids are the older, fainter, low-mass counterpart to the classical Cepheids (e.g., Wallerstein 2002). As such, they fall into the instability strip between the RR Lyrae and the RV Tau stars, and their periods are bound to 1 day on the lower end and to 20 days on the upper end, following the classification of Soszyński et al. (2008b). However, the limit between RR Lyrae and Type II Cepheids, on the one hand, and between Type II Cepheids and RV Tau stars, on the other, are not clearly defined. Type II Cepheids are themselves divided in two sub-classes; the BL Her stars have periods ranging from 1 to 4 days while the W Vir stars have periods between 4 and 20 days, again according to Soszyński et al. (2008b). In our current understanding, the different classes correspond to stars in different evolutionary stages: BL Her stars are currently evolving from the horizontal branch (HB) to the asymptotic giant branch (AGB) and can be considered as post early-AGB stars (Castellani et al. 2007). W Vir stars cross the instability strip in their so-called “blue-nose” from the AGB while they are undergoing He-shell flashes. Finally, RV Tau stars are about to leave the AGB, so they are crossing the instability strip towards the white dwarf domain (Gingold et al. 1985; Bono et al. 1997; and references therein; see also Maas et al. 2007, for further considerations).

Various criteria have been tested to distinguish between classical and Type II Cepheids, and among the Type II Cepheids, to distinguish the BL Her and the W Vir stars. They are based on the shape of the light curve, on the stability of the period, or on the presence of distinctive features in the spectra. If these criteria have proved to be useful (see for instance Schmidt et al. 2004a), they are not sufficient to secure a robust classification. Indeed for various criteria, the properties of different types of variables overlap over various period ranges (e.g., Schmidt et al. 2005b; Soszyński et al. 2008b). Moreover, the use of external parameters (metallicity, proper motion, distance to the Galactic plane) is hampered by the fact that Type II Cepheids are very heterogeneous, because they span a wide metallicity range, and they can be found in the bulge, the thick disc, the halo, or in globular clusters.

Table 1

Spectroscopic observations of HQ Car and DD Vel.

The difficulty for properly classifying the Type II Cepheids can be illustrated by the two stars in our sample: DD Vel and HQ Car. DD Vel is classified as a classical Cepheid pulsating in the fundamental mode in the ASAS catalogue (Pojmanski 2002) and in the Machine-learned ASAS Classification Catalog (MACC, Richards et al. 2012) but as a Type II Cepheid in both the General Catalog of Variable Stars (GCVS, Samus et al. 2007−2013) and in the International Variable Star Index (VSX, Watson 2006). It is not listed as a Type II Cepheid by Harris (1985). HQ Car is considered as a classical Cepheid in the GCVS and in both the ASAS and MACC catalogues, while it is listed as a Type II Cepheid by Harris (1985) and in the VSX. It is worth mentioning that neither of the two stars is listed in the Fernie database of classical Cepheids (Fernie et al. 1995).

Surprisingly, Type II Cepheids did not receive much attention (see references in Maas et al. 2007) after the early spectroscopic analyses of Rodgers & Bell (1963, 1968), Barker et al. (1971), and Anderson & Kraft (1971). Indeed, modern high resolution spectroscopic studies have been limited to only a few stars: TX Del (Andrievsky et al. 2002), ST Pup (Gonzalez & Wallerstein 1996), V553 Cen (Wallerstein & Gonzalez 1996), and RT Tra (Wallerstein et al. 2000) before Maas et al. (2007) analysed 19 BL Her and W Vir stars. The study of Maas et al. (2007) is, to date, the unique extensive abundance analysis of Type II Cepheids. Amongst other results, they made three significant points regarding the classification of Type II Cepheids:

  • Amongst stars with short periods, there is a clear separation into two groups. The BL Her stars are relatively metal-rich and show excesses of sodium, carbon, and nitrogen, along with thick disc kinematics. The UY Eri stars are significantly more metal-poor and are similar to stars in the halo;

  • Stars with periods between 10 and 20 days, such as W Vir itself, show metallicities ranging approximately from 1.0 to 2.0 and are similar to variables in globular clusters;

  • Stars with periods longer than 20 days show element separation as do RV Tau stars. A few stars with periods in the 2030 day range, such as TW Cap, are as metal-poor as the 1020-day stars.

Since the Maas et al. (2007) paper, only two new Type II Cepheids have been studied in detail: QQ Per by Wallerstein et al. (2008) and W Vir by Kovtyukh et al. (2011).

After a brief description of the data in Sect. 2, we examine in Sect. 3 different classification criteria and discuss the chemical composition of HQ Car and DD Vel in Sect. 4.

2. Data

The HQ Car spectra have been obtained with different instruments: one spectrum was taken with the echelle spectrograph on the 4m Blanco telescope at Cerro Tololo Inter-American Observatory (CTIO). It has a resolution of 28 000 and covers the 55008000 Å wavelength range with a signal-to-noise, S/N, (per pixel) of 57 in the order containing Hα. Another spectrum was obtained using the 2.2 m ESO/MPG telescope and the FEROS echelle spectrograph at the ESO La Silla observatory (Kaufer et al. 1999). The spectrum covers the 35009200 Å wavelength range with a resolution of 48 000 and a S/N (per pixel) in excess of 150 over the largest part of the spectrum. Finally, two spectra were obtained with the HARPS (Mayor et al. 2003) echelle spectrograph mounted at the 3.6 m telescope at the ESO La Silla observatory, which provides a resolution R = 115 000 over a wide spectral range (38006900 Å). They both reach a S/N of 50 at 650 nm.

We analysed two spectra for DD Vel: the first one consists of four back-to-back FEROS spectra1 coadded in order to increase the S/N. The second2 was obtained with the UVES (Dekker et al. 2000) echelle spectrograph (R = 40 000) using the DIC2 (437+760) configuration. The blue and red arms cover the wavelength intervals [37505000] Å and [56507600/76609460] Å. Relevant information concerning the observations and pulsation parameters of the Cepheids are listed in Table 1. As shown in the next sections, strong emission features become prominent at some phases, and the spectra are therefore not suitable for an accurate abundance determination. We used the CTIO spectrum (φ = 0.361) for HQ Car and the FEROS spectrum (φ ≈ 0.292) in the case of DD Vel.

3. Classification

3.1. Classification based on the location on a colourmagnitude diagram

thumbnail Fig. 1

Location of DD Vel and HQ Car in a K, JK colourmagnitude diagram. The dashed line shows the zero-age-horizontal-branch (ZAHB), while the coloured lines display HB evolutionary models for stellar masses ranging from 0.49 to 0.58 M. The black lines display the instability strip for RR Lyrae and BL Herculis stars.

Open with DEXTER

Following the suggestion of an anonymous referee, we performed a detailed comparison in the K, JK colourmagnitude diagram to constrain the nature of the candidate Type II Cepheids. Figure 1 shows evolutionary prescriptions for α-enhanced HB evolutionary models (Pietrinferni et al. 2004, 2006) at fixed chemical composition (see labelled values) and the two targets. It shows the zero-age-horizontal-branch (ZAHB) and HB evolutionary models for three different values of the stellar masses ranging from 0.49 to 0.58 M. The apparent NIR magnitudes of the targets are based on 2MASS photometry (Skrutskie et al. 2006). They were unreddened using the empirical reddening law provided by Cardelli et al. (1989). The true distance modulus was estimated using the K-band period-luminosity relation for Type II Cepheids provided by Matsunaga et al. (2006). We found MK = −3.81 mag for DD Vel and MK = −3.87 mag for HQ Car. Data plotted in this figure show that the position of the targets agrees quite well, within the errors, with the current evolutionary prescriptions, thus further supporting the working hypothesis that they are Type II Cepheids. It also displays the instability strip for RR Lyrae and BL Herculis stars. The hottest edge shows the first overtone blue edge, while the coolest shows the fundamental red edge. Current pulsation predictions (Marconi et al. 2015) suggest that the edges of the instability strip are independent of the metal content in the NIR bands. The above edges should be cautiously treated, since they have been slightly extrapolated to higher luminosities to cover the magnitude range of the targets.

3.2. Classification based on emission features in the spectrum

3.2.1. Emission in Hα

The presence of emission in the Hα lines of W Vir was first reported by Joy (1937) and Sanford (1953). Joy (1949) noted that hydrogen line emission was a common feature of Type II Cepheids in globular clusters; it has afterwards also been reported in field Type II Cepheids (e.g., Wallerstein 1958; Harris & Wallerstein 1984). Following Schwarzschild (1953), the doubled absorption profiles, together with emission with an inverse P-Cygni profile in Hα, were early modeled as a shock-wave passing through the atmosphere in the rising part of the lightcurve (Whitney 1956a,b; Wallerstein 1959; Whitney & Scalafuris 1963): emission originates in the de-excitation region behind the radiative shock wave. Lèbre & Gillet (1992) used high resolution spectroscopy to follow the evolution of the emission profile of Hα through an entire cycle of W Vir and further improved the shock model. Finally, Kovtyukh et al. (2011) extended this study to many metallic lines including Fe I, Fe II, Na I, and Ba II and concluded that W Vir consists in its inner part of a pulsating star with periodic shocks reaching the upper atmosphere and in its outer part of a circumstellar envelope.

thumbnail Fig. 2

Behaviour of the Hα line in HQ Car at different phases.

Open with DEXTER

thumbnail Fig. 3

Behaviour of the Hα line in DD Vel at different phases.

Open with DEXTER

In Fig. 2, we present the variations in the Hα profile for HQ Car at four different phases. They are very similar to those presented for W Vir by Lèbre & Gillet (1992, their Fig. 3 and by Kovtyukh et al. (2011, their Fig. 12). The shock wave rising in the atmosphere of the star causes a broad emission feature comprising five components (3 in emission, 2 in absorption). The absorption features are associated to the presence of a circumstellar envelope for one and to the fall back of the upper atmosphere located above the shock for the other. In particular, the Hα emission totally disappears at φ = 0.361, in good agreement with Lèbre & Gillet (1992), who mention that the shock emission is present during the entire cycle except between phases φ = 0.38−0.44. In the case of DD Vel (Fig. 3), the Hα emission is weak as expected for φ = 0.292 but, more surprisingly, also weak at φ = 0.798. Given our limited phase coverage for this star, it could also very well be that we missed the phase of strong Hα emission.

3.2.2. Emission in He I at 5875.64 Å

thumbnail Fig. 4

Behaviour of the 5853 Ba II line and the 5876 He I line in HQ Car at different phases.

Open with DEXTER

thumbnail Fig. 5

Same as Fig. 4 for DD Vel.

Open with DEXTER

Emission lines of He I in the spectra of type II Cepheids were first mentioned by Wallerstein (1959). They enabled him to confirm the shock model. (The emission is caused by helium ionized by the shock wave that captures electrons.) More recent observations by Raga et al. (1989) were used to determine the H/He ratio in the atmosphere of W Vir. He I emission lines were extensively studied by both Lèbre & Gillet (1992) and Kovtyukh et al. (2011). The former reported the presence of emission in the He 5875.64 Å line between phases φ = 0.827 and φ = 0.009, while the latter detected emission between φ = 0.865 and φ = 0.201 for the same line. In both studies the emission peaks between φ ~ 0.8 and φ ~ 0.1, when the shock reaches its highest intensity.

In Fig. 4, we present the variations in the He I 5875.64 Å line profile for HQ Car at four different phases. Once again, they are very similar to those presented for W Vir by Lèbre & Gillet (1992, their Fig. 6) and by Kovtyukh et al. (2011, their Figs. 4. Emission is present only at the end of the cycle, at phases φ = 0.891 and φ = 0.961, in good agreement with previous studies. In the same figure, we note the doubling of the Ba II line profile in the same phases. The mechanism responsible for the line doubling was first explained by Schwarzschild (1953): line doubling can be observed when the shock wave moves across the layer of formation of a given absorption line, provided that this layer is thick enough; the blueshifted line is produced by cooling gas moving upwards, while the redshifted line originates in gas already falling down. In the case of DD Vel, Fig. 5 shows no emission for the He I 5875.64 Å line, but the Ba II line is split into two components at φ = 0.798, indicating that the region where the Ba II line is formed is crossed by the shock wave.

3.3. Kinematics consistent with a thick disc membership

Kinematics alone is not sufficient to decide that a star belongs to the thin or the thick disc. It is, however, interesting to investigate the kinematic properties of DD Vel and HQ Car. Therefore we computed their space velocities ULSR, VLSR, WLSR in the local standard of rest3 using proper motions from the Naval Observatory Merged Astrometric Dataset (NOMAD, Zacharias et al. 2004) and the data shown in Table 2. Both stars have a total velocity 70 ≤ vtot ≤ 180 km s-1, making them likely thick disc members (e.g., Nissen 2004). Their velocity along the direction of Galactic rotation VLSR almost falls within 100 km s-1 and 40 km s-1, the range quoted by Reddy et al. (2003) for a probable thick-disc membership. Finally, comparing the stars in our sample to the velocitymetallicity plots of Bensby et al. (2007, see their Fig. 1), we find that ULSR is not conclusive for HQ Car, while VLSR and WLSR both place this star in the thick disc. The situation is a bit less clear for DD Vel because its ULSR is still at the upper limit for a thin disc star, and its VLSR at the lower limit, whereas its WLSR is typical of the thick disc. In conclusion, HQ Car seems to be a very likely thick-disc member ,whereas it cannot be totally excluded that DD Vel is a thin-disc member. The above kinematical evidence therefore supports the hypothesis that both stars are Type II Cepheids in the thick disc rather than classical Cepheids located in the thin disc.

Table 2

Kinematics of HQ Car and DD Vel.

3.4. Classification based on the chemical composition

From their period (with respect to the classification of Soszyński et al. 2008b for Type II Cepheids) and the emission features in Hα and He I at 5876 Å, it already appears to be clear that both HQ Car and DD Vel are W Vir stars. This will be reinforced in Sect. 4 where we examine their chemical composition.

4. Chemical composition

4.1. Method

We used the DECH 30 software package4 to normalize the individual spectra to the local continuum, to identify the lines of different chemical elements, and to measure the equivalent widths (EW) of the absorption lines. The oscillator strengths have been taken from the Vienna Atomic Lines Database (VALD Kupka et al. 1999).

To determine the effective temperature Teff, we employed the line depth ratios method of Kovtyukh (2007), which comes from the work of Kovtyukh & Gorlova (2000). The ratios of the central depths of carefully chosen pairs of lines that have a very different dependence on Teff are entered in previously calibrated relations. This technique allows determining Teff with great precision: the use of several tens (50) of ratios per spectrum leads to uncertainties of 1020 K when S/N> 100 and of 3050 K when S/N< 100. The method is independent of the interstellar reddening and only marginally dependent on the individual characteristics of stars, such as rotation, microturbulence, and metallicity.

To determine the surface gravity (log  g) and the microturbulent velocity Vt, we used a canonical analysis. We sought the surface gravity from the excitation equilibrium of Fe I and Fe II lines, and the microturbulent velocity is determined from the Fe I lines. We note that the excitation equilibrium is also satisfied by V I and V II and, to a slightly lesser extent, by Ti I and Ti II in HQ Car, while it is satisfied for the couples Si I/Si II, Ti I/Ti II (but not Cr I/Cr II) in the case of DD Vel. As far as the microturbulent velocity is concerned, an innovative approach using lines of several elements has been developed by Sahin et al. (2011) and is illustrated in Reddy et al. (2012). In this method, the standard errors are plotted as a function of the microturbulent velocity. We applied it to the stars in our sample, and the results are in good agreement with our values for Vt. They are described in Appendix A.1. The atmospheric parameters for DD Vel and HQ Car are listed in Table 3.

Table 3

Atmospheric parameters derived for HQ Car and DD Vel.

The lines of odd-Z elements can be broadened due to their hyperfine structure (hfs). However, the hfs corrections are negligible in the case of V or Co for the considered EW. This is not true in the case of Sc, Mn, or Cu (e.g., North et al. 2012; Reddy et al. 2012). We therefore computed the abundances of these elements via spectral synthesis using the 5526.79, 5657.90, 5667.15, 6245.62, 6604.60 lines for Sc II, 5420.35, 5432.56, 6013.48, 6021.79 for Mn I and 5105.55, 5218.21, 5782.14 for Cu I, and the STARSP code developed by Tsymbal (1996). We took the hyperfine structure of Sc II (Prochaska & McWilliam 2000), Mn I, and Cu I (Allen & Porto de Mello 2011) into account for the line profile calculations.

Atmospheric models are interpolated for each Type II Cepheid using the grid of 1D, LTE atmosphere models of Castelli & Kurucz (2004). Individual abundances are listed in Table 4 and abundance ratios (with respect to iron) in Table 5. We computed the solar reference abundances using lines in the Sun with EWs < 120 mA and the same atmosphere models (Castelli & Kurucz 2004). They are listed in the Appendix B, together with the prescriptions of Asplund et al. (2009) and the solar abundances of Reddy et al. (2003) that are used by Reddy et al. (2006) in their study of the thick disc.

Table 4

Individual abundances [X/H] in HQ Car and DD Vel.

Table 5

Abundance ratios [X/Fe] in HQ Car and DD Vel.

Table 6

Abundance uncertainties due to uncertainties on the atmospheric parameters, computed for HQ Car (Teff = 5580 K, log  g = 1.6 dex, Vt = 3.1 km s-1, [ Fe / H ] = −0.327 dex).

We used 25 calibrations to determine the effective temperature of HQ Car and 26 calibrations for DD Vel, leading to standard deviations of 95 K and 109 K, respectively, and standard errors of 19 K and 22 K. We adopted 100 K as the uncertainty on Teff. We estimated the uncertainty on log  g as ± 0.2 dex and the uncertainty on Vt as ± 0.5 km s-1. Table 6 lists the variations in the individual abundances [X/H] when changing the atmospheric parameters by ΔTeff = + 100 K, Δlog  g = + 0.2 dex, and ΔVt = + 0.2 km s-1 and their sum in quadrature, which we adopt as the uncertainty on the abundances due to the uncertainties on the atmospheric parameters. It is well documented (e.g., Johnson 2002) that such a method leads to overestimated values for the total error, because by construction it ignores covariances between the different atmospheric parameters. They nevertheless remain lower than 0.10 dex in most cases. The sum in quadrature of the errors associated with the uncertainties on the atmospheric parameters and of the standard deviation associated with the determination of the abundance of a given element gives the total error on the abundance for this element.

4.2. Chemical composition

The two stars in our sample have [Fe/H] in the 0.3 to 0.5 dex range, towards the lower end of the metallicity distribution for BL Her stars, but still in a domain where the metallicities of BL Her and W Vir stars overlap (see Maas et al. 2007). As shown just after this, they are probably affected by dust-gas separation. However, the [S/Fe] we measured for DD Vel (+0.09 dex) and HQ Car (+0.27 dex) are very similar to those already reported for [S/Fe] in different Galactic structures. Below [ Fe / H ] = −1.0 dex, [S/Fe], values are scattered around a plateau at +0.25 dex and decrease at higher metallicities until reaching [S/Fe] = 0.0 dex at [ Fe / H ] = −0.3 dex (e.g., François 1987, 1988; Chen et al. 2002; Nissen et al. 2007; Matrozis et al. 2013; Caffau et al. 2014, and references therein). Also our [Zn/Fe] measurement of +0.28 ± 0.12 dex in DD Vel is very consistent with previous values ([ Zn / Fe ] ≈ +0.1−+0.2 dex) reported for the thick disc (e.g., Mishenina et al. 2002; Bensby et al. 2005; Brewer & Carney 2006; Reddy et al. 2006). Since sulphur and zinc are only slightly depleted into dust (Savage & Sembach 1996), the typical thick-disc values for [S/Fe] and [Zn/Fe] in HQ Car and DD Vel indicate that their iron abundances are probably not very modified by the dust-gas separation. In particular, this allows us to use an average thick disc star for the 0.45 to 0.55 dex [Fe/H] bin for comparison purposes (Reddy et al. 2006, Col. 4 in their Table 7).

Maas et al. (2007) have shown that W Vir stars have [Na/Fe] that is independent of [Fe/H] and consistent with thick disc stars where ⟨ [ Na / Fe ] ⟩ = + 0.12 dex (Reddy et al. 2006). This argument does not apply to the stars with a severe dust-gas separation. They find, in contrast, that BL Her stars are strongly overabundant in sodium with a mean [Na/Fe] = +0.73 dex. For the two Type II Cepheids in our sample, [Na/Fe] varies between +0.07 and +0.15 dex, similar to the representative thick disc value (see Fig. 6). Since iron could be affected by dust-gas separation, we also compare [Na/Zn] for the BL Her and W Vir stars, and again DD Vel has low [Na/Zn] similar to the other W Vir stars, while the BL Her stars show very high (>+0.5 dex) values of [Na/Zn]. The absence of Na overabundance in our sample confirms that they are W Vir stars and not BL Her stars, as could already be inferred from their period (P> 4 d).

When gas cools sufficiently, dust grains can form and the abundances of the elements in the gas phase decrease. Because this happens at different temperatures for different trace elements, the quantity “50% condensation temperature” (50% Tc) has been defined, at which 50% of the element is found in the gas phase and the other 50% is locked in dust grains. We adopted the 50% Tc determined by Lodders (2003). A correlation between the underabundance of a given element and its condensation temperature is then interpreted as a dust-gas separation5. As can be seen in Fig. 7, the Type II Cepheids in our sample show hints of (mild) dust-gas separation: the more volatile elements have abundances similar to those of an average thick-disc star (Reddy et al. 2006), while the refractory elements are underabundant, because they are depleted into dust. For a better visibility, we focus in Fig. 8 on the elements with 50% Tc> 1300 K.

As expected, the signature of dust-gas separation is especially marked for the elements with the highest 50% Tc. In HQ Car, Ca, Nd, Al, and Sc are mildly depleted by 0.2 dex and Y by 0.3 dex with respect to an average thick-disc star. Zr seems to be very depleted because [Zr/H] clearly falls below the abundances of the other neutron-capture elements, but we have no comparison with an average thick-disc star to draw a firm conclusion. The depletion is more severe for DD Vel as the underabundances with respect to an average thick disc star reach 0.30 dex for Ti, 0.45 dex for Nd and Ca, 0.80 dex for Sc and 0.90 dex for Y. To further support the dust-gas separation, we note that most of the individual abundances fall below those of the thick-disc reference star for the elements with 50% Tc> 1400 K.

Hints or even clear evidence of dust-gas separation in Type II Cepheids have already been reported for ST Pup by Gonzalez & Wallerstein (1996) and for CO Pup, V1711 Sgr, MZ Cyg, and SZ Mon by Maas et al. (2007). On the other hand, these authors find the signature of dust-gas separation less convincing in RX Lib and W Vir because it relies mostly on the depletion in Sc. As we discuss in more detail in the next paragraph, severe dust-gas separation has also been reported in most of the RV Tau stars (see Giridhar et al. 2005, and references therein). Maas et al. (2007) found a Type II Cepheid (CC Lyr) with an extreme dust-gas separation, which is larger than in any RV Tau star. It is important to note that all the Type II Cepheids with a signature of dust-gas separation are W Vir stars and not BL Her stars. That the stars in our sample also show possible (HQ Car) or probable (DD Vel) signs of this phenomenon reinforces their classification as W Vir stars.

Dust-gas separation is a common feature in RV Tau stars (see Giridhar et al. 2005, and references therein). In a qualitative scenario (Waters et al. 1992), binary RV Tau stars are surrounded by a dusty disc. The dust-gas separation occurs when radiation pressure traps the dust grains in the disc while some of the gas (deprived from dust) is re-accreted on the star via the viscous disc that allows for transfer of angular momentum. The origin of the circumbinary disc in RV Tau stars is not clear, but it is generally believed to be created during binary interaction when the primary was a giant.

This scenario excludes metal-poor systems ([Fe/H] 1.0 dex) where dust cannot form in sufficient quantities, and indeed no sign of dust-gas separation has been found for metal-poor RV Tau variables in globular clusters (Gonzalez & Lambert 1996). Similarly, Maas et al. (2007) find no evidence of dust-gas separation for TW Cap, a W Vir star with [Fe/H] = 1.8 dex possibly associated to the halo of the Milky Way.

thumbnail Fig. 6

Top: [Na/Fe] vs. [Fe/H] for BL Her stars (open circles) and W Vir stars (filled circles) from Maas et al. (2007). HQ Car and DD Vel are overplotted in red and blue, respectively. The ratio for a representative thick-disc star at [Fe/H] = 0.5 dex is shown as a dashed line. Bottom: same for [Na/Zn] vs. [Zn/H]. Abundances have been rescaled to the solar abundances of Reddy et al. (2003), used as a reference by Maas et al. (2007).

Open with DEXTER

thumbnail Fig. 7

[X/H] vs. 50% Tc for the two Type II Cepheids in our sample. An average thick-disc star (grey dots) is plotted for comparison. Elements are identified by their chemical symbol. Abundances have been rescaled to the solar abundances that we recomputed with the Castelli & Kurucz (2004) models (See Appendix B).

Open with DEXTER

thumbnail Fig. 8

Same as Fig. 7, but focusing on the elements with the highest condensation temperature.

Open with DEXTER

It is not clear that the dust-gas separation has the same origin in W Vir stars as in RV Tau stars. In particular, the observed depletion is generally much shallower in the W Vir stars. Only in the case of CC Lyr does it reach the extreme values more commonly seen in RV Tau stars (e.g., 3.0 dex in HP Lyr and DY Ori, see Giridhar et al. 2005). The RV Tau stars depleted in their refractory elements are known binaries for a large number of them, supporting the hypothesis of a circumbinary dusty disc (Rao et al. 2012). Disentangling orbital velocities from pulsational velocities is very demanding in terms of observing time both in the case of RV Tau and W Vir stars, and indeed only four type II Cepheids are currently known as binaries: AU Peg, IX Cas, TX Del, and ST Pup. It is interesting to note that the only W Vir star in this group (ST Pup) shows obvious signs of dust-gas depletion, while the other shorter-period stars do not.

Recently reported observational (e.g., Marengo et al. 2010b) and theoretical (e.g., Neilson et al. 2012) lines of evidence support the existence of mass loss in classical Cepheids; however, these outflows seem to have a very low dust content (Marengo et al. 2013), possibly indicating that the wind is driven by pulsation and is not dust-driven as generally observed in evolved stars. On the other hand, extended dusty environments have been detected with high angular resolution techniques (e.g., Kervella et al. 2006; Gallenne et al. 2013, and references therein) and from extended emission in the mid- and far-infrared (Barmby et al. 2011). They have been attributed to the presence of a circumstellar envelope around the Cepheids.

As far as Type II Cepheids are concerned, Kovtyukh et al. (2011) analysed hydrogen, helium, and metallic lines in W Vir itself. They were able to reproduce the specifics of spectral line variability in W Vir with the help of a non-linear pulsation model. Results suggest that W Vir consists of two different layers, the inner part being the pulsating star itself and the outer part a very extended and dense atmosphere, and it might even include a circumstellar envelope with a very low expansion rate.

If the most desirable experiment in the near future were to systematically examine the binarity properties of W Vir stars, it would nevertheless be interesting to investigate whether dust-gas separation could also somehow take place in their circumstellar envelopes.

5. Summary and conclusion

The status of the HQ Car and DD Vel Type II Cepheids has remained unclear. Depending on the catalogue (i.e., on the method and criteria used to perform the classification), they are sometimes listed as classical Cepheids and sometimes as Type II. Because we observed emission features in the Hα and in the 5875.64 Å He I lines that are characteristic features of W Vir stars, we conclude that HQ Car and DD Vel are Type II Cepheids from this sub-class. Their periods of 14.06 and 13.19 days, respectively, and the absence of Na overabundance further indicates that they are not BL Her stars. Moreover, they show a possible (HQ Car) or probable (DD Vel) signature of mild dust-gas separation. Such abundance patterns have currently been observed only in long-period Type II Cepheids and RV Tau stars, thus reinforcing our classification.

Several studies of the Galactic abundance gradients in the thin disc using classical Cepheids have reported increased dispersion in the outer disc (Yong et al. 2006; Lemasle et al. 2008; Luck et al. 2011; Luck & Lambert 2011; Genovali et al. 2014); however, these findings are hampered by the possible contamination of the current samples by misclassified Type II Cepheids that are thick-disc members. Including unrecognized Type II Cepheids modifies the abundance patterns not only because they belong to another stellar population but also because their current distances are computed with period-luminosity relations that are only valid for classical Cepheids. Thick-disc members can be identified by their specific location in the [α/Fe] vs. [Fe/H] plane (e.g., Recio-Blanco et al. 2014), and recent studies indicate that the thin disc contamination by thick-disc stars is not negligible (see, for instance, Mikolaitis et al. 2014, their Fig. 7).


1

Prog. ID: 060.A-9120.

2

Prog. ID: 082.D-0901.

3

[U, V, W] = [11.10, 12.24, 7.25] km s-1 (Schönrich et al. 2010).

5

It should be noted that this assumption involves implicit approximations (such as the formation of dust grains under conditions of thermodynamic equilibrium) that may not be met in the surroundings of W Vir stars due to the presence of shocks in the atmosphere (Kovtyukh et al. 2011).

Acknowledgments

The authors thank the anonymous referees for the very valuable comments that helped to improve the quality of this paper. The authors thank Dr. D. Yong and Prof. B. W. Carney for providing the HQ Car spectrum. The authors thank Profs. C. Dominik and L. B. F. M. Waters for useful discussions on the dust-gas separation phenomenon. V.K. acknowledges the support from the Swiss National Science Foundation, project SCOPES No. IZ73Z0152485. G.G. acknowledges the support of the Chilean fund FONDECYT-regular (project 1120190).

References

Online material

Appendix A: Appendix A

Appendix A.1: A.1. An alternative approach to determining Vt

In this Appendix we show the standard deviation around the mean abundance plotted as a function of the microturbulent velocity, following the approach of Sahin et al. (2011). The

thumbnail Fig. A.1

Standard deviation around the mean abundances as a function of microturbulence Vt for HQ Car, shown for several elements.

Open with DEXTER

thumbnail Fig. A.2

Same as Fig. A.1, but for DD Vel.

Open with DEXTER

dispersion of the abundances is computed for the Fe I, Fe II, Si I, and Ni I lines, while the microturbulent velocity Vt is varied from 1 to 6 km s-1. The minimum value of the dispersion is in good agreement within the different elements, confirming the values of the microturbulence (derived solely from Fe I) adopted in this study, namely 3.1 km s-1 for HQ Car and 3.8 km s-1 for DD Vel.

Appendix A.2: A.2. List of lines used

Table A.1

Atomic parameters and EWs of HQ Car and DD Vel.

Appendix B: Appendix B: Solar references

Table A.1

Solar abundance derived from the solar spectrum using the grid of models by Castelli & Kurucz (2004), compared to the solar photospheric abundance by Asplund et al. (2009) and by Reddy et al. (2003).

All Tables

Table 1

Spectroscopic observations of HQ Car and DD Vel.

Table 2

Kinematics of HQ Car and DD Vel.

Table 3

Atmospheric parameters derived for HQ Car and DD Vel.

Table 4

Individual abundances [X/H] in HQ Car and DD Vel.

Table 5

Abundance ratios [X/Fe] in HQ Car and DD Vel.

Table 6

Abundance uncertainties due to uncertainties on the atmospheric parameters, computed for HQ Car (Teff = 5580 K, log  g = 1.6 dex, Vt = 3.1 km s-1, [ Fe / H ] = −0.327 dex).

Table A.1

Atomic parameters and EWs of HQ Car and DD Vel.

Table A.1

Solar abundance derived from the solar spectrum using the grid of models by Castelli & Kurucz (2004), compared to the solar photospheric abundance by Asplund et al. (2009) and by Reddy et al. (2003).

All Figures

thumbnail Fig. 1

Location of DD Vel and HQ Car in a K, JK colourmagnitude diagram. The dashed line shows the zero-age-horizontal-branch (ZAHB), while the coloured lines display HB evolutionary models for stellar masses ranging from 0.49 to 0.58 M. The black lines display the instability strip for RR Lyrae and BL Herculis stars.

Open with DEXTER
In the text
thumbnail Fig. 2

Behaviour of the Hα line in HQ Car at different phases.

Open with DEXTER
In the text
thumbnail Fig. 3

Behaviour of the Hα line in DD Vel at different phases.

Open with DEXTER
In the text
thumbnail Fig. 4

Behaviour of the 5853 Ba II line and the 5876 He I line in HQ Car at different phases.

Open with DEXTER
In the text
thumbnail Fig. 5

Same as Fig. 4 for DD Vel.

Open with DEXTER
In the text
thumbnail Fig. 6

Top: [Na/Fe] vs. [Fe/H] for BL Her stars (open circles) and W Vir stars (filled circles) from Maas et al. (2007). HQ Car and DD Vel are overplotted in red and blue, respectively. The ratio for a representative thick-disc star at [Fe/H] = 0.5 dex is shown as a dashed line. Bottom: same for [Na/Zn] vs. [Zn/H]. Abundances have been rescaled to the solar abundances of Reddy et al. (2003), used as a reference by Maas et al. (2007).

Open with DEXTER
In the text
thumbnail Fig. 7

[X/H] vs. 50% Tc for the two Type II Cepheids in our sample. An average thick-disc star (grey dots) is plotted for comparison. Elements are identified by their chemical symbol. Abundances have been rescaled to the solar abundances that we recomputed with the Castelli & Kurucz (2004) models (See Appendix B).

Open with DEXTER
In the text
thumbnail Fig. 8

Same as Fig. 7, but focusing on the elements with the highest condensation temperature.

Open with DEXTER
In the text
thumbnail Fig. A.1

Standard deviation around the mean abundances as a function of microturbulence Vt for HQ Car, shown for several elements.

Open with DEXTER
In the text
thumbnail Fig. A.2

Same as Fig. A.1, but for DD Vel.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.